Why Dem Primary Anomalies Must Be Thoroughly Investigated Before Choosing a Nominee

[Zynx]

7. Exit polls have missed drastically overseas before as well. The UK comes to mind.

They weren't even close in 1992 and overstated Labour's lead in 1997 by a good margin.

When samples are much larger than they are here, then I think they're useful. In the case of NY, they weren't terribly much larger than the pre-election polling and they by definition aren't random.

A serious problem with our exit polling in this country is that it's poorly funded. You need to get a very large array of different samples in order to correctly project an outcome because of geographic variations. For example, if you draw too much from Madison for the white demographic in Wisconsin, your projection of the state as a whole will be way off.

An on-the-ground sample of about 1,000 for a state probably isn't sufficient because of the shortcomings of the sampling methods. Telephone samples of that size are probably more accurate because they tend to be more properly weighted.

The NY initial exit polls were so out of line with all pre-election polling that I think it's clear that they should be disregarded. The final results were almost bang-on with the pre-election polling.