Surviving Your Dissertation 4

Save your time - order a paper!

Get your paper written from scratch within the tight deadline. Our service is a reliable solution to all your troubles. Place an order on any task and we will take care of it. You won’t have to worry about the quality and deadlines

Order Paper Now

Surviving Your Dissertation A Comprehensive Guide to Content and Process

Kjell Erik Rudestam Fielding Graduate University

Rae R. Newton Fielding Graduate University

FOR INFORMATION:

SAGE Publications, Inc.

2455 Teller Road

Thousand Oaks, California 91320

E-mail: order@sagepub.com

SAGE Publications Ltd.

1 Oliver’s Yard

55 City Road

London EC1Y 1SP

United Kingdom

SAGE Publications India Pvt. Ltd.

B 1/I 1 Mohan Cooperative Industrial Area

Mathura Road, New Delhi 110 044

India

SAGE Publications Asia-Pacific Pte. Ltd.

3 Church Street

#10-04 Samsung Hub

Singapore 049483

All rights reserved. No part of this book may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system, without permission in writing from the publisher.

Printed in the United States of America

Library of Congress Cataloging-in-Publication Data

Rudestam, Kjell Erik.

Surviving your dissertation : a comprehensive guide to content and process / Kjell Erik Rudestam, Fielding Graduate University, Rae R. Newton, Fielding Graduate University. —Fourth edition.

pages cm.

Includes bibliographical references.

ISBN 978-1-4522-6097-6 (pbk. : alk. paper)

1. Dissertations, Academic—United States. 2. Report writing. 3. Research—United States. I. Newton, Rae R. II. Title.

LB2369.R83 2015

378.2—dc23 2014002905

This book is printed on acid-free paper.

14 15 16 17 18 10 9 8 7 6 5 4 3 2 1

Acquisitions Editor: Vicki Knight

Editorial Assistant: Yvonne McDuffee

Production Editors: Laura Barrett, David C. Felts

Copy Editor: Paula L. Fleming

Typesetter: C&M Digitals (P) Ltd.

Proofreader: Sarah J. Duffy

Indexer: Joan Shapiro

Cover Designer: Rose Storey

Marketing Manager: Nicole Elliott

Contents Preface About the Authors Part I: Getting Started

1. The Research Process 2. Selecting a Suitable Topic 3. Methods of Inquiry: Quantitative and Qualitative Approaches

Part II: Working With Content: The Dissertation Chapters 4. Literature Review and Statement of the Problem 5. The Method Chapter: Describing Your Research Plan 6. Presenting the Results of Quantitative Research 7. Presenting the Results of Qualitative Research 8. Discussion

Part III: Working With Process: What You Need to Know to Make the Dissertation Easier

9. Overcoming Barriers: Becoming an Expert While Controlling Your Own Destiny 10. Writing 11. How to Complete Your Dissertation Using Online Data Access and Collection 12. Guidelines for the Presentation of Numbers in the Dissertation 13. Informed Consent and Other Ethical Concerns

References Name Index Subject Index

Preface

We are pleased to present the fourth edition of Surviving Your Dissertation. As with previous editions, we have sought to answer questions that students and faculty have at every stage of the dissertation process. In past editions, we have illustrated the challenge of engaging in such a notable project with book covers that depict a bridge leading into an impenetrable jungle and life buoys close at hand to negotiate the stormy seas. The cover of this edition offers a different, but equally relevant image: the pride of victory achieved by scaling a lofty peak.

In many ways, these images also reflect our own experience in writing the book. The field of research in the social and behavioral sciences has expanded rapidly over the past several years, and we have frequently felt as if we are scrambling to keep up. This edition reflects our experience. We have maintained the overall structure of the book, which has been well received so far, while updating content on those topics that are indispensable to the dissertation process: the selection of an appropriate research topic; the review of the literature; the description of the methodology and research design; the collection and analysis of data; and the interpretation, presentation, and discussion of the results and implications of the study. Within this updated material, we have once again attempted to provide sufficient detail to enable the reader to know exactly what goes into each section and chapter of the dissertation and how to format that information. In addition, we continue to include topics that are not always present in sources of this kind: the many types of quantitative and qualitative research models and approaches that are available to the student, the principles of good scholarly and academic writing, suggestions for how to select and work with committees, and tips for overcoming task and emotional blocks that may impede progress. Throughout, we have replaced older references with newer, more contemporary ones, including many new dissertation examples taken from our students and colleagues.

We have also added significant new content to the fourth edition. We note that the traditional null hypothesis significance-testing model is being challenged and augmented by an emphasis on clinical or practical significance and a corresponding use of measures of effect size and

confidence intervals. We have explained and illustrated this new approach to the presentation of statistical results. Similarly, we have acknowledged an emerging emphasis on theoretical “models” and the influence of model building on the design and presentation of research. We provide several examples of studies that incorporate this approach. We are increasingly impressed by the implications of the Internet for the entire research enterprise. We have expanded our discussion of the Internet as a source of data, an opportunity for data collection, and a vehicle for data analysis, as well as providing recommendations of potentially helpful websites and software programs that may be unfamiliar to the average reader. We also acknowledge the parallel expansion of available data sources in all their varied forms, including data archives, social media, and what is currently known as “big data.” Finally, we have expanded the sections on qualitative and multimodal methods of research, which have an inductive, theory- building focus. With respect to all these topics, we have tried to explain the concepts, illustrate them with new tables and figures, and, in many cases, provide very specific details about how to incorporate them into a research study.

We believe that this book is suitable for a large academic and professional audience. Of course, it is primarily directed at the graduate student who envisions or is involved in writing a research dissertation. Thus, there is significant focus on material that is best suited for the beginning doctoral student—for example, how to develop a research question, how to construct a table or figure, how to report a statistical finding, how to use American Psychological Association formatting conventions, and so on. However, there is also content directed at the more advanced student—for example, how to conceptualize and illustrate a mediation model, how to report multiple regression findings, and how to code text for a grounded theory study. Moreover, we have become increasingly aware that Surviving Your Dissertation serves also as a resource for researchers and practitioners who have either forgotten important details or are motivated to keep abreast of evolving research practices in their fields. Perhaps more important, we view the book as a convenient source of information for faculty who are currently supervising graduate students’ dissertations or research projects.

We remain deeply indebted to our own students, who continue to thrill us with their creativity and force us to keep learning in order to stay a step ahead. We hope that they, and you, find this volume a helpful and steady

companion in your research and writing endeavors.

A large number of individuals have contributed to the completion of this project. We called upon many faculty colleagues to nominate student dissertations that exemplify high levels of scholarship and have sprinkled references to these dissertations throughout the book to illustrate important principles and recommendations. We are very appreciative of these relatively recent graduates for allowing us to share their first major research endeavors in this venue. We also benefitted from the critical reflections and insight of the following reviewers of the third edition of Surviving Your Dissertation: Anne J. Hacker, Bernie Kerr, Karin Klenke, Kaye Pepper, and Udaya R. Wagle. Their observations and suggestions were both reinforcing and helpful in crystallizing changes for this edition.

We are also grateful to be part of the SAGE family, a collaborative, dedicated group of professionals who have facilitated our writing careers in so many ways. The following individuals were notable contributors: Vicki Knight, publisher and senior editor, has always provided us with a balance of inspiring leadership and nurturing support. Her editorial assistant, Jessica Miller, has been consistently responsive to our frequent requests for assistance. Laura Barrett and David Felts, project editors, have gracefully guided the editorial process from start to finish. And Paula Fleming, our copy editor, is truly a paragon in her craft. Her grammatical acuity, common sense, and work ethic cannot be overestimated. Thank you all.

Finally, we must thank our partners in life, Jan and Kathy, for their continuing patience and support as we have devoted our energy and attention to four editions of this volume.

About the Authors

Kjell Erik Rudestam is Professor of Psychology at Fielding Graduate University, Santa Barbara, California, where he served as Associate Dean of Academic Affairs for many years. He was previously a psychology professor at York University, Toronto, and Miami University, Oxford, Ohio, after receiving his PhD in Psychology (Clinical) from the University of Oregon. He is the author of Your Statistical Consultant: Answers to Your Data Analysis Questions, 2nd edition (also with Rae R. Newton), Handbook of Online Learning, 2nd edition (with Judith Schoenholtz-Read), and eight other books, as well as numerous articles in professional journals on topics including suicide, psychotherapy, and family and organizational systems. He is a Fellow of the American Psychological Association (Division 12), a Diplomate of the American Board of Examiners in Professional Psychology (Clinical), a Diplomate of the American Academy of Experts in Traumatic Stress, and holds an Honorary Doctorate of Science from The Professional School of Psychology.

Rae R. Newton is Professor of Sociology Emeritus at California State University, Fullerton. He recently joined the faculty of the School of Psychology at Fielding Graduate University where he serves as a research consultant and statistical advisor to doctoral students and faculty. He received his PhD in sociology from the University of California, Santa Barbara, and completed postdoctoral training in mental health measurement at Indiana University. His primary interests include longitudinal modeling of outcomes for high risk youth and foster care populations, family violence and statistics education. He is author, with Kjell Erik Rudestam, of Your Statistical Consultant: Answers to Your Data Analysis Questions, now in its second edition and numerous articles in professional journals on topics including family violence, child maltreatment, and measurement. In semi-retirement he enjoys traveling with his wife in their RV and surfing throughout Mexico and Central America.

Part I Getting Started

The Research Process Selecting a Suitable Topic Methods of Inquiry: Quantitative and Qualitative Approaches

1 The Research Process

There is a story about a Zen Buddhist who took a group of monks into the forest. The group soon lost its way. Presently one of the monks asked the leader where they were going. The wise man answered, “To the deepest, darkest part of the forest so that we can all find our way out together.” Doctoral research for the graduate student in the social sciences is often just such an experience—trekking into a forest of impenetrable density and making many wrong turns. Over the years, our students have used various metaphors to describe the dissertation process, metaphors that convey the feeling of being lost in the wilderness. One student compared the process to the Sisyphean struggle of reaching the top of a hill, only to discover the presence of an even higher mountain behind it. Another student experienced the task as learning a Martian language, known to the natives who composed her committee but entirely foreign to her. A third student offered perhaps the best description when she suggested that it was like waiting patiently in a seemingly interminable line to gain admission to a desirable event, then finally reaching the front only to be told to return to the rear of the line.

One reason that students become more exasperated than necessary on the dissertation journey is that they fail to understand the procedures and practices that form the foundation for contemporary social science research. Many students who are attracted to their field of interest out of an applied concern are apprehensive about making the leap from application to theory, an indispensable part of the research enterprise. What may not be so evident is that many of the skills that go into being a consummate practitioner are the same ones demanded of a capable researcher. It is well known that curiosity and hypothesis testing are the bedrock of empirical research. In a similar fashion, experienced psychotherapists, to take an example from clinical psychology, are sensitive and keen observers of client behavior. They are persistent hypothesis testers. They are curious about the relationship between family history variables and current functioning. They draw on theory and experience to help select a particular intervention for a particular client problem or moment in therapy.

Dispassionate logic and clear and organized thinking are as necessary for effectiveness in the field as they are for success in research. In fact, the

bridge between research and just plain living is much shorter than most people think. All of us gather data about the world around us, wonder what will happen if we or others behave in particular ways, and test our pet hunches through deliberate action. To a large extent, the formal research enterprise consists of thinking systematically about these same issues.

The procedures outlined in this book are intended to assist the doctoral student in planning and writing a research dissertation, but the suggestions are equally applicable to writing a master’s thesis. In fact, there is considerable overlap between these two challenging activities. For most students, the master’s thesis is the first rigorous research project they attempt. This means that, in the absence of strong, supportive faculty consultation, the student often concludes the thesis with considerable relief and an awareness of how not to do the study the next time! With a doctoral dissertation, it is generally expected, sometimes as an act of faith, that the student is a more seasoned and sophisticated researcher. The consensus opinion is that dissertations are generally longer than theses, that they are more original, that they rely more heavily on theoretically based arguments, and that they make a greater contribution to the field.

In most graduate programs, the prelude to conducting a dissertation study is presenting a dissertation proposal. A research proposal is an action plan that justifies and describes the proposed study. The completion of a comprehensive proposal is a very important step in the dissertation process. The proposal serves as a contract between the student and his or her dissertation or thesis committee that, when approved by all parties, constitutes an agreement that data may be collected and the study may be completed. As long as the student follows the steps outlined in the proposal, committee members should be discouraged from demanding significant changes to the study after the proposal has been approved. Naturally, it is not uncommon to expect small changes, additions, or deletions as the study progresses because one can never totally envision the unpredictable turns that research can take.

There is no universally agreed-on format for the research proposal. To our way of thinking, a good proposal contains a review of the relevant literature, a statement of the problem and the associated hypotheses, and a clear delineation of the proposed method and plan for data analysis. In our experience, an approved proposal means that a significant percentage of the work on the dissertation has been completed. As such, this book is

intended to help students construct research proposals as well as complete dissertations.

The Research Wheel One way to think about the phases of the research process is with reference to the so-called research wheel (see Figure 1.1). The wheel metaphor suggests that research is not linear but is rather a recursive cycle of steps. The most common entry point is some form of empirical observation. In other words, the researcher selects a topic from the infinite array of possible topics. The next step is a process of inductive logic that culminates in a proposition. The inductive process serves to relate the specific topic to a broader context and begins with some hunches in the form “I wonder if . . . .” These hunches typically are guided by the values, assumptions, and goals of the researcher, which need to be explicated.

Figure 1.1 The Research Wheel

Stage 2 of the research wheel is a developed proposition, which is expressed as a statement of an established relationship (e.g., “the early bird is more likely than the late bird to catch the worm”). The proposition exists within a conceptual or theoretical framework. The role of the researcher is to clarify the relationship between a particular proposition and the broader context of theory and previous research. This is probably the most challenging and creative aspect of the dissertation process.

Theories and conceptual frameworks are developed to account for or describe abstract phenomena that occur under similar conditions. A theory

is the language that allows researchers to move from observation to observation and make sense of similarities and differences. A conceptual framework, which is simply a less-developed form of a theory, consists of statements that link abstract concepts (e.g., motivation, role) to empirical data. If not placed within such a context, the proposed study has a “So what?” quality. This is one of the main objections to the research proposals of novice researchers: The research question may be inherently interesting but ultimately meaningless. For instance, the question “Are there more women than men in graduate school today?” is entirely banal as a research question unless the answer has conceptual or theoretical implications that are developed within the study.

Although a study may be worthwhile primarily for its practical implications (e.g., “Should we start recruiting more men into graduate schools?”), a purely applied study may not be acceptable as a dissertation. Kerlinger and Lee (1999), authors of a highly respected text on research methodology, noted that “the basic purpose of scientific research is theory” (p. 5). Generally speaking, a research dissertation is expected to contribute to the scholarly literature in a field and not merely solve an applied problem. Thus, developing a proposition for one’s dissertation typically involves immersing oneself in the research and theoretical literature of the field to identify a conceptual framework for the study.

Having stated our position on the role of theory in dissertation research, we now need to take a step back. As a psychologist and a sociologist, respectively, we are most familiar with research conventions within these two disciplines. Other branches of the social sciences have their own standards of what constitutes an acceptable dissertation topic. We have attempted to keep this book as generalizable as possible and to infuse it with examples from other fields. Ultimately, of course, you will need to follow the rules and conventions that pertain to your discipline as well as to your university and department.

For example, a few major universities allow a doctoral student to submit one or more published articles as the equivalent of a dissertation. Many others encourage studies that consist of secondary data analyses derived from national databases, such as U.S. Census data or the General Social Survey, or data obtained from a larger study. Some fields—notably social work, education, policy evaluation, and professional psychology—may encourage dissertations that solve applied problems rather than make

distinct theoretical contributions. Studies that evaluate the effectiveness of programs or interventions are a case in point because they sometimes contribute little to validating a theory. Political science and economics are examples of fields that are diverse enough to accommodate both theoretically based studies and purely applied studies. Within the subspecialty of international relations, for instance, one could imagine a survey and analysis of security agreements of European nations after the unraveling of the North Atlantic Treaty Organization (NATO) that rely on interviews with foreign policy makers and are largely descriptive and applied. In contrast, a study of the role of a commitment to ideology in the success of political parties in the United States, based on an analysis of historical documents and voting records, might be grounded in a theory of how ideology attracts or alienates the voting public.

Moving forward along the research wheel, the researcher uses deductive reasoning to move from the larger context of theory to generate a specific research question. The research question is the precisely stated form of the researcher’s intent and may be accompanied by one or more specific hypotheses. The first loop is completed as the researcher seeks to discover or collect the data that will serve to answer the research question. The data collection process is essentially another task of empirical observation, which then initiates another round of the research wheel. Generalizations are made on the basis of the particular data that have been observed (inductive process), and the generalizations are tied to a conceptual framework, which then leads to the elucidation of further research questions and implications for additional study.

The kinds of skills called for at the various points of the research wheel are reminiscent of the thoughts about learning presented by Bertrand Russell many years ago. Russell noted that there are two primary kinds of knowledge acquisition: knowledge by description and knowledge by acquaintance. Knowledge by description is learning in a passive mode, such as by reading a book on how to change the oil in one’s car or hearing a lecture on Adam Smith’s theory of economics. This type of learning is especially well suited to mastering abstract information; in other words, it is better for learning about economics than about changing the oil. Knowledge by acquaintance, on the other hand, is learning by doing—the kind of skill training that comes from practicing a tennis serve, driving an automobile, and playing with a computer. This is concrete knowledge acquisition, oriented to solving problems.

The research process demands both skills. First, the researcher needs to apply clear, logical thinking to working with concepts and ideas and building theories. It is our impression that many graduate students, particularly those who have experience as practitioners in their fields, are weaker in this abstract conceptualization, and honing this skill may be the major challenge of the dissertation. Second, the researcher must engage in the practical application of ideas, including by systematically planning a study and then collecting and analyzing data. The ability to focus, problem solve, and make decisions will help bring the study to completion.

2 Selecting a Suitable Topic

The selection of an appropriate topic is the first major challenge in conducting research. In many academic settings, this task is simplified by working with a faculty mentor who is already familiar with an interesting area of study, may have an extensive program of research in that area, and may even have defined one or more researchable questions. It is quite common for students interested in a particular area of research to not only select their doctoral institution but also select their dissertation chair, with the goal of joining the research program of a noted scholar in that field.

On the other hand, you may not be blessed with a faculty role model who is actively engaged in research in an area of interest to you. There are no simple rules for selecting a topic of interest, but there are some considerations with respect to appropriateness. It is generally unwise to define something as important as a dissertation topic without first obtaining a broad familiarity with the field. This implies a large amount of exploring the literature and studying the experts. Without this initial exploration, you can neither know the range of possibilities of interesting topics nor have a clear idea of what is already known. Most students obtain their research topics from the loose ends they discover in reading within an area, from an interesting observation they have made (“I notice that men shut up when a beautiful woman enters the room; I wonder what the effect of physical attractiveness is on group process?”), or from an applied focus in their lives or professional work (“I have a difficult time treating these alcoholics and want to discover how best to work with them”). In short, there is no substitute for immersing oneself in a field of study by having conversations with leading scholars, advisers, and peers; critically reading the existing literature; and reflecting upon the implications of professional and personal experiences.

Some Guidelines for Topic Selection Here are some guidelines for deciding whether a topic is appropriate as a dissertation subject.

1. A topic needs to sustain your interest over a long period of time. A

study on learning nonsense syllables under two sets of environmental conditions may sound appealing in its simplicity, but remember Finagle’s first law of research: If something can go wrong, it will go wrong! Dissertations usually take at least twice as much time as anticipated, and there are few worse fates than slaving for hour after hour on a project that you abhor. Remember, too, that all dissertations are recorded and published by the Library of Congress, and you will always be associated with your particular study.

2. At the other extreme, it is wise to avoid a topic that is overly ambitious and overly challenging. Most students want to graduate, preferably within a reasonable period of time. Grandiose dissertations have a way of never being completed, and even the best dissertations end up being compromises among your own ambition, the wishes of your committee, and practical circumstances. It is not realistic for a dissertation to say everything there is to say about a particular topic (e.g., the European Union), and you need to temper your enthusiasm with pragmatism. As one student put it, “There are two types of dissertations: the great ones and those that are completed!” Sometimes it makes sense to select a research topic on the basis of convenience or workability and use the luxury of the postgraduate years to pursue more esoteric topics of personal interest.

3. We suggest that you avoid topics that may be linked too closely with emotional issues in your own life. It always makes sense to choose a topic that is interesting and personally meaningful. Some students, however, try to use a dissertation to resolve an emotional issue or solve a personal problem. For example, even if you think you have successfully overcome the personal impact of the death of your child, this is a topic to be avoided. It will necessarily stir up emotional issues that may get in the way of completing the dissertation.

4. A related issue is selecting a topic in which you have a personal ax to grind. Remember that conducting research demands ruthless honesty and objectivity. If you initiate a study to demonstrate that men are no damned good, you will be able neither to allow yourself the sober reflections of good research nor to acknowledge the possibility that your conclusions may contradict your expectations. It is much better to begin with a hunch (“I’ve noticed that men don’t do very well with housekeeping; I wonder if that has something to do with being pampered as children”) and to regard the research as an adventurous exploration to shed light on this topic rather

than as a polemical exercise to substantiate your point of view.

5. Finally, you need to select a topic that has the potential to make an original contribution to the field and allow you to demonstrate your independent mastery of subject and method. In other words, the topic must be worth pursuing. At the very least, the study must generate or help validate theoretical understanding in an area or, in those fields where applied dissertations are permissible, contribute to the development of professional practice. Some students are put off when they discover that a literature review contains contradictory or puzzling results or explanations for a phenomenon. However, such contradictions should be taken not as reasons to steer away from a topic but rather as opportunities to resolve a mystery. When people disagree or when existing explanations seem inadequate, there is often room for a critical study to be conducted. An opportunity to design a study that resolves theoretical contradictions within a discipline should not be overlooked.

From Interesting Idea to Research Question Let us assume that you have identified a general area of research and that your choice is based on curiosity and may involve resolving a problem, explaining a phenomenon, uncovering a process by which something occurs, demonstrating the truth of a hidden fact, building on or reevaluating other studies, or testing some theory in your field. To know whether or not the topic is important (significant), you must be familiar with the literature in the area. In Chapter 4, we present a number of suggestions for conducting a good review and assessment of the literature. In the meantime, we have noticed that many students have difficulty transforming an interesting idea into a researchable question, and we have designed a simple exercise to help in that endeavor.

Researchable questions almost invariably involve a relationship between two or more variables, phenomena, concepts, or ideas. The nature of that relationship may vary. Research studies generally consist of methods to explicate the nature of the relationship. Research in the social sciences rarely consists of explicating a single construct (e.g., “I will look at everything there is to know about the ‘imposter phenomenon’”) or a single variable (e.g., voting rates in presidential elections).1 Even the presence of two variables is apt to be limiting, and oftentimes it is only when a third

“connecting” variable is invoked that an idea becomes researchable. As a caveat, however, we acknowledge that research questions that are qualitative rather than quantitative in nature might not be as focused on the relationship between variables as on “how” processes develop or are experienced. We will have more to say about this distinction in forthcoming chapters.

An example might help to demonstrate how the introduction of an additional variable can lead to the birth of a promising study. Let us assume that I am interested in how members of a younger generation perceive the elderly. At this level, a study would be rather mundane and likely to lead to a “So what?” response. So far, it implies asking people what they think of the elderly, perhaps using interviews or tests or even behavioral observations. But we really won’t learn much of value about the nature of perceptions of the elderly in contemporary society and what influences those perceptions. Introducing a second variable, however, can lead to a set of questions that have promising theoretical (as well as practical) implications: I wonder what the role of the media is in shaping social perceptions of the elderly? I wonder if living with a grandparent makes any difference in how the elderly are viewed? I wonder how specific legislation designed to benefit the elderly has changed our perception of them? I wonder if there is a relationship between how middle-aged adults deal with their aging parents and how they view the elderly? The new variables introduced in these potential research questions are, respectively, the slant of the media, presence or absence of a grandparent, type of legislation, and treatment of one’s own parents. These variables impart meaning to the research because they offer suggestions as to what accounts for variability in perceptions of the elderly.

As an example of generating a research question using three primary variables, let’s say that you have inferred that many women lose interest in sexual relations with their husbands after the birth of a child. At this level, the proposed study would consist of checking out this hunch by assessing the sexual interest of women (Variable 1) before and after childbirth (Variable 2). But what would this finding mean? The introduction of a third variable or construct could lead to a much more sophisticated and conceptually meaningful study. An investigator might ask, “I wonder if the partner’s involvement in parenting makes a difference? What’s the role of his sexual initiative? How about childbirth complications? Father’s involvement in the birthing? The length of time they have been married?

Presence of other children in the home? Mother’s level of fatigue? Her body image?” There is no end to the number of interesting questions that can be raised simply by introducing another variable into the proposed study. This variable would then help to explain the nature of the relationship between the primary variables. In fact, one could brainstorm a whole list of third variables that could contribute to a better understanding of the relationship between childbirth and sexuality.

Note that the precise function of the third, or connecting, variable depends on the logic of the conceptual model or theory underlying the study. In this regard, a distinction can be made between two terms, mediator and moderator, which play important roles in research questions. A moderator variable pinpoints the conditions under which an independent variable exerts its effects on a dependent variable. Strictly speaking, a moderator effect is an interaction effect in which the influence of one variable depends upon the level of another variable (Frazier, Tix, & Barron, 2004). One commonly employed moderator variable is gender, which has two levels, male and female. The relationship between provocation and aggression, for example, may be very different for men and women. The role of context can also be conceptualized as a moderator variable. The famous Kinsey report on sexual behavior would certainly have generated very different results if the interviews with participants about their sex lives had taken place in the presence of family members. Identification of relevant contextual variables has important implications for the design of a study because such variables will affect the generalizability of research findings.

A mediating variable, on the other hand, tries to describe how or why rather than when or for whom effects will occur by accounting for the relationship between the independent variable (the predictor) and the dependent variable (the criterion). The mediator is the mechanism through which the predictor affects the outcome.2 As such, one can think of mediators as process variables. For example, in the counseling psychology field, maladaptive perfectionism can be regarded as either a moderating variable or a mediating variable (Wei, Mallinckrodt, Russell, & Abraham, 2004). Conceptualized as a moderator, attachment anxiety could be seen to exert negative effects on depressive mood only under conditions of high maladaptive perfectionism (i.e., there is a statistical interaction between maladaptive perfectionism and attachment anxiety). Conceptualized as a mediator, maladaptive perfectionism acts as an intervening variable

between attachment anxiety and depressive mood (i.e., there is an indirect relationship between anxiety and depression). As Wei et al. stated,

It is possible for maladaptive perfectionism to serve as both an intermediate link in the causal chain leading from attachment insecurity to depressive mood (i.e., as a mediator) and as a variable that alters the strength of association between attachment insecurity and depressive mood (i.e., as a moderator). (p. 203)

The diagram in Figure 2.1 captures the distinction between moderating and mediating variables in a theoretical model. In the case of mediation, the mediating variable (maladaptive perfectionism) is placed between attachment anxiety and depressed mood. In the case of moderation, the arrow from maladaptive perfectionism points to another arrow, that from attachment anxiety to depressed mood, indicating that the relationship between attachment anxiety and depressed mood depends on the level of maladaptive perfectionism.

Figure 2.1 The Distinction Between Moderator and Mediator Variables, Represented in a Causal Diagram

Figure 2.2 Classification Plot Representing Moderation of Relationship Between Attachment Anxiety and Depressed Mood, Moderated by Maladaptive Perfectionism

Source: Author created using data from Wei, W., Mallinckrodt, B., Russell, D., & Abraham, W. T. (2004). Maladaptive perfectionism as a mediator and moderator between adult attachment and depressive mood. Journal of Counseling Psychology, 51(2), 201–212.

We have illustrated one potential moderated outcome in the classification plot shown in Figure 2.2. Note that under conditions of low maladaptive perfectionism, there is only a small difference in depression between those with low and high attachment anxiety; however, under conditions of high maladaptive perfectionism, those with high levels of attachment anxiety are more likely to be depressed than those with low levels. In other words, maladaptive perfectionism moderates the relationship between attachment anxiety and depression, or, put another way, maladaptive perfectionism and attachment anxiety interact.

One research study is not likely to establish and verify all of the important elements of a complex conceptual model. As one of our colleagues puts it, you would need a video camera to capture the entire Grand Canyon on film, whereas the dissertation is more like a snapshot, perhaps of a mule

and rider descending one small section of one canyon trail. Yet the proposed model can provide a useful context for current and future research studies. Most ambitious research studies rely heavily on just such theoretical models.

As you might imagine, a researcher is in no position to test a model of this scope in a single study. For example, Gerald Patterson and his colleagues (Patterson, DeBaryshe, & Ramsey, 1989) spent many years developing and testing a model to explain aggressive and deviant behavior among young males. The model hypothesizes that such antisocial behavior can be causally linked to disrupted parental discipline and poor family management skills. Moreover, the relationship between these two sets of variables is not direct but is mediated by a network of other variables. The process is thought to begin with parents “training” a child to behave aggressively by relying on aversive behaviors in both punishment and negative reinforcement contingencies. The inability of the parents to control coercive exchanges among family members constitutes “training for fighting,” which leads, in turn, to aggressive behavior and poor peer relationships. This lack of social skills generalizes to antisocial behavior in the classroom, which makes it next to impossible for the youth to obtain basic academic skills, thus preparing him poorly to cope with life outside school. Ultimately, this set of factors leads to high rates of delinquent behavior. An abbreviated summary of one version of the model is shown in Figure 2.3.

Figure 2.3 A Model of Antisocial Behavior

Source: From Patterson, G. R., DeBaryshe, B. D., & Ramsey, E. A developmental perspective on antisocial behavior. American Psychologist, 44. Copyright © 1989, American Psychological Association. Reprinted with permission.

Over the years that Patterson and his colleagues (Patterson et al., 1989) spent elaborating the nature of these relationships, they conducted numerous studies that each constituted a “snapshot” of one aspect of this complex model, perhaps focusing on a particular set of relationships. For instance, the investigator might ask whether a relationship exists between physical fighting and poor peer relationships. Each variable would have to be operationalized, probably by obtaining more than one measure of both fighting and peer relationships. In Patterson’s work, he asked mothers, peers, and teachers to rate levels of physical fighting because their perspectives might differ. Likewise, peers, teachers, and self-reports are used to obtain measures of peer relations. The objective of the study—that is, to determine the nature and form of the relationship between the primary variables—determines the research method that is employed. In the early years of his career, Patterson focused on the relationships among contextual variables, parental beliefs, parenting practices, and child outcomes. He concluded that parenting practices—such as discipline, monitoring, problem solving, involvement, and positive reinforcement— serve as mediating variables between parenting beliefs and attitudes and the child’s behavior. Once this model was supported by sufficient data, Patterson proceeded to establish links between children’s behavior problems and subsequent chronic juvenile and adult offending (Reid, Patterson, & Snyder, 2002).

Whether or not a particular dissertation is designed to test a theory or model derived from the research literature, we believe that the creation of a visual model, which shows how the network of relevant variables and constructs may be related to one another, can serve as a powerful tool for guiding the study. Arranging your ideas spatially helps to organize your thinking, which in turn helps position your proposed study within a larger framework.

Research models are developed to account for the relationships among variables at a conceptual level and then used to guide the construction of research designs by which the relationships will be tested, usually, but not always, using contemporary statistics. The process is iterative so that the

models are modified on the basis of data and then reevaluated in further studies. Two primary types of relationships can be identified and explored within a causal model (Jaccard & Jacoby, 2010): predictive relationships and causal relationships. A predictive relationship implies that an association or correlation exists between two (or more) variables without assuming that one causes the other. For example, we may determine that traveling frequently as a child is related to (predicts) being more proficient in languages as an adult, without knowing whether travel actually causes improvement in language skills. Above, we used the terms independent variables and predictor variables more or less interchangeably. Strictly speaking, however, when the issue is prediction, the relationship is between one or more predictor variables and a criterion variable (Jaccard & Jacoby, 2010).

Causal relationships imply that one variable “causes” another; that is, changes in the primary variable, usually referred to as the independent variable, elicit changes in the second variable, the dependent variable or outcome variable. Although the principle of causation is the foundational bedrock for the experimental method in social science research, philosophers of science have argued for centuries whether causality can ever be truly demonstrated. Jaccard and Jacoby (2010) made a persuasive argument that the concept of causality is ultimately a heuristic that enables us to maintain an organized view of our world and of human behavior. By inferring causality, we can identify systematic relationships between variables and produce socially significant changes by manipulating some variables to influence others. Whether or not causality can be definitively demonstrated, much of contemporary research is conducted to give us confidence in theoretical models that purport causal relationships.

There are several types of causal relationships, and each type can play a role in developing a causal model. How causal models can be employed and evaluated in dissertations is a subject for a later chapter (Chapter 6). How to think about and construct causal models as a way of describing research ideas constitutes much of Jaccard and Jacoby’s (2010) very useful book, and the following discussion is stimulated by their work. Jaccard and Jacoby observed that most researchers begin by identifying an outcome variable that they want to understand better. An example might be the level of concern people express for the environment, including by engaging in behaviors that are environmentally sensitive, such as recycling waste or reducing pollution. The next challenge is to identify some

variables that could potentially influence or relate to environmentally sensitive behavior. One could imagine a study, for example, that seeks to determine which interventions would increase the motivation to recycle trash (or, more modestly, just to understand differences between those who readily recycle and those who do not). Perhaps we predict that having a neighborhood trash collection system that mandates sorting trash into recyclable and nonrecyclable categories will directly affect environmentally pro-social behavior. Of course, not all studies begin by identifying a dependent variable; it is also possible to choose an independent variable and speculate about its effects. For instance, a study might address the implications of working in a highly polluting industry for health, socioeconomic status, and social relationships.

Indirect causal relationships have an effect through the influence of an intermediary variable, which we have referred to above as a mediating variable. Moderated causal relationships are a third type of causal relationship, again defined earlier in the chapter. Most causal models contain a combination of the various types of relationships. Models can get very complex because of the number of variables and their subtle relationships with one another. Thus, simple path diagrams evolve into sophisticated theoretical networks. Many of these models have been developed over the course of a career by dedicated researchers, such as Gerald Patterson, who started with studies that explored subsets of variables within a model that then evolved in complexity. Because computer software allows for the manipulation of multiple variables simultaneously and relatively effortlessly, researchers usually begin by proposing one or more theoretical models and evaluating them empirically.

One example of a research model comes from the dissertation of Bill MacNulty (2004), one of our doctoral students. MacNulty generated this model from existing research literature and then tested it empirically using a number of well-validated self-report scales. The study employed the schema-polarity model of psychological functioning to assess how self- schemas (cognitive representations of self and others) influence the experience of gratitude and forgiveness and whether these variables mediate relationships between self-schemas and physical health and well- being. The model is summarized in Figure 2.4. The plus and minus signs refer to the direction of the hypothesized relationships among the variables. Although the results supported most of the initial hypotheses, the proposed model needed to be amended to accommodate the data. This

is typical of the research enterprise, in which theories and conceptual models are continually tested and refined to serve as increasingly sophisticated representations of real-life phenomena.

Another dissertation example comes from Jenny Knetig’s (2012) study of active duty military personnel who are at risk of experiencing symptoms of posttraumatic stress disorder (PTSD). Knetig speculated that being psychologically minded (a component of what is referred to as having a mentalizing capacity) allows some soldiers to perceive and interpret cognitive and affective states of themselves and others in a way that might facilitate resilience and help-seeking behaviors, which, in turn, mitigate how they are affected by severe stress. At the outset of her study, on the basis of the available literature and her own experience, Knetig postulated that the relationships among these variables might look something like the diagram in Figure 2.5.

Figure 2.4 A Theoretical Framework Presented as a Causal Diagram

Source: From Self-Schemas, Forgiveness, Gratitude, Physical Health, and Subjective Well-Being, by W. MacNulty, 2004, unpublished doctoral dissertation, Fielding Graduate University, Santa Barbara, CA. Copyright 2004 by W. MacNulty. Reprinted with permission of the author.

At the conclusion of the study, after collecting her data and performing a number of statistical analyses (canonical correlation analysis), Knetig amended her proposed model, as shown in Figure 2.6, to reflect the experiences of her participants more accurately.

Figure 2.5 Proposed Mediational Model Relating Psychological Mindedness to PTSD

Source: Knetig, 2012, p. 52. Reprinted with permission of the author.

In short, the data suggested that soldiers who are more psychologically minded are less apt to conceal their thoughts and feelings. It also suggested that the relationship between psychological mindedness and symptoms is mediated by self-concealment.

Figure 2.6 Final Mediation Model Relating Psychological Mindedness to PTSD

Source: Knetig, 2012, p. 54. Reprinted with permission of the author.

Generating Researchable Questions To help students generate researchable questions from their interesting ideas, we use a brainstorming exercise that begins with labeling one or two variables and generating a second or third. Brainstorming consists of

openly and noncritically listing all possible ideas in a given period of time. Later you can return to a more critical analysis of each idea and delete those that are uninteresting, not meaningful, or impractical. Ultimately, of course, it is contact with the literature that determines whether or not a research question is viable, because the literature houses the scholarly inquiry that goes beyond the limits of your own knowledge.

We suggest that you do this brainstorming exercise in a small group so that the person receiving the consultation merely serves as a scribe to record the ideas thrown out by the other group members (see Box 2.1). After 5 or 10 minutes, move on to the next person’s partially formed research topic. We generally use this exercise in groups of three or four so that group members can frequently shift groups and draw on the spontaneous reactions of a larger number of peers, uncontaminated by prior ideas or a particular mind-set.

The exercise involves suspending critical thinking and allowing new ideas to percolate. It should especially suit divergent thinkers, who will find the demand to be expansive in their thinking exciting and creative. Convergent thinkers may experience the exercise as a bit overwhelming, but they will find fulfillment in other stages of the research process that demand compulsivity, care, and precision. Every chapter of a dissertation contains both divergent and convergent elements.

Note that not all worthwhile research studies focus on three (or more) primary variables. Many studies look at the relationship between two variables or concepts, and a few descriptive studies make do with one variable or construct. The latter generally occurs in the early stages of research in an area, when little is known about a topic. Some investigators are pathfinders in terms of opening up new topics of research by trying to understand as much as possible about a phenomenon and generating more informed hypotheses for others to test in the future. Nevertheless, we believe that most students underestimate what is currently known about most topics and that the most interesting, practical, and theoretically meaningful studies are likely to consider relationships among several variables.

Our exercise for generating research questions is only one option. Jaccard and Jacoby (2010) have listed 26 different heuristics for how to think creatively about questions or phenomena that may be of interest to you. Items on their list include analyzing your own experiences, using case

studies, interviewing experts in the field, role-playing, and conducting thought experiments. The adoption of thought experiments might be particularly helpful. Jaccard and Jacoby viewed these as experiments that you conduct in your mind as if you are really obtaining the data and analyzing the results. You can then play with the possibilities of adding new variables or introducing different contingencies into the situation. One contingency is the introduction of so-called counterfactuals into the thought experiment (Tetlock & Belkin, 1996). Counterfactuals refer to “what-might-have-been” scenarios, such as what might happen if parents rather than administrators ran the public schools. Researchers can use this strategy to address theoretical alternatives that might otherwise not be raised or appreciated.

We conclude this chapter with the outline in Box 2.2, which asks you to look at the kinds of issues that need to be considered and responded to during the course of developing the research proposal. By and large, your dissertation committee will need to be convinced of three things to be comfortable with your proposal:

1. Is the question clear and researchable, and will the answer to the question extend knowledge in your field of study?

2. Have you located your question within the context of previous study such that you have demonstrated a mastery and consideration of the relevant background literature?

3. Is the proposed method suitable for exploring your question?

Box 2.1 Brainstorming Exercise Begin by defining one or two variables (or constructs) of interest. Then generate a list of additional variables (or constructs) that in some way amplify the original variables or illuminate the relationship between them. The new variables you list may be independent variables, dependent variables, moderating variables, or mediating variables in the research questions you eventually select. After brainstorming this list, go back and eliminate those variables that do not interest you or do not seem promising to pursue. Finally, see if you can now define one or more research questions that speak to the relationship among the two or three variables (or constructs) you have specified. Ultimately, each of these variables will need to be operationally defined as you develop your research study.

Here are some examples of the application of this brainstorming exercise to topics from different disciplines.

Political Science Begin with an interest in citizen participation in city council meetings. List variables or phenomena that might influence, be influenced by, or be related to this variable. A sample research question is “What is the impact of citizen participation in city council meetings on legislative decision making?”

Education Begin with an interest in single mothers who return to school while receiving Aid to Families with Dependent Children (AFDC). List variables or phenomena that might influence, be influenced by, or be related to this variable. A sample research question is “What is the effect of the availability of child care on whether single mothers receiving AFDC return to school?”

Criminal Justice Begin with an interest in the relationship between neighborhood crime watch programs and robbery rates. List variables that might influence or amplify the relationship between these two variables. A sample research question is “What is the effect of neighborhood crime watch programs, in both urban and rural environments, on the rate of burglaries?”

Psychology Begin with an interest in the relationship between physical attractiveness and self-esteem. List variables that might amplify or influence the relationship between these two variables. A sample research question is “What is the role of body image and physical attractiveness in self-esteem?” Another sample research question is “What is the role of body image in mediating the relationship between physical attractiveness and self-esteem?”

Selecting a Suitable Topic: Student Suggestions

Over the years, our students have provided many useful suggestions for completing a dissertation. Here are some of the suggestions they have offered to one another that pertain to the earliest stages of developing a dissertation. Other suggestions are noted at the appropriate places in subsequent chapters.

1. Start a computer file where you can store good ideas for future reference. Use the file for noting books and articles to get from the library, good quotations, inspirations for future studies, half-baked notions that might be useful in the future, and so on.

2. Think of your topic as a large jigsaw puzzle with a piece missing. That piece is what you want to research to fill in the gap in your field. To discover which piece is missing, you must read as much of the literature as possible in your field.

3. Before you begin, read several well-written dissertations recommended by your chairperson. 4. As you progress through the dissertation process and your question shrinks due to the necessity of

maintaining a manageable project for one person, don’t lose heart. Even very small questions can serve much larger purposes.

5. To keep the perspective of meaningfulness throughout, keep imagining an audience of individuals who would want to know the results of your work. Even if you can imagine only 25 people in the world who would care, keep that group alive in front of your eyes.

6. A few students and I have found that beginning dissertation work early on and setting “mile marker dates” is very helpful. This has helped us keep each other accountable by regular “check-ins” with each other. Sharing these dates with friends and family as well as posting them in a conspicuous place so you see them pretty much daily has helped as well.

7. Make contact with researchers who may be of interest to you. It cannot hurt, and it might be quite helpful.

8. Travel to at least one professional meeting where research in your area of interest is being presented. I went to a paper presentation in my area of family violence. I realized that I knew more about my topic than I thought and shared my research ideas with one of the presenters. I came home reinvigorated and anxious to complete my dissertation.

Box 2.2 Outline of Issues for a Student Researcher to Complete in the Development of the Research Proposal

Review of the Literature The classic, definitive, or most influential pieces of research in this area are . . . The journals that specialize in the kind of research in which I propose to engage are . . . The body(ies) of research to which I wish to add is (are) . . . The experts in the field of my research are . . .

Statement of the Problem The intellectual problem(s) I may help solve through this research is (are) . . . The moral, political, social, or practical problem(s) I may help alleviate through this research is (are) . . .

Method The method I propose to use to answer my question, prove my point, or gain more detailed and substantive knowledge is . . . An alternative way to do this study would be . . . Three important research studies that have been carried out using the method I propose are . . . The reason(s) this method is a good one for my question, proposition, or subject is (are) . . . Possible weaknesses of this method are . . . The skills I will need to use this method are . . . Of these skills, I still need to acquire . . . I propose to acquire these skills by . . .

Notes 1. A construct is a concept used for scientific purposes in building theories. Constructs (e.g., self-esteem), like concepts, are abstractions formed by generalizing from specific behaviors or manipulations. When constructs are operationalized in such a way that they can be “scored” to take on different numerical values, they are referred to as variables.

2. There has been extensive discussion in recent literature regarding the definition of, appropriate statistical analyses for, and interpretation of moderating and mediating effects. We recommend the following discussions: Frazier, Tix, and Barron, 2004; Hayes, 2009; Jaccard and Jacoby, 2010; Kazdin, 2007; Kim, Kaye, and Wright, 2001; and Preacher and Hayes, 2008.

3 Methods of Inquiry Quantitative and Qualitative Approaches

The principal characteristic of scholarly and scientific inquiry—as opposed to informal, intuitive kinds of inquiry—is the use of rationally grounded procedures to extend knowledge that a community of scholars regards as reliable and valid. The dissertation process is a ritual of socialization into that community of scholars, so it is necessary for you, as a student, to master the scholarly procedures within your discipline. The specific methods chosen to attack a problem will depend on your discipline and the nature of the specific problem. There is no universally accepted approach within the social sciences, although there are rich research traditions that cannot be ignored, as well as a common understanding that chosen methods of inquiry must rest on rational justification. This means that scientific methods differ from more informal methods of inquiry by their reliance on validated public procedures that have been determined to produce reliable knowledge.

Currently, there are many disagreements in the social sciences regarding what constitutes knowledge and the procedures for gaining it. One way to think about how research generally contributes to the knowledge base of a discipline is in terms of the following three-level hierarchy of knowledge, suggested by our colleague Marilyn Freimuth.

Axiologic/Epistemic Level.

This is the underlying level of basic world hypotheses that form the foundation for content and method within a field of inquiry. Epistemology refers to the study of the nature of knowledge, whereas axiology refers to the study of ethics, values, and aesthetics. Examples of constructs at this level include the explanatory principle of cause and effect and the notion of open systems.

Theoretical Level.

This is the level of models and theories. Theories are premises to account for data or, more informally, explanations of how things work based on

data. Examples are the theory of loss aversion in economics (Tversky & Kahneman, 1991) and the five-factor theory of personality in psychology (McCrae & Costa, 2003). The distinction between theories and models is murky because these terms are often used interchangeably within the social sciences. At the most basic level, both theories and models refer to relationships between concepts. For our purposes, the term model refers to a higher-order theory, that is, a representational system at a higher level of abstraction that can inform and be informed by alternative theories. (This concept is similar to the framework or worldview that guides researchers, identified as a “paradigm” by Thomas Kuhn [1996].) Thus, psychoanalysis could be seen as a model, a wide lens with which to view and understand the mysteries of human behavior. Each model carries with it certain sets of assumptions. In the case of psychoanalysis, these assumptions include the unifying importance of causal determinism and unconscious motivation. Note that this use of the term model differs somewhat from that in the discussion of working models in Chapter 2.

Empirical Level.

In the field of epistemology, empiricism refers to a commitment to obtaining knowledge through sense experience (literally, “based on experience” in Greek). Empiricism is frequently contrasted with rationalism, which refers to knowledge derived purely through thought and reason, and to more natural philosophical and religious traditions of reaching conclusions. In the present context, the empirical level includes hypotheses and methods and data of scientific research. Hypotheses are tentative answers to questions, generally based on theory.

The primary role of research within this three-level schema is to link the theoretical and the empirical. Theories need the support of data to remain viable, whereas methods carry assumptions that are theoretical in nature. Note that research findings do not contribute directly to the axiologic/epistemic level or even to basic models. Those levels reflect fundamental value commitments and personal preferences that are rarely modified on the basis of additional data, especially the kind of data generated by scholarly research. It is hard to imagine a psychoanalyst becoming a behaviorist or a Republican joining the Democrats without a significant shift in values unlikely to be compelled by the accumulated wisdom imparted by a series of research studies. Because most researchers strongly identify with particular values and carry many personal

preferences into their work, it becomes especially important to learn to discriminate between beliefs and opinions, on the one hand, and verifiable, data-inspired support for ideas, on the other hand.

A brief look at the history of science is a humbling experience that should put to rest the misguided notion that research discovers truth. Drilling holes in the skull (trephining) used to be an acceptable way to dismiss the demons responsible for mental illness, and it wasn’t that long ago that the sun was thought to circle the earth. One wonders what remnants of contemporary scientific truth will be regarded as equally ludicrous tomorrow. Instead, what research contributes is a series of thoughtful observations that support or question the validity of theories, which are in turn based on a set of largely untestable beliefs and assumptions. Every once in a while, at opportunistic moments of scholarly upheaval, a new paradigm appears that seems to do a better job of explaining the available data and guiding further inquiry.

Each social science discipline and set of investigators seems to have its own favored approach to generating knowledge. For instance, public opinion studies usually rely on survey research methods, psychoanalytic studies of infants make use of observational techniques, studies of organizational effectiveness may employ action research methods and case studies, historical investigations of political and social events rely on archival records and content analysis, and laboratory studies of perceptual processes stress experimental manipulation and hypothesis testing. Within your chosen field, it is important to ask how a piece of research acquires legitimacy as reliable knowledge. No doubt part of the answer comes down to underlying epistemological assumptions and values. Certainly research strategies will differ in terms of the problems they address and the outcomes they produce. As we later show, one important distinction in the choice of method seems to be the nature of the relationship between the researcher and the topic of study.

We would argue that researchers in the social sciences have generally been myopic in defining the kinds of studies that might legitimately lend themselves to research dissertations. Most students in the social sciences are taught early on about the difference between independent and dependent variables and how experimental research implies active manipulation of independent variables to observe a subsequent impact on dependent variables. This basic and time-honored strategy has an earthy

history in the systematic evaluation of fertilizers for agricultural productivity (Cowles, 2000). It remains a cornerstone in conducting social science research with human subjects. Yet it is certainly not the only way to conduct research.

The only universal in scientific knowledge is a general commitment to using logical argument and evidence to arrive at conclusions that are recognized as tentative and subject to further amendment. Good scientists in action often deviate from an “official” philosophy of science and a prescribed methodology. William Bevan (1991), former president of the American Psychological Association, noted,

If you want to understand what effective science making is about, don’t listen to what creative scientists say about their formal belief systems. Watch what they do. When they engage in good, effective science making they don’t, as a rule, reflect on their presuppositions; they engage in a practical art form in which their decisions are motivated by the requirements of particular problem solving. (p. 478)

The key to evaluating a completed study is to assess whether the selected method is sufficiently rigorous and appropriate to the research question and whether the study is conceptually and theoretically grounded. The more familiar you are with the full range of alternative research strategies, the more enlightened and appropriate your choice of a particular method is apt to be. Too often, students become so enamored with an approach to research that they choose the method before determining the question. Unless the dissertation is designed to illustrate the use of a promising and innovative methodology, this is putting the cart before the horse. In general, the method needs to evolve out of the research question and be determined by it.

Quantitative Methods The epistemological foundation of most social science inquiry throughout the 20th century was logical positivism, a school of thought that maintains that all knowledge is derived from direct observation and logical inferences based on direct observation. To a great extent, the notion of

objectively studying human beings is derived from a love affair that social scientists have had with the natural sciences, which sought to understand nature by isolating phenomena, observing them, and formulating mathematical laws to describe patterns in nature. Current research in the social sciences is deeply steeped in the empirical and quantitative traditions.

Statistical methods are especially useful for looking at relationships and patterns and expressing these patterns with numbers. Descriptive statistics describe these patterns of behavior, whereas inferential statistics draw on probabilistic arguments to generalize findings from samples to populations of interest. Kerlinger (1977) focused on the inferential process when he defined statistics as

the theory and method of analyzing quantitative data obtained from samples of observations in order to study and compare sources of variance of phenomena, to help make decisions to accept or reject hypothesized relations between the phenomena, and to aid in making reliable inferences from empirical observations. (p. 185)

Note that the focus in the natural science model of research is the study of average or group effects, not of individual differences. The kinds of inferential statements that derive from this model of research refer to groups of people or groups of events; that is, they are probabilistic (e.g., “Surveys find that most people believe that police officers use excessive force in dealing with criminals,” or “Emotional expressiveness is related to coping effectively with natural disasters”).

In experimental research, quantitative research designs are used to determine aggregate differences between groups or classes of subjects. Emphasis is placed on precise measurement and controlling for extraneous sources of error. The purpose, therefore, is to isolate a variable of interest (the independent variable) and manipulate it to observe the impact of the manipulation on a second, or dependent, variable. This procedure is facilitated by the “control” of extraneous variables, thus allowing the researcher to infer a causal relationship between the two (or more) variables of interest.

Methodological control is generally accomplished by two procedures that rely on the principle of randomness. One is random sampling, which uses subjects that have randomly been drawn from the potential pool of subjects so that each member of the population has an equal chance or known probability of being selected. Random selection of subjects permits the researcher to generalize the results of the study from the sample to the population in question. The second procedure is randomization, which assigns subjects to groups or experimental conditions in such a way that each subject has an equal chance of being selected for each condition. Subject characteristics are thus randomly distributed in every respect other than the experimental manipulation or treatment, allowing the researcher to infer that resultant differences between the groups must be the result of the isolated variable in question.

Unfortunately, these efforts at experimental control are often impractical in social science research with human subjects. Psychology, for instance, has an honorable tradition of laboratory research using tight experimental designs, but research in the clinical or social arena may not permit the kind of control stipulated by the experimental method. This dilemma is equally prominent in field studies in disciplines such as sociology, education, and political science. Jared Diamond (2005), Pulitzer Prize–winning geographer and biologist, for example, conducted quantitative “natural experiments” to investigate the problem of deforestation on Pacific islands. He and his colleague Barry Rolett numerically graded the extent of deforestation on 81 Pacific islands and statistically predicted this outcome from a combination of nine input variables, such as the amount of rainfall, isolation from human populations, and restoration of soil fertility. In a different context, one cannot practically conspire to rear children using two distinct parenting styles, nor can one ethically inflict child abuse to study its immediate impact in a controlled fashion. Researchers can, however, study analogs of these variables using pure experimental designs (e.g., one can ask parents to use specific interventions at the onset of particular child behaviors). Change studies, in which a treatment or program is being evaluated for its effectiveness, may also lend themselves well to experimental designs. Even so, it may not be possible to randomize subjects into groups that receive a treatment or intervention and those that do not. A number of ingenious solutions have been proposed to deal with the ethics of denying treatment to those who need it, including the use of placebos and waiting-list controls (Kazdin, 2002).

More typically, the research method of choice in the social sciences seems to be a quasi-experimental design that compromises some of the rigor of the controlled experiment but maintains the argument and logic of experimental research (Kline, 2009; Shadish, Cook, & Campbell, 2001). This kind of research uses a systematic, empirical approach in which the investigator does not employ a control group or does not randomly assign subjects to conditions because events have already occurred or cannot be sufficiently manipulated. So-called causal statements become correlational statements in quasi-experimental research, although it is often possible to infer a sequence of events in causal form. That is one reason why it is crucial to have a theoretical model as a foundation for an empirical study. The model itself informs your attempt to meaningfully interpret the results of the study. However, because it is difficult to ascribe causality with confidence from quasi-experimental designs that lack true experimental manipulation, independent variables are often termed “predictor” variables in these studies (Kline, 2009).

Caution is also needed in interpreting the meaning of results whenever subjects assign themselves to groups. A colorful example is the apparent negative correlation that exists between the numbers of mules found in the various states and the number of PhDs living there. The fact that states that have a lot of mules don’t have so many PhDs, and vice versa, is an empirical observation that can be statistically expressed in the form of a correlation coefficient. A researcher would be hard-pressed to argue a causal relationship between these two variables unless he or she drew on an underlying theoretical model that links the two variables through a third (mediating) variable, such as the degree of urbanization. Note that this simple correlational study could, at least theoretically, be transformed into an experimental study by, for example, flooding some states with mules to see if the PhDs leave or wooing the PhDs across state lines to see if the number of mules in the new state of residence decreases.

This is not a book on research design, but the adoption of a particular research strategy will affect the final form of your dissertation. Whether a study employs a true experimental design, a quasi-experimental design, or a cross-sectional survey design, the most popular strategy in the social sciences is a comparison between groups. That is, independent (randomly assigned) groups of subjects are used for each experimental or control condition. The best-known variant of this strategy, the pretest-posttest control group design, uses two equivalent groups of subjects that both

receive pretests and posttests and differ only in the experimental treatment that is given to one group (see Table 3.1).

In this design, it becomes possible to evaluate the impact of an intervention because the control group offers a baseline for comparison. One could use this design to evaluate, for example, whether the inclusion of spouses in an aftercare program for heart bypass surgery patients encourages greater compliance with medical regimens. Or one could design a study to evaluate the effect of introducing air bags in automobiles on the rate of physical injury to passengers. Some automobiles of a given make would receive air bags, some would not, and the change in types and rates of injuries would be the dependent measure.

The straightforward pretest–posttest–control group design makes it possible to attribute the effect of experimental interventions to those interventions rather than to some extraneous variable. The interpretation of results of studies using this design may be compromised, however, if the subjects have not been assigned to conditions in a truly random manner. In the proposed air bag study, for example, if automobiles and drivers are not randomly assigned to conditions, inherently safer drivers may well choose automobiles with better safety features. Because randomization is not always possible, it becomes crucial to argue for the “equivalence” of the two groups, even if they do not derive from the identical population of subjects. One way in which researchers attempt to make this argument is by matching the groups on key variables that are critical to the understanding of the study, such as age, sex, symptomatology, or, in the current example, the previous driving records of the participants.

The basic pretest–posttest–control group design does not adequately control for any effect that the pretest evaluations might have on the subjects. Some assessments can sensitize subjects by making them aware that they are now participating in a study or by providing a practice

experience that contaminates the validity of posttest results. A simple posttest-only design may get around this difficulty and is probably underused (Campbell & Stanley, 2005). In any case, choosing a basic research design does not eliminate the need for you to think carefully and creatively about potential sources of error and alternative explanations to account for findings.

Most experimental designs are variants of the treatment and control group format described earlier.1 Such designs permit the researcher to make causal inferences regarding relationships among the variables. In contrast, correlational (or observational) studies do not generally enable the researcher to demonstrate causal relationships among variables. Any conclusions regarding causality must be inferred from the underlying theory rather than from the results of the study.

Studies built around experimental or correlational designs generate data that are subsequently analyzed using appropriate inferential statistics. Statistical techniques that are used to evaluate the effectiveness of an intervention or a difference between groups, such as an analysis of variance (ANOVA) or t test, compare the size of between-group differences (e.g., the treatment effect) with the size of within-group differences due to individual variability. These techniques express the experimental tradition. The logic of the correlational paradigm is quite different (Cronbach, 1975). Correlations depend on comparing two distributions of scores that are broadly dispersed along two dimensions, such as longevity and alcohol use. Statistical techniques that emerged from this tradition, such as multiple regression, are especially popular in social science research that relies on questionnaires, surveys, or scales and the relationship between continuous variables. Because correlational studies typically cannot randomly assign subjects to groups, we have a second major type of control in social science research—statistical control. Statistical control attempts to use complex statistical procedures to remove variability from measures of group difference or relationship that could be attributed to variables other than the major independent variables of interest. Be aware, however, that it is the design of the study and not the choice of statistical method that principally governs the types of statements that can be made about the relationships among variables.

Both experimental and correlational traditions have a rightful place in the evaluation of quantitative data, and a detailed comparison of them goes

beyond the scope of this book. It is important to remember that although statistics is an indispensable tool for scientific inference, the appropriate application of statistics cannot make up for a faulty research design. In many instances, statistical methods drawn from both the experimental and correlational paradigms are equally legitimate choices. In fact, the same data usually can be analyzed in multiple ways. If you are looking at the relationship between locus of control and frequency of medical visits for preventive health, for example, you could express this relationship using a correlation coefficient or by dividing your sample into two or more subgroups on the basis of the personality construct of locus of control and comparing the resulting groups on medical visits. Generally speaking, it is not a good idea to “throw away” data (you are throwing away data if you arbitrarily reduce a continuum of locus of control scores to two or more discrete values, such as internal or external categories), but these kinds of decisions require statistical expertise and theoretical grounding.

Table 3.2 summarizes the methodological and statistical methods of controlling for extraneous factors in a research design.

We wish to make two additional points regarding the use of quantitative research. One is that there is a tendency in the social sciences to overemphasize the importance of statistically significant findings and to underemphasize the importance of clinically or socially significant findings. In other words, simply because a difference is significant at a certain probability level (typically,.05 or.01) does not mean that the difference is meaningful in practical terms. For instance, a difference of 5 points on a depression scale might be statistically meaningful but may not be meaningful clinically. Too often, students assume that the object of research is to achieve statistical significance rather than to make meaningful inferences about behavior. The primary reason that Jacob Cohen (1990), the father of power analysis, was drawn to correlational analyses is that they yield an r, a measure of effect size. That is, unlike probability (p) values, correlation coefficients can straightforwardly indicate the magnitude of the relationship between variables, which may be far more informative than the presence or absence of statistical significance. Cohen went on to note that researchers sometimes learn more from what they see than from what they compute, and he argued for an increased use of the graphic display of data, using simple scatter plots and so-called stem-and-leaf diagrams before or instead of performing complicated statistical analyses. (We have more to say about this topic in

Chapter 6, our discussion of strategies for presenting results.)

Second, as you consider the kinds of designs and controls that are available to the social science researcher, we urge you to be aware of a fundamental dilemma. Good research is a constant balancing act between control and meaningfulness. At one extreme is an emphasis on controlling the observation and measurement of a variable by eliminating the influence of as many confounding variables as possible. What results might be a tight laboratory study in which the findings inspire confidence but are not particularly interesting. At the other extreme is the observation of complex human behavior in the field, without invoking any controls, so that the results seem fascinating but are highly unreliable and difficult to replicate. The fashion in social science research has moved back and forth between these poles of emphasizing precision of measurement and generalizability of findings versus emphasizing depth of coverage and description of context. Today the pendulum seems to be swinging in the direction of meaningfulness, hastened by the availability of a greater number of permissible research strategies together with a reevaluation of

research epistemology.

Qualitative Methods The researcher who employs experimental and quasi-experimental designs attempts to control the playing field of the study as much as possible, restrict the focus of attention to a relatively narrow band of behavior (often manipulating experimental conditions to further narrow the object of study to a single variable), and do no harm as a detached and objective observer of the action. A countervailing trend in social science research calls for sidestepping the artificiality and narrowness of experimental studies by promoting methods of inquiry that allow researchers to be more spontaneous and flexible in exploring phenomena in their natural environment. Some of these methods of inquiry challenge the epistemological and philosophical foundations of traditional social science research, which is more compatible with a research culture that maintains a belief in a knowable world, universal properties of social behavior, and the attainment of truth through method (K. J. Gergen, 2001). The commitment to a logical empirical approach to research is not necessarily seamless with a postmodern worldview, which challenges the sanctity of the scientific method as a vehicle for attaining truth and promotes an awareness that beliefs and apparent “realities” are socially constituted rather than given and, therefore, can show up differently in different cultures, times, and circumstances (Neimeyer, 1993). The term constructivism is a name for the epistemology associated with the view that what people may consider objective knowledge and truth are a result of perspective. For the constructivist, knowledge is not “found” or “discovered” from existing facts but constructed as the invention of an active, engaging mind.

There are many flavors of constructivism,2 but they all focus on how humans create systems of meaning to understand their world and their experience. The term social constructionism is usually used to refer to the fact that meaning is typically created not by individual cognitive processes but within human relationships as part of a social exchange process. Thus, a focus on the isolated knower is replaced by an emphasis on how knowledge is situated within, and is dependent upon, historical factors, cultural factors, and contextual factors. Describing a person or an event is not mirroring “what is out there” but understanding how meaning is socially constructed and mediated by language and values.

Qualitative methods are usually linked to a constructivist theory of knowledge because qualitative methods tend to focus on understanding experiences from the point of view of those who live them. But that is not necessarily the case. The world of qualitative research is rich with alternative perspectives. At one extreme are those who, in their questioning of the validity of a logical positivistic science applied to human behavior and social systems, take issue with the ideal or even the possibility of having a neutral, disengaged investigator (see Feyerabend, 1981; Popper, 1965; Toulmin, 1972). Taking their hint from modern physics, they suggest that the presence of an observer inevitably alters that which is being observed—that, in fact, one cannot separate the investigator from the object of inquiry. Feminist theorists have other reasons for criticizing the traditional experimental method, claiming that it creates a hierarchy of power in which the omnipotent researcher, often a male, instructs, observes, records, and sometimes deceives the subjects (Peplau & Conrad, 1989). It should be noted, however, that whether a study uses experimental or nonexperimental methods does not necessarily imply anything about the researcher’s commitment to nonsexist research.

The impact of these developments in the philosophy of science on method has been profound, especially within the last 20 years or so. A host of alternative research paradigms have evolved and are now being applied to dissertation research in the social sciences. The labels given to these approaches include “phenomenological,” “hermeneutic,” “naturalistic,” “experiential,” “dialectical,” and so on. The generic label most commonly used to incorporate these diverse research strategies is “qualitative research.” Crotty (1998) maintained that the fundamental distinction between quantitative research and qualitative research is seen at the level of method, rather than the level of theory or epistemology. Moreover, qualitative researchers do not possess a distinct set of methods that are all their own (Denzin & Lincoln, 2011). They can make use of interviews, text analysis, surveys, participant observation, and even statistics. Over time, different research traditions have evolved that bring to bear particular perspectives from which to investigate particular topics, such as psychoanalytic studies of children and ethnographic studies of cultures. Within these domains, the researcher may draw on many specific methods; an example is the ethnographer who employs both interviews and observational descriptions. In general, qualitative research implies an emphasis on processes and meanings over measures of quantity, intensity, and frequency (Denzin & Lincoln, 2011). As suggested earlier, the newer

generation of qualitative researchers emphasizes the socially constructed nature of reality, a close relationship between the researcher and the object of study, and the context that influences the inquiry.

The boundaries between quantitative research and qualitative research have become increasingly fuzzy as various disciplines have adopted their own perspectives on adapting methodologies to serve their needs. At the risk of overgeneralization, we are listing eight distinctions between quantitative and qualitative research that are often highlighted. These distinctions are also summarized in Table 3.3.

1. The most obvious distinction is that data in quantitative studies are expressed in numbers, where numbers are a metric for measuring, describing, testing, and generalizing about variables of interest to the researcher. In qualitative research, the currency of choice is words. However, in some qualitative (or hybrid) studies, those words may be coded, categorized, expressed in numerical form, and analyzed quantitatively.

2. Quantitative research tends to use the hypothetico-deductive approach to research design, which prescribes specification of variables and hypotheses before data collection. Counterexamples include survey research methods and factor analytic studies that are more exploratory and rely on inductive rather than deductive procedures to interpret findings. In contrast, qualitative research begins with specific observations and moves toward the identification of general patterns that emerge from the cases under study. The researcher does not impose much of an organizing structure or make assumptions about the interrelationships among the data prior to making the observations. This is not to imply, however, that the study is not thoroughly planned.

3. The quantitative researcher usually tries to control the site and context of the study to focus on a limited number of variables. This is particularly true in experimental laboratory research and, to a lesser extent, quasi- experimental studies. The qualitative researcher, on the other hand, is intent on understanding phenomena in their naturally occurring context with all of its inherent complexity. Just because a study is conducted in the field, however, does not mean that it is necessarily qualitative in form.

4. Quantitative research seeks to define a narrow set of variables operationally and isolate them for observation and study. This contrasts with qualitative research, which is more holistic and aims for a psychologically rich, in-depth understanding of a person, program, or situation by exploring a phenomenon in its entirety.

5. Quantitative research seeks objectivity and pursues this ideal by standardizing procedures and measures as much as possible and by distancing the researcher from the participants. The qualitative researcher values the subjectivity of the participants and sees their unique characteristics not as “error” to be removed or minimized but as valued aspects of the research situation.

6. The aim of quantitative studies is prediction, control, or explanation/theory testing, or all three. Predicting under what circumstances events lead to other events or variables are associated with other variables helps to explain important phenomena in the social sciences. In qualitative studies, the goal is more likely to focus on description, exploration, search for meaning, or theory building. Qualitative research tends to be a discovery-oriented approach.

7. The stance of the researcher is different in qualitative research than in quantitative research. The quantitative researcher drives the study by manipulating and controlling the conditions of the study as well as the information provided to the research participants. The qualitative researcher usually invites the subject to participate, sometimes as a formal collaborator, by contributing knowledge about unobservable aspects of his or her experience that are not accessible to the researcher in other ways.

8. Quantitative research relies on statistical analysis to analyze data. This includes the use of descriptive and inferential statistics to determine the relationship between variables or the significance of group differences or the effect of an intervention. In qualitative research, some kind of text analysis is employed to categorize responses and identify themes, which are then evaluated subjectively to shed light on a phenomenon of interest. Although individual differences may be explored to further understand the phenomenon, those differences between individuals or groups are usually not the focus of the study. Instead, they are used to build theory or add to theory development.

The appropriate selection of methods of inquiry is contextual and depends to a large extent on learning the standards used in your own discipline. For example, qualitative methods have an especially comfortable home in the ethnographic and field study traditions of anthropology and sociology that emerged in the 19th century. Psychologists and psychiatrists also developed detailed case histories of their patients at about that time. Today, qualitative dissertations are widespread, although any classification of qualitative methods is apt to be a simplification. The following approaches are frequently adopted in contemporary social science dissertations: phenomenological research, ethnographic inquiry, grounded theory, and narrative research. They are described in more detail throughout the book.

As Crotty (1998) clarified, all methodologies and methods (methodologies

can be regarded as the strategies, action plans, or designs that inform the choice of specific methods, that is, procedures and techniques for data collection and analysis) flow from philosophical positions that provide a theoretical context for the choice of methodology. Theories and methods need to be logically linked. However, it is also possible for different theoretical perspectives to employ very similar methods. For instance, case studies have a rich tradition in the literature as a method of collecting data. But there are big differences between observing a well-known political figure to learn about campaign tactics, interviewing the Dalai Lama about the role of spirituality in world affairs, and measuring the social behavior of an autistic child before and after a treatment intervention. All of these examples can formally be described as case studies, but they emanate from different perspectives on research. We remind you to be tolerant of overlapping categorizations because there is considerable inbreeding among research paradigms. And we urge you again to select methods, regardless of their source, based on their sensitivity and application to the research questions you are asking.

Phenomenology Phenomenologists take issue with positivist science and maintain that the scientific world is not the “lived” world that we experience on a daily basis. Edmund Husserl (1970), the reputed founder of phenomenology, argued that traditional science distances people from the world of everyday experiences. By setting aside theories, conceptualizations, and hypotheses, one could begin with a direct and unbiased appreciation of pure human experience. As such, the phenomenological movement was inspired by Husserl’s well-known dictum, “[Back] to the things themselves!”

The reader who seeks a historical perspective of the philosophical basis of phenomenology is referred to analyses by Crotty (1998), Giorgi (2009), and Gubrium and Holstein (1997). Crotty, in particular, maintained that the practice of phenomenological research, especially in North America, has evolved to the point that the everyday experiences of participants are accepted much more subjectively and uncritically than the theory of phenomenology would suggest. Gubrium and Holstein discussed how phenomenology has become a philosophical basis for interpretive research strategies that include ethnomethodology (the study of the meaning of ordinary talk and social interactions) and conversational analysis (the study of the structure of such talk and interactions).

As it is most commonly understood, the focus of phenomenological research is on what the person experiences and its expression in language that is as loyal to the lived experience as possible. Thus, phenomenological inquiry attempts to describe and elucidate the meanings of human experience. More than other forms of inquiry, phenomenology attempts to get beneath how people describe their experience to the structures that underlie consciousness, that is, to the essential nature of ideas. Phenomenologically oriented researchers typically use interviews or extended conversations as the source of their data. Important skills for the researcher include listening, observing, and forming an empathic alliance with the subject. The investigator remains watchful for themes that are presented but resists any temptation to structure or analyze the meanings of an observation prematurely. Once the basic observations are recorded, the data may be reduced, reconstructed, and analyzed as a public document.

Most writers distinguish between at least two strains of phenomenological research (Polkinghorne, 2010). One, called “empirical” phenomenological research, is directly descended from Husserl’s philosophical position. It is represented by a tradition of studies from Duquesne University, starting with van Kaam’s (1966) study of “feeling understood.” Giorgi’s (2009) ongoing work is illustrative: The researcher collects naïve descriptions of a phenomenon from open-ended questions and dialogue with a participant and then uses reflective analysis and interpretation of the participant’s story to describe the structure of the experience. This is in line with Husserl’s observation that the mind identifies objects as indicators of categories rather than as raw sensory data.

The second primary type of phenomenological research is “existential” or “interpretive” phenomenological research (Polkinghorne, 2010). This increasingly popular approach within the research community draws upon the existential contributions of Heidegger, a student of Husserl. Heidegger was interested in the uniqueness of individuals rather than the classification schemes found across people. Interpretive phenomenology refers to how different individuals understand and give meaning to similar life events. An example might be how different participants in a study understand and relate to the experience of returning home from military service.

Clark Moustakas (1994), one of the founding fathers of phenomenological

research, referred to his own version of phenomenological inquiry as heuristic research, meaning “to discover” or “to find.” The process begins with a question or a problem that is personally meaningful to the researcher in terms of understanding the relationship between oneself and world. Moustakas’s early study of loneliness serves as an example. According to Moustakas, heuristic research has a somewhat different flavor than does the Duquesne approach: The process maintains closer contact with the individual stories of the participants than does structural analysis. At the same time, it is broader in scope than a single situation in the life of a participant and may go beyond narrative description to include stories, self-dialogues, journals, diaries, and artwork as sources of data.

Several of our doctoral students have developed dissertations based on phenomenologically oriented qualitative interviews. As one example, Lauri Francis (2012), for her dissertation in educational leadership, used interviews and a writing activity to determine which pedagogical experiences impact the ability of teaching leaders to nurture the implementation of academic rigor in the classroom. Another student explored how people make meaning from experiences of unanticipated mortal danger. Veronica Clark (1997) conducted open-ended interviews with 10 participants who had experienced life-threatening events in the arena of sports. Her analysis of and reflection on these interviews, presented in both prose and prose trope (a form of narrative poetry), revealed how the events had forced participants to experience multiple realities and get to a deeper understanding of the layered human experience. Finally, Sharon Sherman (1995) completed a largely phenomenological dissertation on the meaning of living with asthma. Her interviews with asthmatic adults led to the development of a conceptual model by which to understand this experience.

Ethnographic Inquiry The ethnographic paradigm includes anthropological descriptions, naturalistic research, field research, and participant observations. Ethnographers attempt to capture and understand specific aspects of the life of a particular group by observing their patterns of behavior, customs, and lifestyles. The focus is on obtaining full and detailed descriptions from informants about ordinary behavior within naturally occurring settings. There is a strong emphasis on exploring the nature of a specific social phenomenon rather than testing hypotheses (Atkinson & Hammersley,

1994). Ethnographers tend to work with uncoded, unstructured data to produce explicit interpretations of the meanings of human actions. Ethnography is a prominent research method within the fields of cultural anthropology and sociology.

Ethnographic inquiry can be found on a continuum ranging from relatively pure description to more theoretically guided explanations of cultural, social, and organizational life. On the more inductive end of the continuum, the researcher develops theory out of the descriptive and interpretive process; on the deductive end of the continuum, the researcher builds a study out of an established theoretical framework. Typically, the ethnographer initiates prolonged contact and immersion in a setting of interest while maintaining as much detachment as possible from the subject matter. The naturalistic setting could be the mental hospital explored in the work of Erving Goffman (1961) over 50 years ago or the street corner populated by unemployed Black men in a classic study by Liebow (1968/2003). More traditional anthropological examples would be a study of health practices among Native Americans living on a reservation or the immersion in non-Western cultures by Mead, Malinowski, or Franz Boas, the renowned ethnographer who advanced social relativism as the prevailing form of American anthropology. The investigator might obtain some preliminary understanding of the history of the culture by referring to archival records and artifacts in preparation for living among the informants for several months. During the time in the field, the researcher would keep field notes of all observations and interactions and perhaps follow up the observations with intensive, qualitative interviews. The data are recorded verbatim, if possible, using the language of the participant, and then reduced for analysis and presentation. More detailed overviews of ethnographic research can be found in contemporary texts such as Schensul, Schensul, and LeCompte (2013) and Fetterman (2010).

When conducting ethnographic studies, there is a fundamental tension between being an objective, detached observer and an emotionally involved participant. The researcher simultaneously adopts two distinct roles while trying to understand the actions, beliefs, and knowledge of a particular group of people (the insider perspective is called emic, and the outsider perspective is called etic). George Herbert Mead (1934), a social psychologist and philosopher from the late 19th and early 20th centuries, argued that to “enter the attitudes of the community” one must “take the

role of others,” and this adoption of the perspective of others found its way into ethnographic inquiry. Today, ethnography is being transformed by an infusion of critical inquiry, which means going beyond trying to understand a culture to addressing political dimensions within it (Crotty, 1998). Thus, whereas traditional ethnography positions the researcher in the background of the study as an objective recorder of facts (i.e., the “realist” position), some contemporary ethnographers take a position of advocacy toward their subjects, who often represent marginalized groups in society. The latter is known as the “critical” perspective (Madison, 2012). Another extension of ethnographic inquiry is called autoethnography, whereby the researcher becomes the object of study. Stacy Holman Jones (2005) showed how a qualitative researcher might subject his or her own gender, class, and cultural beliefs and behaviors to the same study as those of other participants.

Ethnographic inquiry was the basis of Sarah MacDougall’s (2005) creative dissertation on the transformational capacity of a contemporary group process called PeerSpirit circling. MacDougall drew on evidence from ancient and contemporary indigenous cultures indicating the efficacy of circle council as a means of effective problem solving, and used focus groups, participant observation, interviews, and autoethnography to demonstrate how the practice fosters personal transformative experiences that lead to collaborative social action. In an excellent dissertation in sociology at the University of California at Santa Cruz, Rebecca Scott (2007) attempted to understand how the culture of West Virginia coalfields contributes to endorsement of mountaintop-removal coal mining, which leads to both environmental and social destruction. Her study involved spending time in the coal-mining culture and interviewing the stakeholders.

Grounded Theory One of the more prominent types of qualitative research is referred to as grounded theory. In Crotty’s (1998) opinion, grounded theory is a form of ethnographic inquiry that relies on a clearly formulated series of procedures for developing theory. When researchers use the term grounded theory, they are usually referring to those analytical steps (described in Chapter 7), but the term can also apply to a method of inquiry itself. As such, grounded theory has its roots in the theory of symbolic interactionism, which also influenced ethnographic inquiry

(Crotty, 1998). Symbolic interactionism evolved as a pragmatic approach to the study of social interactions through the original contributions of George Herbert Mead. The theory argues that every person is a social construction; that is, people become persons through their interactions with society, using the vehicles of language, communication, and community. From the social interactionist perspective, the researcher must put himself or herself in the role of the other person to view the world from that person’s perspective and understand the meaning of his or her actions (Crotty, 1998).

As a research methodology, the grounded theory approach is a way of conceptualizing the similarities of experience of an aggregate of individuals. It is a discovery-oriented approach to research that offers a set of procedures for collecting data and building theory. The researcher has a research question but rarely a set of theoretical propositions or hypotheses to color the interpretation of findings that emerge from the study.

Grounded theory became popular as a research methodology through a successful 1967 book by Glaser and Strauss. A few years thereafter, the authors ended their collaboration and published independently, Strauss with his colleague Juliette Corbin (Strauss & Corbin, 1998) and Glaser (1998) on his own. The differences between their approaches make for interesting reading (see, e.g., Rennie, 1998). One of the key differences is the extent to which theory is truly discovered, without the preconceptions of the researcher, as opposed to verified, as is more the case in the traditional hypothetico-deductive paradigm. Thus, some grounded theorists became concerned that Strauss and Corbin, among others, became overly prescriptive in developing elaborate coding procedures for analyzing qualitative data. These coding procedures were seen as adding a deductive element to the research process because the categories themselves may reflect the researcher’s interests and biases. The antidote, from the traditional grounded theory perspective, was to immerse oneself in the lived experience of the participants (i.e., the data) in a more direct but flexible way.

To make matters even more complicated, whereas most authorities view Strauss and especially Glaser as quite positivistic and objective in their orientations to research, recent writers are more explicitly constructivist and postmodern. Willig (2013), for example, has noted that the term discovery implies that the researcher is seeking to find meaning that is

already extant within the data, whereas meaning does not emerge from a phenomenon but is always constructed by the researcher in an interaction with the data. Thus, one cannot completely avoid the influence of the researcher on the interpretation of the data, no matter how disciplined one’s attempt to do so. This social constructivist wing of contemporary grounded theory research is well illustrated by the work of Charmaz (2005, 2014), who very clearly focuses on interpreting a phenomenon rather than reporting it or verifying it. She insists that theory that is generated in grounded theory research is shaped by the researcher and derived through deliberate interaction with the data. The resulting theory, then, is inevitably only one slice of the pie, so to speak, rather than the only “truth.” Charmaz (2005) also made the point that, in contrast to her orientation, grounded theory methodology originally gave researchers a way of doing qualitative studies with positivist approval.

As a student, it may not matter so much which approach to grounded theory methodology you adopt, as long as you have a good understanding of what you are doing, why you are doing it, and that you are doing it consistently. The procedures for conducting a grounded theory study are presented in more detail in Chapter 5.

A good example of the classic grounded theory approach espoused by Glaser and Strauss (1967) is Victor Chears’s (2009) dissertation Taking a Stand for Others. The author allowed the theory to emerge from the data rather than attempting to verify any preconceived concepts in his exploration of “standers,” individuals who assume leadership roles with respect to other individuals or organizations explicitly to help facilitate important transitions in others’ lives. The theory comes from the strategies adopted by the standers as they built these relationships and were present and available to help develop the capacities of their clients. A dissertation by Virginia Hedges (2003) used a grounded theory approach to examine the journeys of Latino students who were unusually successful in navigating the public school system. Data from one-on-one, open-ended interviews were analyzed using the constant comparative method (see Chapter 7). A grounded theory consisting of the conceptual categories of encouragement, familia, meaningful relationships, and goal orientation emerged that described a process by which Latino students enhance their cultural identity. Another example of a grounded theory dissertation is Candice Knight’s (2005) exploration of significant training experiences that contributed to the perceived competency development of exceptional

humanistic psychotherapists. Transcribed data from videotaped interviews with 14 participants from throughout the United States and Canada led to the emergence of a multivariate theoretical training model.

Narrative Inquiry We have added narrative inquiry as a fourth major qualitative methodology, in part because of its increasing visibility in the research literature and because many of our students seem to be employing this model for their dissertations. Simply put, narrative inquiry can be regarded as a qualitative methodology that deals with biographic experiences as narrated by the person who has lived them (Chase, 2012). Forerunners of narrative inquiry include the life history method espoused by sociologists and anthropologists early in the 20th century. Life histories are often based on extensive autobiographies from noteworthy cultures or subgroups. Lewis’s (1961) well-known study of a Mexican family, published as The Children of Sanchez, introduced the “culture of poverty” as a concept. Other influences on the development of narrative inquiry include sociolinguists who have studied oral narratives of everyday experience and feminists who have addressed the distinctiveness of women’s narratives. An example of the latter is Belenky, Clinchy, Goldberger, and Tarule’s (1986) honored study Women’s Ways of Knowing.

According to Chase (2012), a narrative may be oral or written and derived from naturally occurring conversation, an interview, or fieldwork. It can be a story that refers to a specific event, such as a job interview or a romantic liaison; it can be a story that reflects on an important life issue, such as athletics or dying; it can even be a story about one’s entire life. What is distinct about the contemporary narrative approach to research is the focus on meaning making, as opposed to merely documenting a history or an experience. Narrative researchers need considerable training in interviewing skills because they must draw out and listen to the thoughts, feelings, and interpretations of the narrator as he or she constructs and organizes previous life experiences. Each person’s narrative is unique, not only because of the uniqueness of that person’s thought processes but also because of the uniqueness of the setting in which it is produced. Chase (2005, p. 657) referred to narratives as “socially situated interactive performances” to capture the notion that narratives are a product of a narrator and the listener coming together at a specific time and place for a specific purpose.

In the final stages of narrative inquiry, researchers also become narrators, as they interpret and make sense of the narratives they have elicited. In this endeavor, the subjectivity of the researcher and of those who are studied is part of the research process. The researcher’s reflections, including how he or she makes interpretations and judgments, become part of the data pool and are also documented. This turning back and reflecting on oneself is known as reflexivity (Josselson & Lieblich, 2003) and has become a fundamental construct in contemporary narrative research.

Specific approaches to narrative research may differ somewhat depending on the academic discipline (Chase, 2012). Psychologists tend to emphasize the content of stories and may be interested in the relationship between life stories and the process of identity development (i.e., the life and the story differ from, but may impact, one another). For example, a dissertation by Denise Humphrey (2003) used a narrative approach to explore the intricate relationships of women who had been adopted in a closed adoption system with their adoptive mothers, birth mothers, and biological children. Humphrey interpreted the interview narratives of these women through the lens of Kohut’s (1978–1991) concepts of self-object needs and functions. Humphrey concluded that becoming a mother serves a restorative function for the adoptee that helps overcome deficiencies in the adoptive process. A second student, Ellen Schecter (2004), observed that little is known about how women in general, and lesbians in particular, negotiate sexual fluidity in terms of their sexual identity. Through in-depth, qualitative interviews, she examined the experience of long-time lesbians who, in midlife, became intimately partnered with a man. Common themes in the narratives were found, leading to a new conceptual model that shows how social and personal constructions are used to create idiosyncratic sexual identities that fit the individual.

In contrast to these psychological studies, sociologists may focus on how participants construct their experience within specific institutional or organizational contexts (i.e., narratives as lived experience) or how they understand certain aspects of their lives. An example is Catherine Riessman’s (1990) classic study of men’s and women’s divorce stories. The link between narrative inquiry and ethnography has been captured best by anthropologists who become involved with one or more members of a community over time and construct narratives about those encounters.

Dissertation Implications of Qualitative Research

The distinctiveness of qualitative research has implications for the write- up of the research proposal and dissertation. Qualitative research designs typically are not intended to prove or test a theory, and it is more likely that the theory will emerge once the data are collected (an inductive approach rather than a traditional deductive approach). This does not mean that the researcher can ignore the theoretical perspectives of previous work cited in the literature review. Note, however, that some qualitative researchers discourage the consideration of any theoretical knowledge based on inferences from existing research before analyzing data from the proposed study. We are in general agreement with Miles and Huberman (1994), who take a moderate position on the role of theory in naturalistic studies. They view a conceptual framework as the “current version of the researcher’s map of the territory being investigated” (p. 20). This means that the framework may change as the study evolves. The amount of prestructuring depends on what is known from the literature about the phenomenon being studied, the measures or instruments that are available, and the time allotted for the study. Very loose designs imply the collection of great amounts of data that may initially look important but turn out to be tangential or irrelevant, along with great amounts of time to sift through these data. At the very least, a conceptual framework allows different investigators who are exploring a similar phenomenon to communicate with one another and compare experiences and results.

Adopting a tentative conceptual framework allows the researcher to focus and bound the study with regard to whom and what will and will not be studied. Miles and Huberman (1994) chose to express their conceptual frameworks in terms of graphic “bins” that consist of labels for events, settings, processes, and theoretical constructs. They reasoned that the researcher will come to the study with some ideas about the content of these bins. For instance, a qualitative study on prison behavior could reflect working decisions focusing on current behavior rather than prior history (events), high-security prisons (settings), interactions among prisoners and between prisoners and guards (processes), and authority relations and organizational norms (theoretical constructs). These choices and distinctions are, of course, informed by the theoretical and empirical literature.

Research questions can then be formulated as a way of explicating any theoretical assumptions and orienting the investigator (and the student’s committee) to the primary goals and tasks of the study without dampening

the process of curiosity and discovery. For example, one cannot study every aspect of prison life. Furthermore, the issues adopted by the researcher and expressed as research questions have direct implications for the choice of methodology. A focus such as “how prisoners and guards negotiate conflict and express power in relationships” has implications for the behavioral events that will be sampled and the research tools that will be used to obtain information (e.g., field notes, interview transcripts, diaries, prison documents). Research questions in qualitative research can be revised or reformulated as the study proceeds.

Students selecting a qualitative design need to convince their committees that they understand the role of the qualitative researcher. This includes experience with the sensitive kind of interviewing found in naturalistic studies, whereby the investigator enters the world of the participant subject without a fixed agenda and maintains sufficient scientific rigor in the process. Because the researcher is regarded as a person who comes to the scene with his or her own operative reality, rather than as a totally detached scientific observer, it becomes vital to understand, acknowledge, and share one’s own underlying values, assumptions, and expectations. This perspective should become clear in the Review of the Literature and Method chapters of the dissertation. Moreover, researcher subjectivity can be reduced by a variety of data-handling procedures. Will there be audio- or videotaping to augment written field notes? How will these materials be reduced in scope? Will process notes be included that describe the researcher’s reactions at various points of the study? Will pilot studies be used to test the suitability of procedures? Will conclusions be provided to informants for verification prior to publication (member checking)? Specification of these ingredients can be convincing documentation of the rigor of the proposed study that do not compromise the necessary open contract of the proposal.

Because qualitative data may consist of detailed descriptions of events, situations, and behaviors, as well as direct quotations from people about their experiences and beliefs, the Results chapter of the dissertation will be directly influenced as well. We have found that students often mistakenly believe that a qualitative study might be easier to conduct because there are no specific hypotheses and no statistical tests to perform. However, the sifting and resifting of transcripts with huge amounts of open-ended responses into a coherent pattern generally takes as much effort and leads to as much frustration as the statistics that were being avoided. Good

research is always taxing in some way.

Other Possible Approaches to the Dissertation

Hermeneutics Hermeneutics has been described as the interpretation of texts or transcribed meanings (Polkinghorne, 2000). One engages in a hermeneutic approach to data to derive a better understanding of the context that gives it meaning. Hermeneutics, as a specialized field of study, was pioneered by biblical scholars in the 17th century who used textual analysis and interpretation to elicit the meanings of religious text. More recently, researchers in the social sciences, as well as scholars in the field of literary criticism, have extended the application of hermeneutics to the interpretation of secular texts.

There is ongoing debate within the field of hermeneutics between objectivists, who consider the text to contain meaning independent of the interpreter, and others, who view active interpretation as primary to all understanding. The latter position is quite similar to modern constructivist thinking in the philosophy of science (Winograd & Flores, 1986). From this orientation, understanding is the fusion of the perspective of the phenomenon and the perspective of the interpreter. Everyone brings life experiences and expectations to the task of interpretation, but because even people’s self-understanding is limited and only partially expressible, interaction with the meaning of the text can produce a deeper understanding of both the observer and the observed. As Mahoney (1990) put it, “New or changed meanings arise from the active encounter of the text and its reader” (p. 93).

Texts from ancient cultures, for instance, may be analyzed in their historical context with the goal of applying their meanings to current issues. This understanding, which must show the meaning of a phenomenon in a way that is both comprehensible to the research consumer and loyal to the frame of reference of the subject, may then lead to more formal research questions. In hermeneutics, the data are given to the researcher, whereas in a standard phenomenological study, the researcher helps to create the transcribed narrative, which has usually been obtained by interviewing the participant(s) (Hoshmand, 1989). As we have

seen, phenomenological research can have a hermeneutic basis that is more interpretive than descriptive. A good example of a dissertation taking this approach is Smith’s (1998) study of how family/divorce mediators can remain internally balanced and focused while trying to resolve challenging disputes between separating partners. Smith conducted three in-depth interviews with seven nationally recognized mediators and performed an inductive analysis of the interview transcripts that revealed layers of voices existing within the mediators’ consciousness. Hermeneutic phenomenology, as a research method, can also make use of data sources such as literature, poetry, visual arts, and video while retaining the participants’ oral or written descriptions of their experiences (Hein & Austin, 2001). At the dissertation level, this kind of hermeneutic approach is exemplified by J. M. Elliott’s (1997) study of five Renewal of Canada conferences, in which the materials that were studied included videotapes, formal and informal papers and reports, press releases, and media coverage of the conference workshops and meetings. The outcome is an understanding of the conditions that contribute to or hinder the quality of the communicative interaction in a discursive attempt to bridge differences.

A hermeneutically informed approach to research is quite complex. Because language is regarded as the core of understanding, the researcher needs to return repeatedly to the source of data, setting up a dialogue with it, so to speak. The investigator asks what the data mean to their creator and tries to integrate that meaning with their meaning to the researcher. This kind of inquiry is sometimes referred to as the “hermeneutic circle method,” originally proposed by Wilhelm Dilthey (1996) in the 19th century as a series of steps to educe how the meaning of an entire text informs the meaning of segments of the text and how the meaning of segments of the text elucidate the meaning of the entire text. Whereas Dilthey took an objectivist stance in trying to create a “science of subjectivity” that could be used to reconstruct the meaning of texts, subsequent hermeneuticists such as Gadamer (2013) and Habermas (Habermas & McCarthy, 1985) amended these ideas to acknowledge that we can never really get into the mind of the writer of a text. Our interpretation must be grounded in understanding our own situational circumstances because no single correct interpretation or objective meaning exists (Packer, 2010). Although we are all hermeneutically inclined whenever we seek to learn the contexts of things, ideas, and feelings, hermeneutic inquiry is relatively rare as a formal approach to

research in the social sciences. Ambitious, well-known examples of hermeneutic studies are psychodynamically guided biographies, such as Erik Erikson’s Young Man Luther, and the work of Carl Jung, who used an archetypal, mythic perspective to describe contemporary problems.

It can be argued that hermeneutics is more of a theoretical perspective than a particular research methodology. According to Martin Packer (1985, 2010), the hermeneutic approach is applicable to the study of all human action, where the action is treated as though it has a textual structure. The investigator studies what people do when they are engaged in everyday, practical activities. What sets hermeneutics apart from more empirical or rational orientations to the study of human behavior is the belief that a particular activity can be understood only in conjunction with understanding the context in which it occurs rather than as an abstraction or a set of causal relationships. As Packer (1985) put it,

The difference between a rationalist or empiricist explanation and a hermeneutic interpretation is a little like the difference between a map of a city and an account of that city by someone who lives in it and walks its streets. (p. 1091)

The mapmaker’s product is formal and abstract; the inhabitant’s map is personal and biased.

David Rennie (2012), moreover, recently proposed that all qualitative research can be seen from the perspective of “methodical hermeneutics.” Rennie divided qualitative research into three kinds of approaches: (a) “experiential” methods, which conceptualize the meaning of experiences into structures, narratives, categories, or themes and include phenomenology, narrative analysis, and the grounded theory method; (b) “discursive” methods, which are used to study pragmatics or function of language and include conversation analysis and discourse analysis; and (c) “experiential/discursive” methods, which include thematic analysis and the case-study method. Rennie argued that the hermeneutic circle, originally proposed as a method of analysis by Dilthey (1996), pertains to all discovery-oriented analyses of verbal text and such analyses characterize almost all contemporary qualitative research.

Case Studies The term case studies usually refers to studies that focus on a single individual, organization, event, program, or process or what Stake (2000, p. 436) called a “specific, unique bounded system.” Many academic departments are wary of supporting case studies as dissertations because departments are dubious of the likelihood of learning much of conceptual value from a single instance or example. On the other hand, case studies are frequently found in practice-oriented disciplines—such as education, social work, management science, urban planning, and public administration—in addition to some traditional social science disciplines (Yin, 2013). Indeed, there are many ways of thinking about case studies from both quantitative and qualitative perspectives. A quantitative approach in the classic experimental tradition could include what has been called a single-subject or N = 1 design. This empirical approach is associated with specific statistical procedures (see Gast & Ledford, 2009; S. B. Richards, Taylor, Ramasamy, & Richards, 2013). Single-subject quantitative studies can be used to assess changes in a phenomenon over time through the use of repeated measures or to assess the impact of a particular treatment by removing or reversing the intervention and then evaluating differences in the dependent variable. Single-subject research strategies are especially appropriate for developing or refining novel interventions and for closely examining the behavior of individual subjects.

Case studies, however, are more commonly associated with qualitative designs, in which there is an intensive effort to understand a single unit of study within a complex context. Research questions vary, but the goal is always to obtain a comprehensive understanding of the case. As Stake (2005) advised, “Place your best intellect into the thick of what is going on” (p. 449) and use your observational and reflective skills to excavate meanings.

How important is generalizing to a larger population? It depends. Stake (2000) described the intrinsic case study as one in which generalization is irrelevant because the attraction is understanding the unique (or even typical) person, group, or event. He described the instrumental case study as one that is intended to shed light on an issue or test a generalization rather than focus on the case per se. In our opinion, a purely descriptive or exploratory case study does not fulfill the expectations of a doctoral

dissertation unless it includes an explanatory element with theoretical implications. This means that the researcher needs to generalize to the world of theory as opposed to other possible cases. It also means that the research question is more apt to be of the “how” or “why” category than the descriptive “who,” “what,” and “where” questions that pertain, for example, to survey research and many other applied endeavors. However, we are aware that this is not a universal standard. The interested reader is referred to authors on case studies such as Stake (2000) and Yin (2013), who discuss these and related issues from somewhat different perspectives.

It should be said that any number of specific data collection methods might be included in a good case study. These would include interviews, behavioral observations, participant observation (as in ethnographic research), documentation, and the examination of archival records. Classic case studies include the sociological description of Middletown, a small Midwestern town (Lynd & Lynd, 1929); W. F. Whyte’s (1955) Street Corner Society; and Freud’s (1905–1909/1997) Dora: An Analysis of a Case of Hysteria. Thus, it is better not to think of your potential dissertation as using the case study method but rather to think of applying a method to a single case. Among case study dissertations at our own institution is a psychobiography of Richard Price, cofounder of the Esalen Institute. This study used the theoretical perspective of intersubjectivity theory and drew from archival documents, personal histories, and interviews with colleagues, friends, and family members to identify the recurring themes and patterns in Price’s subjective world so as to illuminate their influence on his contributions to Gestalt theory and practice and the evolution of Esalen (Erickson, 2003). A very different case study dissertation comes from Paula Holtz (2003), who conducted an ex post facto study of three brief psychodynamic psychotherapies that investigated the self- and interactive regulation and coordination of the timing of vocal behaviors of therapist and patient throughout the course of each therapy session. The study used a repeated single-case design, computerized scoring of the vocal behaviors, and time-series analyses. Among other findings, the analyses provided substantial evidence in support of the psychoanalytic dyadic systems view that each therapist or patient self-regulates the timing of his or her vocal behaviors with those of the partner. Finally, Cristina Balboa (2009), a graduate of Yale University, received a prestigious dissertation award (the Gabriel G. Rudney Memorial Award) for her qualitative comparative case study research on environmental nongovernmental organizations (NGOs) and their

governance operations. She studied and assessed the accountability of three private conservation networks in Papua New Guinea, Palau, and the Philippines while drawing upon contemporary theories of organizational structure and ethos.

Mixed Model: Quantitative and Qualitative Study An increasingly popular approach to designing a dissertation is to use a combination of quantitative and qualitative methodologies. This approach combines the rigor and precision of experimental, quasi-experimental, or correlational designs and quantitative data with the depth understanding of qualitative methods and data. Thus, the methods can inform one another or deal with different levels of analysis. There are many ways of mixing models. Teddlie and Tashakkori (2009) have enumerated several possible designs, including mixed methodology studies that combine aspects of both paradigms throughout the study. Theirs is a pragmatic approach in which questions of method are secondary to the adoption of an overriding paradigm or worldview to guide the investigation. Thus, it might be possible to mix research hypotheses of a confirmatory nature with general questions of an exploratory nature, structured interviews and scales that are quantitative with open-ended interviews and observations that are qualitative, and methods of analysis that draw on both traditions to expand the meaningfulness of the findings. An early example of an innovative mixed methodology was employed by Mary Gergen (1988) to study the way in which women think about menopause. Gergen held a research event by inviting several women to her home to complete questionnaires that addressed attitudes toward menopause, followed by a group discussion on the topic. The research report combined a quantitative analysis of the responses to the questionnaire with a qualitative analysis of themes generated by the discussion. An example from another field would be an analysis of the effect of timber dislocation on a logging community by quantitatively assessing the economic impact and qualitatively assessing the emotional impact on workers in the industry and their families.

The mixing of methods within the mixed model dissertation occurs in the data collection phase, the data analysis phase, and the data interpretation phase of the study. A simplified summary might include two main options: (a) whether the quantitative and qualitative elements of the study are sequential or concurrent and (b) whether one method is nested within the other or is used to confirm the findings obtained by the other. In a

sequential strategy, a researcher might begin with one approach and subsequently use the other approach to elaborate on or expand those findings. One variation is to add a qualitative component to a fundamentally quantitative study to help explain or extend the findings. Another option is to begin with a qualitative phase and add quantitative data collection at a later point. This design makes it possible to submit an emergent theory from a qualitative study to quantitative validation (Morgan, 1998). It may also be the method of choice when a researcher is designing an assessment instrument using largely rational or qualitative methods for constructing and choosing items and then validating the instrument statistically.

In a concurrent (or “parallel”) design, the researcher collects or analyzes both forms of data at the same time. In the most common variation, the quantitative and qualitative approaches are used to supplement one another in the same study, with each method seeking to confirm or validate the findings from the other and strengthen the outcomes of the study. The researcher hopes that advantages of one approach compensate for weaknesses of the other.

In an embedded design (Bazeley, 2009), there is one predominant method, and the other method is nested within it to enable the researcher to obtain a richer perspective on the phenomenon being studied. The researcher may use the embedded method to look at a different question than explored with the dominant method. A common application is assessing a larger group quantitatively and then interviewing a subsample of that group qualitatively to procure further information. Another application is the collection of quantitative data in a predominantly qualitative study to learn more about the participants. Both Bazeley and Teddlie and Tashakkori have enumerated the various kinds of mixed methods designs that have recently found their way into the research literature.

Mixed model studies present many logistical challenges, one of which is simply the burden of collecting data using two very different methodologies. Another is that students choosing this approach must become knowledgeable about and conversant with two different research paradigms. Nonetheless, we find that an increasing number of students are electing this approach to dissertation projects in spite of the increased task demands.

Perhaps the most common application of mixed methodology is to assess a

large number of participants using standardized scales and measures in a field study or an experimental study and then conduct open-ended interviews with a subset of the original sample to derive a richer understanding of the phenomenon in question. A good example is a study by one of our doctoral students who sought to understand what makes “extreme” athletes (e.g., those who scale vertical cliffs without supporting ropes) engage in what laypeople view as self-destructive behavior (Slanger, 1991). The resulting dissertation combined validated measures of sensation seeking and perceived competence, for objective data, with open-ended interviews conducted with a random subsample of the total group, to obtain a qualitative perspective. Slanger discovered that the methods complemented one another: Data from the quantitative scales revealed how the key predictive variables discriminated among extreme- risk, high-risk, and recreational athletes, and the qualitative interviews introduced the concepts of spirituality and flow (Csikszentmihalyi, 1991).

Another graduate (Christensen, 2005) adopted a mixed method design to study conflict at the governance level within Friends schools, which educate children from a Quaker perspective. Christensen gathered intensive data from interviews with trustees and a focus group with consultants who worked with Friends school boards and supplemented those stories with quantitative data from an electronic survey sent to a larger number of school representatives. The combined data enabled her to identify predictors of growth in organizational dynamics and then design a module-based program for board preparation and education.

Similarly, Hardy (2011) studied the experience of boundary crossings and violations in supervisory relationships among graduate students in counseling and clinical psychology. Data collection was performed using a web-based questionnaire that included case vignettes. Three hypotheses relating to the incidence of boundary violations and how they are defined and perceived were tested statistically. Participants also provided narrative accounts of their own experiences with supervisory boundary crossings and violations and the personal and professional impact of these experiences. The qualitative component of the study was a hermeneutic analysis of these narratives.

Finally, David Nobles (2002) took a very different mixed model approach to his dissertation on speech acts of President George W. Bush related to drug control policy implementation. Nobles analyzed 33 rhetorical

artifacts consisting of speeches, exchanges with the media, and other public remarks from the perspective of three research models: dramatism and metaphorical analysis, both approaches to rhetorical criticism, and communication theory, in the form of coordinated management of meaning (CMM) and social constructionism. The findings describe the impact of the war on drugs metaphor on drug use and drug control policies.

Students who decide to take a mixed model path to their dissertations have a number of decisions to make, including which method, if any, receives priority; how to decide on a data collection sequence; how to explain and integrate findings that may not be congruent; and whether a larger, theoretical perspective should frame the entire research design. Creswell and Plano Clark (2011) is a helpful reference for those seeking criteria for making these strategic choices.

A major reluctance to adopt the mixed model approach comes from scholars with strong epistemological commitments to either quantitative or qualitative research. They often view the underlying assumptions of the approaches as fundamentally incompatible. At the risk of repetition and oversimplification, quantitative studies generally rest on an objectivist epistemological tradition, which seeks to validate knowledge by matching the knowledge claims of the researcher with phenomena in the real world (the correspondence theory of truth). In this tradition, theories are proposed as universal hypotheses to be tested empirically. Qualitative studies, on the other hand, tend to derive from the constructivist tradition associated with the postmodern movement. Here knowledge is not discovered but invented. Moreover, it is situated within a specific context heavily determined by local practices and validated through internal consistency and social consensus. In practice, this means that the researcher maintains an open curiosity about a phenomenon and the theory emerges from the data; there is no one true reality against which one can validate theories deductively.

Morgan (2007) has discussed the navigation of this dilemma as a paradigm shift within the research community, arguing that a pragmatic approach to research design outweighs the merits of adhering to a rigid epistemological position. She has recommended that a pragmatic approach would substitute

abductive reasoning for connecting theory to data for the purely

inductive reasoning associated with qualitative research or the purely deductive reasoning associated with quantitative research; intersubjectivity with regard to the research process for qualitative subjectivity or quantitative objectivity; and transferability when making inferences from data for solely emphasizing the context (qualitative) or generalizing from samples to populations (quantitative).

In this context, abduction refers to going back and forth between induction and deduction, such that observations lead to theories, which then lead to actions in the real world. Intersubjectivity refers to emphasizing shared meanings within specific groups rather than either seeking “truth” or relying entirely on the subjectivity of knowledge. Transferability, in terms of making inferences from data, refers to emphasizing how things that are learned in one context can be applied to another context; this meaning is in contrast to either generalizing from a sample to a population or being restricted by the contextual limitations of knowledge gained in the study. We recognize that not everyone will agree with Morgan’s assumptions, but we appreciate her efforts to find a pragmatic way to improve communication and understanding between what are frequently regarded as irreconcilably polarized research paradigms.

Our own position is that both quantitative and qualitative studies can be approached from a myriad of philosophical perspectives. We encourage students to think clearly about a research topic and then apply the methods that make the most sense for answering their questions of interest and that are consistent with their values. We suggest that you begin your research by asking an essential question and then asking what you must do to convince yourself and others of the validity of the evidence supporting it. Along the journey, be wary of rigid methodological rules and draw on any method with a clear understanding of its advantages, its limitations, and whether it compromises assumptions about the phenomena you are researching.

Theoretical Dissertations Another possible approach to writing a dissertation is to write a theoretical dissertation and bypass the need for data collection entirely. This is by no means an easy alternative. Making an original theoretical contribution is a profound intellectual challenge. One way to consider the difference

between the knowledge of the literature required for a standard quantitative or qualitative study and the knowledge required for a theoretical study is to think about the difference between being a native of a country and a tourist in that country. As a tourist in a foreign environment, you might learn as much as possible about the country by studying maps, reviewing the customs, and learning the language, but chances are you will never master the country as well as the native. It’s the same with research. To make a genuinely original theoretical contribution, you need to know an area of inquiry inside out and be intimately familiar with the issues and controversies in the field. If you are beginning to review an area of interest to formulate a study, you are probably better off adopting an empirical design. Of course, most doctoral dissertations need to be derived from theory and have theoretical implications, and the data you gather and analyze may create the opening for a brand-new way of thinking in your field. That, however, is quite different from starting with the expectation of creating, let’s say, a new theory of consciousness or, a bit more modestly, a revised theory of short-term memory.

If you choose to pursue a theoretical dissertation, you will be expected to argue from the literature that there is a different way of understanding a phenomenon than has heretofore been acknowledged. Some of the more viable theoretical dissertations in the social sciences are those that bring together or integrate two previously distinct areas. For instance, one of our graduate students believed that there was a significant breach between the theory of psychotherapy and the practice of psychotherapy, and this view led to an ambitious, high-quality theoretical dissertation on the relevance of personal theory in psychotherapy (Glover, 1994). Another student completed a very scholarly, book-length theoretical dissertation titled Organic Constructionism and Living Process Theory: A Unified Constructionist Epistemology and Theory of Knowledge (Krebs, 2005). On a less abstract level, Rainaldi (2004) developed a new theory of incorporative female sexuality informed by psychoanalytic drive theory and recent advances in the biological sciences.

Meta-Analysis Meta-analysis is a form of secondary analysis of preexisting data that aims to summarize and compare results from different studies on the same topic. Meta-analyses have become increasingly common in the social science literature because they pool the individual studies of an entire research

community, thus providing the reader with a richer understanding of the status of a phenomenon than any single study can offer.

The term meta-analysis has been attributed to Glass (1976), who used it to mean an “analysis of analyses.” A more complete description of the various meta-analytic methods is available in Newton and Rudestam (2013). Meta-analyses differ in terms of the units of analysis they use (e.g., a complete study, a finding within a study) and the statistical techniques they use to integrate the results from separate studies to draw conclusions about the entire body of research.

The first step in conducting a meta-analysis is to screen and select existent studies for their methodological rigor. Then statistical techniques are used to convert the findings of all the studies to a common metric. Finally, the summary analysis yields information about the strength of relationships among variables (the effect size) across studies, using the newly expanded sample.

All dissertations, of course, involve a critical review of the literature on the topic in question. In a meta-analysis, it is this review of the literature, including a finely tuned statistical analysis, that constitutes the study. In our opinion, there is no reason why a carefully conducted meta-analysis could not serve as a suitable dissertation. For further information about conducting a scholarly meta-analysis, we recommend an introductory text by Borenstein, Hedges, Higgins, and Rothstein (2009).

Action Research Action research provides another possible approach to completing a doctoral dissertation, although it may be too prodigious a challenge for most graduate students. Action research has been defined as “a form of research that generates knowledge claims for the express purpose of taking action to promote social change and social analysis” (Greenwood & Levin, 2006, p. 6). Because action research is generally stimulated by a wish to address a particular problematic situation within an identifiable organization or community, it is distinct from theoretical research that is carried out as a purely academic exercise. Another distinguishing feature is that action research is never done “to” someone but is done by or in collaboration with insiders from the organization or community. This systematically undertaken, reflective process includes creating theory

within a practice context and testing the theory using specific experimental interventions (Herr & Anderson, 2005; Stringer, 2013).

Most action researchers acknowledge the seminal contributions of Kurt Lewin (1948) and his commitment to social change. Action research can be either quantitative or qualitative in nature, drawing on such diverse techniques as surveys, interviews, focus groups, ethnographies, life histories, and statistics. In the early days of action research, the researcher tried to initiate change in a particular direction; more recently, the goals and targets of change are determined by the group members through participatory problem solving. Members of an organization or community that constitutes the focus of the research become coresearchers in the process. Thus, the researcher is a facilitator who needs to possess good group process skills to effectively mobilize a group of participants to study their own behavior, including their own defensive reactions to change.

A good action research project proceeds according to a cycle of steps, introduced by Lewin (1948), known as the plan-act-observe-reflect cycle.

1. The planning stage involves the identification of a problem and the formulation of hypotheses and procedures for achieving one or more goals.

2. The action stage consists of implementing the intervention(s). 3. The observe stage consists of recording the actions and their impact

on achieving the goal(s). 4. Finally, the reflection stage allows for reviewing the data and the

action plan and developing new inferences. These lead to a new cycle of research, and research is a continuous learning process.

Herr and Anderson (2005) advised that to serve as a dissertation, an action research study should contribute generalizable, transferable knowledge as well as knowledge that is useful to those in the setting of the study, and we endorse this point. Action research may, for instance, generate new theory that is applicable to similar problems in other contexts, as well as new tools or products that are recommended for broader use. Herr and Anderson also noted that students who envision conducting an action research study for their dissertations should be conscious of certain potential complications. One is that action research studies can be “messy,” in the sense that procedures and outcomes are difficult to predict. Thus, committee members may need to stay flexible regarding potential outcomes and understand that the methods and procedure may need to be

revised as one goes along. Second, students need to realize that they may be walking a tightrope as they serve the multiple roles of student, researcher, and participant in the research and maybe even employee in the organization. Students must be prepared to make choices with a full awareness of the possible consequences and their ethical implications. Finally, it is important to identify the contributions of the author of the dissertation in spite of the fact that several other individuals may have served as coresearchers.

Within our institution, most action research dissertations have taken place in the fields of education and organization development, although the fields of social work, nursing, and criminology also attract this approach. The action research cycle was used by Judy Witt in her dissertation to explore a community college’s use of collaborative organizational learning in its planning and decision-making processes (Witt, 1997). The student worked as a coresearcher with members of the college administration, faculty, and staff. Each member of the team brought specific skills to the project. The dissertation student, of course, provided her expertise in action research. The team analyzed archival data, as well as data from meetings, journals, interviews, and participant observation field notes, to evaluate the effectiveness of the institution’s learning processes.

Notes 1. Numerous other statistical models control for extraneous variables; only two of the most common are presented here.

2. See Holstein and Gubrium (2008) for an overview of many approaches to constructionism.

Part II Working With Content The Dissertation Chapters

Literature Review and Statement of the Problem The Method Chapter: Describing Your Research Plan Presenting the Results of Quantitative Research Presenting the Results of Qualitative Research Discussion

4 Literature Review and Statement of the Problem

The previous chapters provided an orientation to research in the social sciences and offered suggestions on how to develop an appropriate topic. In this chapter, the research question begins to take shape using the vehicle of the Review of the Literature.

The Introduction The Review of the Literature is generally preceded by a brief introductory chapter. The Introduction consists of an overview of the research problem and some indication of why the problem is worth exploring or what contribution the proposed study is apt to make to theory or practice, or both. The Introduction is usually a few pages. Although it may begin by offering a broad context for the study, it quickly comes to the point with a narrowly focused definition of the problem. The form of the Introduction is the same for both the research proposal and the dissertation, although the understanding of the research problem will likely undergo some changes after the study is completed. Ironically, it is usually impossible to write a final Introduction chapter before completing the Review of the Literature and Method chapters because those chapters will inform the problem and its operationalization.

The wording of the research problem should be sufficiently explicit to orient the most inattentive reader. One might, for example, begin the chapter with a sentence such as “In this study, I attempted to evaluate the impact of environmental protection legislation on atmospheric pollutants in the chemical industry.” The chapter would proceed to stipulate the assumptions and hypotheses of the study, identify the key variables, and explain the procedures used to explore the questions. It should include a synopsis of the arguments that explain the rationale for the research question and the study. It is perfectly acceptable to cite one or more studies that are directly relevant to the proposed investigation and may have inspired it or lent it empirical or theoretical justification. However, this is not the place to conduct a literature review. Avoid technical details and

keep the Introduction short.

Review of the Literature Often the lengthiest section of the research proposal, the Review of the Literature is placed just after the introductory overview of the study. This chapter of the dissertation provides a context for the proposed study and demonstrates why it is important and timely. Thus, this chapter needs to clarify the relationship between the proposed study and previous work conducted on the topic. The reader will need to be convinced not only that the proposed study is distinctive and different from previous research but also that it is worthwhile. This is also the place where the student’s critical abilities as a scholar become evident. Many students erroneously believe that the purpose of the literature review is to convince the reader that the writer is knowledgeable about the work of others. Based on this misunderstanding, the literature review may read like a laundry list of previous studies, with sentences or paragraphs beginning with the words “Smith found . . . ,” “Jones concluded . . . ,” “Anderson stated . . . ,” and so on. This not only is poor writing but also misses the whole point of an effective Review of the Literature.

A colleague of ours, Jeremy Shapiro, noted that much of the labor that goes into writing is often wasted effort because it is not based on a clear understanding of the purpose of an essay or thesis (Shapiro & Nicholsen, 1986). As a general rule, if you have difficulties in your basic writing skills—that is, in constructing grammatical sentences, using appropriate transitions, and staying focused and concise—a research dissertation will glaringly reveal these weaknesses, and the logic and persuasiveness of your arguments will be diminished. Grammar does not receive nearly as much attention in our educational system as it did in the past. Perhaps the best justification for learning good grammar is some sage advice forwarded by Lynne Truss (2006): “Punctuation is a courtesy designed to help readers to understand a story without stumbling” (p. 7). Thus, punctuation marks serve as traffic signals to let the reader know when to pause, pay attention, take a detour, or stop. Consider the following popular example of two sentences with the same words and different punctuation. Notice the different meanings.

A woman, without her man, is nothing.

A woman: without her, man is nothing. (Truss, 2006, p. 9)

Effective academic writing is an acquired skill. One suggestion is to obtain remedial help in strengthening basic writing skills. Furthermore, be aware that the style of writing that is appropriate to research papers is different from the style of writing associated with literary prose. Scientific writing tends to be more direct and to the point and less flowery and evocative. Writing style and process are taken up in depth in Chapter 10.

A good way to formulate a question that is appropriate to a research study is to determine what concerns you, bothers you, or tweaks your curiosity. As you consider one or more possible questions and draw on the observations and ideas of others who are interested in the same and related questions, you are in fact formulating the argument. The forum for the argument is the literature review, which is played out in the form of a dialogue between you and the reader. To dialogue effectively, the writer must anticipate the kinds of questions and apprehensions that the reader might have when critically examining your argument. It is common for critical evaluations of academic papers to be peppered with comments such as “What is your point here?” “What makes you think so?” “What is your evidence?” and “So what?” The more you can anticipate a reader’s questions, the easier it will be to formulate your arguments in a way that produces mutual understanding. Dissertations go through many drafts, and the revision process consists of asking and responding to questions from the point of view of a circumspect and knowledgeable reader.

The literature review is not a compilation of facts and feelings but a coherent argument that leads to the description of a proposed study. There should be no mystery about the direction in which you are going. (“Where are you going with this?” is a good question to ask yourself repeatedly when writing the Review of the Literature.) You always need to state explicitly, at the outset, the goal of the paper and the structure of the evolving argument. By the end of the literature review, the reader should be able to conclude, “Yes, of course, this is the exact study that needs to be done at this time to move knowledge in this field a little further along.” The review attempts to convince the reader of the legitimacy of your assertions by providing sufficient logical and empirical support along the way. You will continually need to determine which assertions the reader can accept as common understanding and which assertions require data as support. For instance, if you were to assert that survivors of suicide

attempts need professional help, a peer reader probably would want to know the basis of your assertion and request some evidence about the needs of those who attempt suicide and why professionals (as opposed to nonprofessionals) are necessary. Avoid statements based on “common knowledge” that can easily be verified as false. For example, someone might make the statement that the divorce rate has skyrocketed in recent years. Not only is this statement false, but also the time frame represented by “recent years” is unclear. On the other hand, the claim that Freud was the father of psychoanalysis is likely to be well established as a fact in the professional psychological community and thus not require further backing.

It is perfectly permissible to draw on the thoughtful arguments of others and incorporate them into your research project. This is very much in keeping with researchers’ understanding of the incremental, cumulative process that characterizes the development of most science (Kuhn, 1996).1 On the other hand, a skillful researcher draws on original source material rather than relying on review articles and secondary sources. We advise great caution when reporting summaries of statistical findings from secondary sources because these can often be both incomplete and misleading. For any study whose results are critical to your central arguments, we recommend careful inspection of the design, analysis, and results from the primary source material.

Becker (1986) described this process with the metaphor of a jigsaw puzzle: Imagine that you design some of the pieces and borrow others in their prefabricated form from the contributions of other scholars. In addition, it is worth noting that becoming overly preoccupied with the literature can deform your argument so that you lose your privileged place at the center of the study. Do not let your anxiety about one missing reference delay your forward momentum; you can always insert missing or new material into the review later. In any case, do not neglect to give proper credit to the source of ideas by citing complete references in your writing.

Common Problems A principal failing of novice researchers at every stage of a project, which is especially evident in the Review of the Literature, is that they give away their own power and authority. As a researcher, you need to accept that you are in charge of this study and that, in the case of dissertations, you

will likely be the world’s leading expert on the narrow topic you address. One way of giving away authority is to defer to the authority of others in the review, assuming, for instance, that because Émile Durkheim or John Dewey said something, it is necessarily valid. You need to adopt a critical perspective in reading and relating the work of others. The main reason why sentences beginning with “Jones found . . .” are best kept to a minimum is that they shift the focus of the review from your own argument to the work of others. A preferable strategy is to develop a theme and then cite the work of relevant authors to buttress the argument you are making or to provide noteworthy examples of your point or counterexamples that need to be considered. Consider the following fictional examples:

Illuminatus (2010) conducted a study on the effects of seasonal light on major depression by comparing the rates of depressive illness among residents of Seattle and San Diego during different times of the year. He hypothesized that there would be more depression in the northern city than the southern city and more depression during winter than during summer. His findings confirmed his hypotheses.

This presentation turns the focus of the review toward the opinions of another researcher.

There is reason to believe that differences in seasonal light can affect the extent of depressive illness. For instance, residents of Seattle, a northern city, are reported to have higher rates of depression than residents of San Diego, a southern city, and these differences are accentuated during the winter (Illuminatus, 2010), suggesting that relative darkness may exacerbate mood disorders.

This presentation keeps the focus on the phenomenon and uses the finding of the cited research as empirical support.

Another way to limit your own authority is by using quotations in excess. The overuse of quotations tends to deflect the argument away from the

control of the author. Restrict the use of quotations to those with particular impact or those that are stated in a unique way that is difficult to capture through paraphrasing. Besides, using your own words to present difficult concepts will help convince you (and others) that you really understand the material.

Once you have read the literature in an area, it may be tempting to report everything you now know. Do not succumb to this temptation! A good literature review needs to be selective, and it is taken for granted that the majority of source material you have read will not make it directly into the literature review. That does not mean that it wasn’t necessary to read all those books and articles; they provide the expertise required to make your contribution. But remember, in the dissertation itself your task is to build an argument, not a library. One of our colleagues likens the process to a courtroom trial, where all admissible testimony by the witnesses must be relevant to the case and question at hand. Consistently ask yourself, “Why am I including this study or reference?” Similarly, each sentence in the dissertation needs to be there for a purpose—sometimes to provide relevant content and sometimes to facilitate communication to the reader, but never as filler.

Although the primary task is to build an argument and you are expected to present your own point of view, it is not fair to exclude references that contradict or question your case. You must be objective enough to present both sides of an argument and acknowledge where the weight of the evidence falls. Some of the most important and most cited dissertations are those that noted conflicting empirical findings, addressed the findings from differing theoretical perspectives, and reconciled the differences with a new, more comprehensive design.

Throughout the review, leave enough signposts along the way to help orient the reader. One way to do this is to inform the reader of what you have done and what conclusions you have drawn on the basis of the available evidence. You also need to convince the reader that your knowledge of the existing literature is sufficiently extensive and intensive to justify your proposed study. That a study on your topic or question does not yet exist is never sufficient justification: Many things are simply not worth studying.

Part of the purpose of the literature review is to enable you (and the reader) to theorize and generate hypotheses. As such, almost all doctoral

dissertations need to be theoretically grounded. That does not mean simply reviewing a number of theories that might be applicable to your study. It means taking a position regarding a theoretical orientation that supports the proposed study. Finally, students sometimes believe that academic writing needs to be pedantic, dry, or just plain boring, probably because that’s what they learned from their professors and from reading much of the research literature. We take issue with this recommendation and have more to say about it in Chapter 10. Certainly the first objective is to write clearly and accurately, but if you can do so with style and without jargon, so much the better. After all, the literature review tells a story, and that story can be told in an interesting manner. A good example of a well- written literature review that builds the reader’s interest can be found in a journal article by D. T. Gilbert, Pinel, Wilson, Blumberg, and Wheatley (1998). Their review exemplifies the goal of directing a literature review to a reasonably intelligent audience so that readers will comprehend the gist of what you are saying even without having specialized knowledge in your area of focus.

Critiquing a Research Article The relevant studies need to be critiqued rather than reported. The critique serves to inform the reader about the status of reliable knowledge in the field and to identify errors to avoid in future research. As you read the available research in an area, you need to maintain a critical perspective, evaluating the study on its own merits and in comparison with other studies on the same or a similar problem. A critique does not imply that you must discover and identify a major flaw or weakness in every study you read. Sometimes students offer critiques that read like a list of “weaknesses” of a particular method cited from a research methods text. These are seldom of value.

You are evaluating the content for its application to your research. That means paying particular attention to the following three elements of all empirical studies:

1. How was the problem defined? Is this definition similar to or different from the way in which you are conceptualizing and defining the problem and the associated concepts and variables?

2. What measures were used to operationalize the variables and assess the differences between groups or effects of interventions? Are these

measures similar to or different from those you intend to employ? 3. What population was studied, and how was the sample chosen? Is the

population similar to or different from the one you intend to address? Was the sample chosen randomly, out of convenience, or in a biased manner?

Answers to these questions will help you evaluate the relevance and limits of generalizability of the study you are reviewing in relation to your proposed study. In addition to these major points, it is always a good idea to ask what the author of the study can properly conclude based on the design. Does the study, for instance, enable you to conclude that there is a cause-and-effect relationship between the study variables or merely that they correlate with one another?

The outline in Box 4.1 consists of a comprehensive set of recommendations for critiquing a research article. Not all of these items will apply to any given citation within the literature review. The amount of attention a study receives will depend on its direct relevance to the proposed research question and should not detract from the flow of the argument. Nevertheless, this list can serve as a reminder of how to read and evaluate critically a research article’s contribution to a proposed study.

Box 4.1 Recommendations for Critiquing a Research Article 1. Conceptualization

a. What is the major problem or issue being investigated? b. How clearly are the major concepts defined/explained?

2. Theoretical Framework and Hypotheses a. Is there a clearly stated research question? b. Are there hypotheses? Are they clearly stated? c. Are the relationships between the main variables explicit and reasonable? d. Are the hypotheses stated in a way that makes them testable?

3. Research Design a. Does the research design adequately control for extraneous variables? b. Could the design be improved? How? c. Are the variables clearly and reasonably operationalized? d. Is the choice of categories or cutting points defensible? e. Are the reliability and validity of the measures discussed? f. Is the choice of measures appropriate? g. Is the population appropriate for the research question being studied? h. Is the sample specified and appropriate? i. Can the results reasonably be generalized on the basis of this sample, and to what population?

4. Results and Discussion a. Are the data appropriate for the study? b. Are the statistical techniques appropriate and adequately described? c. Are the control variables adequately handled in the data analysis? d. Are there other control variables that were not considered but should have been? e. Are the conclusions of the study consistent with the results of the statistical analyses? f. Are alternative conclusions that are consistent with the data discussed and accounted for? g. Are the theoretical and practical implications of the results adequately discussed? h. Are the limitations of the study noted?

5. Summary a. What is your overall assessment of the adequacy of the study for exploring the research

problem? b. What is your overall assessment of the contribution of the study to this area of research?

Long Shots and Close-Ups Our colleague Joseph Handlon drew an analogy between doing a literature review and making a movie. In filmmaking there are, depending on the distance between the camera and the subject matter, long shots, medium shots, and close-ups. As a metaphor, a long shot suggests that the material is background for a particular topic. Background material needs to be acknowledged but not treated with the same detail as foreground material; it is not psychologically figural. A study on the stressful impact of relocation, for instance, might begin with the following observation:

There have been three basic ways of approaching the topic of stress empirically. One is by regarding stress as an independent variable and focusing on the nature and strength of the stressor, exemplified by the empirical contributions of Holmes and Rahe (1967). A second approach is to view stress as a dependent variable, focusing on the physiological and psychological impact of stressful events, illustrated by the seminal work of Hans Selye (1956). An alternative approach is to view stress as a transaction between a stimulus and a response, which is moderated by a set of cognitive variables. This approach, elaborated in the work of Lazarus and his colleagues (Lazarus & Folkman, 1984), forms the conceptual foundation for this study.

The references cited in this example above are quite dated. However, they are regarded as classic studies and, therefore, deserve citation in this context.

Here is another example of a long-shot approach that briefly cites representative studies that do not need to be examined in greater detail because they are not directly relevant to the proposed research question. They are taken from a dissertation proposal on the MMPI and sex offenders by Peter Ellsworth (2013):

Several studies utilizing MMPI/MMPI-2 scores have been conducted with the sexual offender population (Craig, 2005). Studies have compared MMPI/MMPI-2 scores for sexual offenders and non-offenders (Davis & Archer, 2008); recidivist and non-recidivist sex offenders (Erickson, 1987; McCreary, 1975); high-risk and low-risk sex offenders as measured by static factors (Coxe & Holmes, 2009); clergy sex offenders and other offenders (Langevin et al., 2000); clergy child molesters and a national norm (Plante & Aldridge, 2005); and clergy child molesters in treatment centers and a variety of other groups (Terry et al., 2011).

The medium shot is somewhere between the long and the short focus and requires a bit more descriptive material than does the long shot. As an example, let us assume that a researcher wishes to explore the effect of social protest and threats of violence on the well-being of workers in abortion clinics. It would be appropriate to obtain a good overview of the impact of potentially violent social protest in other contexts as well as a good understanding of the emotional demands of working in a clinic serving women with unwanted pregnancies. Studies that bear on these relevant issues may not need to be presented in critical detail, but they certainly need to be summarized sufficiently to give the reader a clear indication of the status of the research as it pertains to the orientation of the proposed study.

Finally, the close-up requires a careful examination of the research and is reserved for those studies that have the most direct relevance to the proposed research question. In some cases, one or two studies are being modified or amended in some critical way to form the basis for the current study. More frequently, a collection of work on a relatively narrow topic is clearly central to the proposal. In either case, these studies are not merely referenced but critically examined so that the reader obtains a clear sense of what is already known about the phenomenon, how reliable and valid the conclusions based on that work are apt to be, and how the proposed study will deal with previous limitations and move the field ahead. The researcher who is interested in exploring the impact of infertility treatments on communication between husbands and wives might present the following close-up statement after having carefully described the samples, measures, and procedures of the two most relevant (fictitious)

studies in that literature.

Of the two studies that bear directly on the proposed question, Sterile (2010) found that couples reported improved communication after experiencing prolonged infertility treatment, whereas Ripe and Fertile (2012) concluded that behavioral exchanges between infertile couples more frequently escalated into arguments the longer that medical interventions continued. Of particular concern in Sterile’s study is the fact that because men and women were interviewed together, the couples may not have been totally honest and responses by one member of a couple may have been prejudiced by those of the other member. Beyond this threat to validity, the conflicting findings of the two studies suggest the need for a more definitive investigation of the impact of infertility treatment on communication patterns within couples.

A good strategy for reviewing the literature can be found by referring to a Venn diagram (see Figure 4.1) of three intersecting circles, which is derived from the previously discussed exercise on formulating research questions. The long shots, or those studies seen through wide-angle lenses, are represented by the portions of the three primary variables that are independent of the other two variables. The medium shot is represented by the intersections of any two variables. The close-up, or studies viewed through a narrow-angle lens, is represented by the joint intersection of all three variables. As a general rule, any studies in the existing literature that incorporate all of the major variables or constructs in the proposed study require very careful scrutiny because they are particularly relevant. Studies that relate some of the variables (e.g., two) also deserve a short description. Studies that deal with only one of the selected variables, perhaps in conjunction with other, less relevant variables, are merely background. They are generally too numerous to examine in detail and include a great deal of content that does not pertain to the current study.

Figure 4.1 Venn Diagram Guide to the Literature Review

As one more example, one certainly would not need to review all studies dealing with sexual dysfunction to focus on male impotence in midlife. Nor would one need to consider all previous work on men or midlife. Yet gender issues and midlife development issues may provide important background material and a theoretical foundation for the proposed study. Moreover, the researcher would not have to introduce every study on impotence but would probably need to be familiar with a broad range of previous work in the area.

Review of the Literature in Qualitative Dissertations Each research tradition has its own idiosyncratic approach to writing the dissertation. What you have just read constitutes the prevailing model in the social sciences. However, research studies that are qualitative in nature may have a different approach to the literature review. Because many of these studies are inductive or theory building, as opposed to theory testing, much of the formal literature review may be found in a discussion chapter toward the end of the dissertation, after a theory has emerged and the author is seeking to position it within the existing literature.

Ruthellen Josselson (Josselson & Lieblich, 2003), who comes out of the narrative inquiry tradition, sees the need for a review that orients the reader to the existing literature but is concerned about a review that restricts rather than opens the inquiry, fearing that an overly

comprehensive or overly focused review preempts the researcher from greeting his or her data with the appropriate level of openness, curiosity, and wonder. The review chapter she favors is more like a research proposal, which demonstrates sufficient sophistication with respect to the literature, including existing theory and relevant empirical studies, but focuses on orienting the reader to the boundaries of the proposed study and launching it. More informally, this background material would say, in effect, “Here’s a phenomenon that interests me, here is one or more theories that have been proposed to try to understand this phenomenon, here are some empirical ways in which previous researchers have attempted to understand the phenomenon, this is what they have discovered so far, and this is what I, very tentatively and with great curiosity, think is important here.” The researcher is then encouraged to keep reading the literature as the study progresses and tie it in to his or her findings in a very thorough and evolved discussion chapter.

In another context, David Rennie (1998), a well-known grounded theory researcher who served as external examiner for the dissertation of one of our doctoral students, admonished the student to reduce significantly the 100-page literature review in her proposal. Rennie viewed it as overly inclusive and inappropriate for a discovery-oriented qualitative study and encouraged her to reintroduce those topics in the literature review that were subsequently referred to by her interviewees once the data were collected and examined. Along this line, William Glaser (1992), a classic grounded theorist, took the position that

there is a need not to review any of the literature in the substantive area under study. This dictum is brought about by the concern to not contaminate, be constrained by, inhibit, stifle or otherwise impede the researcher’s effort to generate categories, their properties, and theoretical codes from the data that truly fit. (p. 31)

As we have seen, however, even within the grounded theory tradition, there are those practitioners who take a less polarizing position with regard to the advisability of obtaining some exposure, if not mastery, of the relevant literature prior to conducting the study.

As one might imagine, the same inductive, discovery-oriented mind-set

finds its way into the Statement of the Problem and Method chapters of the dissertation as well. These distinctions become apparent as we tackle these respective sections of the dissertation in the following chapters of this book. In any case, all students need to give priority to the structural and writing conventions promulgated by their respective disciplines and academic departments.

Statement of the Problem At the conclusion of the literature review, the reader should have obtained a fairly clear idea of the study. By this time, you have carefully crafted your argument and moved the reader along as you built your case. You have convinced your reader of your mastery of the subject matter by having reviewed and critiqued the existent literature that pertains to your study and gives it a suitable context. The next immediate challenge is to form a transition between the literature review and the next section of the dissertation, the Statement of the Problem. One way to build this transition, so that the literature review chapter appears connected to the proposed study, is to write a summary of your review. This summary would highlight your main conclusions, reference the most relevant literature (which you have previously reviewed), and leave the reader anticipating the next steps.

The Statement of the Problem is sometimes written as a separate chapter and sometimes located at the very end of the Review of the Literature. Although you probably have offered a general statement of the problem early in the introductory chapter of the dissertation, this is the place for a more specific statement. The specificity of the problem statement is very important. A research problem consists of much more (and less) than a misunderstood collection of unidentified relationships. The statement is usually framed in the form of one or more research questions and research hypotheses. Although we recommend the inclusion of formal hypotheses as a general standard, whether or not to include them may depend on the type of study, what is known about the question, and the conventions of your discipline and department. Similarly, the statement of the problem may contain conceptual definitions of major concepts. This is particularly true when competing definitions of the concepts exist within the field of inquiry (e.g., it might be important to point out that the study will focus on trait anxiety as opposed to state anxiety or other conceptualizations of the

construct of anxiety).

It is critical that a research question have an explanatory basis. This means that the Statement of the Problem contains a brief summary of the conceptual underpinnings for the proposed research. Dust bowl empiricism is the somewhat derogatory term used to refer to a shotgun approach to research, in which the investigator levels his or her sights to see what is out there without developing a convincing chain of presuppositions and arguments that leads to a prediction. There is no research problem in “wondering” how the variables of gender, voice quality, and persuasion intercorrelate. This will not serve as a suitable problem statement. Hypotheses, on the other hand, have the virtue of being explanatory expressions of research questions because they imply a commitment to a particular understanding of how variables relate.

An example of a research question without a specific hypothesis is “What is the role of male significant others on the criminal activities of female criminals?” This research question implies a study that obtains information from or about women who have been convicted of crimes regarding the influence of boyfriends and male acquaintances in their criminal activities. A study that poses this question without predictive hypotheses (perhaps because of a lack of available information about this topic) might be termed exploratory. Too often, we find, students pose their dissertations as exploratory to avoid the challenging tasks of thinking deeply about underlying concepts and rigorously linking their ideas with previously published work in related areas.

In most instances, it is possible to project hypotheses. Even in those instances where there is a relative lack of research in an area, it is likely that studies and theories exist on related topics that can inform the proposed study. In the previous example, the investigator may have developed some reasonable hunches about the research question from his or her knowledge of women’s developmental theory and the role of the peer group in criminal behavior. These hunches would be reflected in one or more hypotheses.

An example of a research hypothesis is “There is a negative relationship between positive body image and motivation for augmentation mammoplasty.” A second example is “Couples in stable, unhappy marriages use more conflict avoidance methods than couples in stable, happy marriages.” The first hypothesis suggests a study in which the

variables of body image and motivation for breast augmentation will be statistically correlated, whereas the second hypothesis suggests a study using two groups of couples who will be compared on how they manage conflict. In either case, the variables specified in the hypotheses will need to be operationalized, or clarified with regard to how they are to be measured. Such specification usually takes place in the Method chapter. In the first example, the researcher might predict a negative relationship between scores on a body image scale and the motivation to seek augmentation mammoplasty. In the second example, the terms stable/unstable marriage and conflict avoidance methods would need to be conceptually defined and operationalized, and the two groups of couples will need to be identified.

It usually takes several rewritings to come up with research questions and hypotheses that are optimally clear, concise, and meaningful. Note that hypotheses typically are written as positive assertions in the present tense. They are not written as null hypotheses. The reader is probably aware that inferential statistics work on the assumption of rejecting null hypotheses, that is, hypotheses that assume there are no significant differences between or among groups or no significant relationships among variables. Research hypotheses, on the other hand, should be stated not as null hypotheses but as hypotheses that follow from the argument that has been established in the preceding chapter. As research hypotheses, null hypotheses are confusing because they reflect the opposite of the argument you have been proposing. If the logic behind the stated hypotheses is not totally evident, it is always a good idea to follow or precede each hypothesis with a short rationale that reminds the reader how it emerged from theoretical propositions established in the Review of the Literature. A common failing is to include hypotheses that seem to emerge full-blown without building an adequate foundation and referents for them in the literature review.

Recommended criteria for a good hypothesis are that it be free of ambiguity, express the relationship(s) between two or more variables, and imply an empirical test (Locke, Spirduso, & Silverman, 2013). A common pitfall is to have more than one hypothesis embedded in a single, complex statement (e.g., “Women who earn more than their husbands have more self-confidence, have more friends, and receive more help with household tasks than women who earn less than their husbands”).

Some dissertations contain both research questions with hypotheses and

research questions that stand alone. The hypotheses might cover those relationships that directly challenge previous work or test a theory, whereas research questions without hypotheses are more open-ended opportunities to satisfy one’s curiosity. For instance, a student studying self-disclosure patterns among psychotherapists might have specific hypotheses about the relationship between self-disclosure and friendship (e.g., “Therapists who are high in self-disclosure with their patients have fewer personal friends”) but no clear expectations about the relationship between self-disclosure and stages of therapy (e.g., “What is the relationship between therapist self-disclosure with patients and the stage of psychotherapy?”).

A more popular way of combining research questions and hypotheses is to use research questions as more general investigatory themes and then follow up with specific hypotheses that make predictions in a testable form. An example of an experimental dissertation stating both research questions and hypotheses comes from Jennifer Mitchell’s (2012) study of the impact of yoga as an alternative intervention for mitigating prenatal depressive symptoms. Mitchell’s research questions and hypotheses are expressed in this straightforward manner:

Research Question 1: Does participation in a yoga treatment regimen lead to a significant decrease in depression?

H1: The treatment group receiving the yoga sessions will show a greater difference in pre- and post-treatment scores on the CES- D summary scale than the treatment control group receiving parenting education sessions.

Research Question 2: What type of depressive symptomatology is most impacted by a yoga treatment regimen?

H2: The treatment group receiving the yoga sessions will show a greater difference in pre- and post-treatment scores on the CES- D depressed affect and somatic vegetative signs subscales than the treatment control group receiving parenting education sessions. (p. 12)

In another dissertation example, Ellen Goldberg (2003) began with the general research question “What are the parent and child factors that predict diabetes self-care behaviors, particularly around prevention, detection, and treatment of hypoglycemia at home and in the school environment?” In contrast to those in Mitchell’s two-group experimental

design, the primary hypotheses of Goldberg’s study are written to describe relationships between variables and to suggest a correlational analysis of the results:

Hypothesis 1(a): Parental fear of hypoglycemia is inversely related to parental expectations for child responsibility for diabetes self-care autonomy in school. Hypothesis 1(b): Parental fear of hypoglycemia is inversely related to parental expectations for child responsibility for diabetes self-care at home. Hypothesis 2(a): Parental perception of the child’s temperamental regularity and task orientation are inversely related to parental expectations for the child’s responsibility for diabetes self-care autonomy in school. Hypothesis 2(b): Parental perception of the child’s temperamental regularity and task orientation are inversely related to parental expectations for the child’s responsibility for diabetes self-care autonomy at home. Hypothesis 3: Parental expectations for diabetes self-care autonomy are lower in the home environment as compared to the school environment. Hypothesis 4: Parental expectations for diabetes self-care autonomy are inversely related to metabolic control. (pp. 51–52)

The variables noted in the hypotheses also need to be operationalized. In this case, Ellen Goldberg (2003) tested the hypotheses using measures such as the Fear of Hypoglycemia Survey and the Diabetes Family Responsibility Questionnaire. The research questions and hypotheses in the previous example represent fairly straightforward research designs representing basic experimental and correlational paradigms. We are now witnessing an increasing number of dissertation studies that include moderating and mediating variables and more complex analyses. Examples of these types of studies are included in our discussion of the presentation of results in Chapter 6.

Lack of clarity in the writing of hypotheses is one of the major stumbling blocks that our students face. Here we present a couple of examples and offer our suggestions for achieving a better hypothesis statement.

Example 1.

Save your time - order a paper!

Experience

Our mission

Latest Posts

Talk to us...