Don2 (Don1 Revised)
Contributor
All?
Iowa Caucus Results Riddled With Errors and Inconsistencies - The New York Times
The counts are made by volunteer staff. This is the first year they have released all this data other than SDEs. The past caucuses may would have numerous errors as well if they were looked at as closely as they are this time.
Here's a breakdown of all the errors, and a visualization of how they are distributed (from Twitter):
I put together a spreadsheet of all precinct issues I (and others) have identified. Currently there are 28 precincts w/ Delegate allocation errors.
The impact of correcting them would be +3.954 SDEs for Bernie. More than enough to overtake Pete.
View attachment 26030
Does that look random to you?
It does not look random. Caveats:
1. This assumes trusting the data which I do not currently.
2. I would rather do a regularly recognized type of test, such as a t-test. Do you have a set of data for Buttigieg and Sanders, like a list of numbers for each one? Something like B={-3,-3,-3,-2,-2,-2,-1,0,1}; S={-2,-2,-1,0,1,1,1,1,1,2,2,2,2,2,3} or whatever. Then, I can see if it is statistically significant or not.
That still won't tell me if the data posted to the Internet is reliable, but it would be half-way there for me. I would then need to try to reproduce the data. (point#1 above)