Could it be that both types of measures are tapping the same construct, but tapping entirely different aspects of that construct?

Implicit and explicit or self- report measures of the same construct often correlate poorly, or moderately at best with each other. So are our implicit measures assessing the same construct as our explicit measures? Many studies have produced results in which implicit evaluations fluctuated while explicit evaluations remained stable, and vice versa, but does that mean that the different types of measures are assessing different features? Could it be that both types of measures are tapping the same construct, but tapping entirely different aspects of that construct? Is there any way we can determine whether this is true? How would you test the convergent and discriminant validity of an implicit measure? Would you expect one implicit measure to be more likely to correlate with another implicit measure than with an explicit measure? Why?