Problems with Simon Poll
Monday, Feb 24, 2020 - Posted by Rich Miller
* We discussed the Simon Poll results last week, but Rick Pearson took a look at the numbers…
However, the poll, which has a margin of error of 4.5 percentage points among Democrats, relies on a methodology that raises questions about its accuracy.
The survey does not weight its results based on the demographic makeup of Illinois voters, including by gender, geography or race. For example, of the poll’s 1,000 registered voters, 56% were identified as male and 44% female. The federal census, however, shows a majority of the state’s population is female while 2016 exit polling showed the state’s Democratic electorate was 54% female.
The polling sample also did not accurately reflect the percentage of the vote that came from strongly Democratic Chicago compared with less Democratic suburbs and Downstate. Only a quarter of the poll’s Democratic results came from the city of Chicago, while actual voting in 2016 showed the city accounted for 35% of the state’s total Democratic primary turnout.
In addition, the new survey skewed heavily toward older voters, compared with 2016 exit polling that showed roughly 40% of the Democratic electorate was under the age of 45. The new poll’s racial breakdown of 60% white, 24% black and 6% Latino tracked closer to Illinois voting demographics in the last presidential primary.
Crosstabs are here. Discuss.
- Oak Parker - Monday, Feb 24, 20 @ 2:26 pm:
The poll numbers could change so much after South Carolina and Super Tuesday. Some candidates may be out of the race by 3/17 and if Bernie dominates on Super Tuesday, he’ll likely get more votes as his nomination looks inevitable.
- Not a Billionaire - Monday, Feb 24, 20 @ 2:44 pm:
It means that Prizger is even more popular and Trump even less and who knows for the primary.
- Precinct Captain - Monday, Feb 24, 20 @ 2:58 pm:
It’d be interesting if Simon had done any weighting, even showing results under different forms of weighting, not too dissimilar from part of what was done with the New York Times live polling experiment in 2018 when they reported the polls under different turnout scenarios.
- Oswego Willy - Monday, Feb 24, 20 @ 3:06 pm:
=== Discuss===
Could this be that POTUS’ overall numbers are measurably worse, and the Guv might have better overall numbers too?
Trying to arbitrarily look at possible shifts
- Hot Taeks - Monday, Feb 24, 20 @ 3:07 pm:
Extrapolating from this, I believe Senator Sanders has a decent lead (5-10%) in Illinois at the moment. This isn’t surprising as he’s currently the Democratic front-runner.
- IL4Life - Monday, Feb 24, 20 @ 3:11 pm:
In terms of the Democratic primary results (only comparison available), the Paul Simon Poll basically had the same support, plus or minus the margin of error, as a “Victory Research” poll of 1,200 likely Democratic primary voters at about the same time: https://projects.fivethirtyeight.com/polls/20200221_IL.pdf.
- Responsa - Monday, Feb 24, 20 @ 10:46 pm:
Pollsters need to do better. People need to rely less on polls. I’ve had problems with Simon polls in the past so it’s no surprise to me there are again questions being raised about their methodology on this one. Frankly, when it was posted last week people in the comment section here were much more interested in throwing spitballs back and forth to each other than they were in considering the poll itself, and whether it was a worthwhile contribution toward understanding the composition and attitudes of the current Illinois electorate.