April 27, 2016 (updated May 26)
There were three exit polls yesterday in CT, MD, PA. Sanders exit poll share declined from the poll to the vote in two of the three elections. As usual, the exit polls were forced to match the recorded vote.
The difference between Clinton’s adjusted exit poll and recorded share were:
CT 0.01%; MD 0.10%; PA -0.17%
In 21 of 23 primaries, Sander’s exit poll share exceeded his recorded share.
The probability of this being due to chance:
P = 1 in 30,000 = binomdist(2,23,0.5,true)
In 9 of 23 primaries, Sanders exit poll share exceeded his recorded share by more than the margin of error. The probability of this being due to chance:
P = 1 in 441 million = 1-BINOMDIST( 8,23, 0.025, true)
The pollsters ALWAYS force the unadjusted exit polls to match the recorded vote. Where are the unadjusted exit polls? The American Association for Public Opinion Research (AAPOR) discusses the matching process in Explaining Exit Polls.
In close races, the projection models also employ actual vote totals, first in sample precincts as it becomes available and then at the county level for all counties in a state as they become available. It is important to note that after the votes have been counted, the exit poll results are adjusted to match the actual election outcomes. It is in this way that the final exit poll data can be used for its primary and most important purpose – to shed light on why the election turned out the way it did.
Note: It’s not just in close races. Unadjusted exit polls are adjusted to match the recorded vote in ALL races. The pollsters need to provide the actual exit poll respondent data for all precincts. They need to provide the data and then explain the rationale for making the adjustments to match the vote. But they won’t. Ask yourself WHY?
The AAPOR never mentions election fraud as a likely cause of the discrepancies. In actuality all of the adjusted exit poll crosstabs contaminate the true statistical results and are misleading as they do not reflect the the actual responses of those exit polled.
Exit pollsters at Edison Research should not be making adjustments. But it is standard operating procedure. It is unscientific and hides the actual exit poll results. It serves to cover-up the fraud which is measured by the recorded vote discrepancy .
Final Adjusted Exit Polls (forced to match the recorded vote)
|1234 total respondents|
|1364 total respondents|
|1425 total respondents|
May 26 update:
THE CONNECTICUT EXIT POLL
How did Sanders early poll drop from 55% to 47%? It would be nice to know how many were polled at this point. According to Edison Research, about 2/3 of the total polled (1223).
The MoE for 815 respondents is 4.5%. For 1223, it is 3.6%. So Sanders exit poll dropped by 8%, far above the 4.5% MoE.
Should we add CT to the list of 11 of 26 exit polls which exceeded the MoE for Sanders? The 1 in 77 billion probability would become 1 in 2.4 TRILLION.
“Edison Research exit poll interviewers call in exit poll results three times during election day – once in the late morning, once in the mid-afternoon, and once shortly before the polls close in a state. The exit poll data that is released at 5PM to the news organizations comprising the National Election Pool (NEP members are ABC, CBS, CNN, FOX, NBC and the Associated Press) and any other news organizations subscribing to the exit poll include about two-thirds of the interviews that will be conducted on an election day. The exit poll results that are released around poll closing include nearly all of the voter interviews that are conducted during election day.”
The table below was created by Theodore de Macedo Soares (firstname.lastname@example.org)
CNN is the source of the state exit polls which were downloaded shortly after closing.
The NY Times is the source of the reported vote counts.