Overall relationship between party identification and presidential vote

To determine the overall relationship between party identification and presidential vote in Table 1A, focus on the change in percentages as you move across the table from left to right. We would expect the percent for Clinton to decline and the percent for Trump to increase as we go from strong Democrats to strong Republicans. For the most part, that is exactly the pattern that we find in Table 1A. However, independent Democrats do not quite fit the expected pattern. They are slightly more for Clinton than are weak Democrats. Thus, we do not have a perfect monotonic relationship—i.e., there is not a constant decline in the percentage for Clinton (or a constant increase in the percentage for Trump) as we go across the table.

How should we interpret this anomaly concerning independent Democrats?

1. It could be that the relationship truly is monotonic, but Table 1A fails to capture that because of sampling error. In other words, perhaps if all independent Democrats had been interviewed, we would have found a lower percentage voting for Clinton. Samples often do not perfectly reflect the population that they represent, and the 285 independent Democrats in this sample easily could be a little different from all independent Democrats.
2. It could be that our conceptualization of party identification is a little simplistic. Perhaps independent Democrats really are not less Democratic in their voting behavior than weak Democrats. They may be very similar in behavior. In fact, we can see that there is a similar pattern between independent Republicans and weak Republicans in Table 1A, with independent Republicans voting more strongly for Trump than is true for weak Republicans. This also suggests that weak partisans and independent leaners are very similar in their voting behavior.