A Thanksgiving Polling Experiment

One of the really enjoyable aspects of returning to polling this year has been experimenting to see what can be done with the new polling technologies and platforms. It’s amazing how far the industry has come in the five years since I left my old company.

This year, Thanksgiving provided an opportunity for some experimentation. Back in the days when I conducted phone polls, I would never have considered running a survey on Thanksgiving Day. People hated it when pollsters interrupted a regular dinner with a phone call and the outrage would be off the charts for a big family feast. On top of that, the chances of getting credible data from a Thanksgiving Day phone poll were between slim and none.

Now, however, ScottRasmussen.com and HarrisX conduct surveys using a digital platform (see¬†Methodology). So, I wanted to see if would could get reliable data on one of the year’s biggest holidays. To test this, we ran one survey on¬†Wednesday night and into the early morning of Thanksgiving day. A second was conducted Thanksgiving night and into the early morning of Friday.

In both cases, we included questions we’ve asked before. Some involved the president’s Job Approval and a Generic Ballot test while others asked about holiday shopping. I was delighted to find that we obtained reasonable samples and results with both surveys. The Job Approval and Generic Ballot results were consistent with what we have consistently obtained in our tracking data since August 6.

The shopping numbers showed a gradual increase in the number of people who had started shopping and who had finished their shopping. It’s quite consistent with what we expected.

One other aspect of the test was to compare the results from the two surveys on Thanksgiving related topics. On Wednesday night, we asked about things that people expected to do. On Thursday night, we asked what they had done. The results were very similar. In both cases 72% said turkey was the main course and 49% said 3 to 8 people were around the table.

One fun result is that on Wednesday night, 62% expected to eat too much. However, after the meal was finished, just 48% said they had eaten too much. Perhaps 14% restrained themselves a bit or perhaps their definition of eating too much changed. However, based upon the other results from the surveys, the difference was not caused by a flawed sample.

And, even more important, 92% said they have a lot to be thankful for this Thanksgiving.

I am one of them! Hope you are too.

Posted in Deeper Currents

Scott's Newsletter
Sign up for Scott's newsletter and get his political insight delivered right to your inbox!