Reply #68: First of all, we have no idea which data TIA is using -

[Lurker321]

since no one knows what the raw data or others look like, since Mitofsky has not released it yet. From what I understood, he is using raw data that was weighted but not fully adjusted yet. The weighting accounts for it being a little more accurate than the 8% glaring discrepancy in the numbers I posted before.

Basically, since the raw numbers or the model, or the precincts breakdown or anything else really was not released by Mitofsky, any analysis of the data is premature and very likely to be wrong. Mitofsky promised to release the data (I think with the Roper center) within 3 months of the elections - so there are a couple of weeks to go. You and I know that simple weighting does not make the data accurate, the adjustments that need to be made to make it truly accurate are complex, have to take into account a slew of factors, and are more of an art than a science.

I am waiting for the raw data numbers to be released. It will be interesting to check the per-precinct raw data collected with the precincts' official election results, for example. Those should match fairly closely, within the sample MOE, I would presume, if the sampling in those precincts was truly random. No adjustments necessary.