The Commonwealth Leaders Fund (CLF), a newly formed political action committee, released poll data earlier this week showing that the race for governor between Scott Wagner and Governor Tom Wolf is within three percentage points. These results differ significantly from other polling data that has been made public. This raises the question, what poll should you believe?
We took a look “under the hood” of publicly available polling data for the governor’s race. After comparing the older polls to the newly released data, we are more inclined to believe the results of the CLF polling results than the others. There are two main reasons we came to this conclusion. First is sample size, CLF obtained more than 2,000 responses. The earlier polls were 500 responses or less. In polling, talking to more people is better because it smooths out the results. This is reflected in a lower margin of error.
The second reason is a little more technical and has to do with which respondents were selected to participate in the poll. In polling jargon, this is known as sample selection, and in predicting voter sentiment, this can make or break the accuracy of the poll. Because it isn’t practical to talk to every voter, pollsters have to make assumptions about who is going to come out to vote and try to match the characteristics of the actual voters to the characteristics of the sample. When we looked at the details of the CLF sample, we saw that the demographics of their sample more accurately reflect the demographics of people who have historically come out to vote in gubernatorial elections.
The larger sample size along with the historically accurate demographics and a couple other technical items lead us to believe that the CLF polling results are probably closer to reality than other publicly available data. In other words, the race is much more competitive than has been previously reported. This is excellent news for Scott Wagner.
A PDF of the poll results and methodology can be found here.