The MSN reader poll shows Clinton leading by 5 percentage points with just five days until Election Day. I feel pretty good about that insofar as the trend and difference is similar to the aggregation of traditional polls. But, more interesting, we are using the poll to track sub-demographics. On the national level Trump moved up in early-September, crashed through mid-October and and bounced back the last two weeks. By gender, we can see that the Trump’s movement among women was much more subtle, lightly following the noise of the cycle, than his strong upward movement among men (his stronger demographic). A closer look at the women by age, shows Trump almost flat among 18-29 women, but gradually gaining among 65+ women (his strong sub-demographic again). Meanwhile, Clinton’s progression on all three of these levels: national, gender, and gender-by-age is anything but equal and opposite.


By gender, we can see that the Trump’s movement among women was much more subtle than his movement among men. Notice that Trump’s support among men basically grew at a consistent clip. But, his support from women rose and then dipped, a modified version of the national trend. Meanwhile, Clinton’s movement with men was basically a mirror image of Trump’s movement among women. This is both of their weaker demographic; the demographic that both of them are losing mirrors each other, moving with the noise of the race, while the strong demographics diverge and progress upwards. Unlike Trump among men, does not really gain much ground with women, basically holding flat, as Trump gobbles up “other” and “undecided” voters in the form of men.


A deeper dive into 18-29 women and 65+ women shows a similar patter of the demographics within the demographics. Again, Trump’s stronger combination 65+ women shows growth during the cycle, but his weaker demographic, 18-29 women is basically flat. Clinton’s stronger demographic, 18-29 women actually is a bit noisy this time, while her weaker demographic, older women, is basically a flat line.


Once the election is over we will have more time to examine these sub-demographic in details. We hope to learn more about both how candidates grow support over time and reflect noise during the course of the cycle. Why are some demographics actually moving upwards while other cycle around and around. And, can we spot real movement from noise while it happens!