538's projections on undecideds

[daboese]

They have done it considering their deviations from the polling numbers.
Hence, the effects that they have taken into seem to match the numbers even better.
This is just another variable to play around it when fitting the data, and they felt like they should take it into account now. There's nothing wrong about that, possibly they will refine the model later.

[daboese]

Anyone who had to do with modelling will see that 538 is just trying to adjust their models in order to predict the majority of the polls better.
Allocating the undecideds is a good idea at this point in time, and they have a reasoning to do so- even if they really take into account some other effects.

Their numbers base on:
--snip--
Specifically, Obama overperformed:

1. In states with high African-American populations;
2. In states that share a border with Illinois (no, Kentucky doesn't count);
3. In states with highly educated electorates;
4. To a lesser extent, in the South (as indicated by the number of evangelicals), even after accounting for the above variables.

Meanwhile, he underperformed his polls:

1. In the Appalachian states (as indicated by the number of respondents who identify their ancestry as 'American', a practice concentrated in the Appalachian region);
2. In states with low education levels;
3. And in states with a high number of Catholics.
--snap--