I had trouble transferring my personal popularity or support to the broader cause of the Democratic party.This caused me to ponder the irony that a supposedly popular president could not convince voters to elect Clinton who he described as the best-prepared candidate we've seen in a long time. Let's consider a politically incorrect thought, shall we?
Suppose the Bradley Effect works in two ways. The first is of course the original way in which voters report to pollsters they will vote for a black candidate for whom they have no intention of voting.
I hypothesize the second way is, when asked by pollsters their opinion of a black incumbent office holder, voters will voice approval-of-performance they do not, in fact, feel. The reason for making an incorrect statement is identical in both cases, namely a desire not to appear racially biased.
If the Bradley Effect works for both incumbents and candidates, then just maybe President Obama had substantially less "personal popularity" than pollsters reported. That would explain the "trouble" Obama claimed to have. Perhaps he couldn't transfer popularity he didn't actually have as his poll numbers were partially an artifact of social desirability bias.