Classroom activity on poll tracking

by Andrew Gelman on September 26, 2012 · 2 comments

in Campaigns and elections,Education

Mark Foster writes:

I teach elementary statistics and regularly use the “Guessing Ages” activity from your book early in the semester. What I would like to know are your thoughts on a research project idea I had about a week ago involving the upcoming presidential election.

Students could collect polling tracking data from various sources for each state throughout the month of October. The object being to see if any perceived trends or patterns would be useful in making predictions. The results would be presented just prior to election day.

Another idea would be to collect voter registration percentages of Democrats, Republicans, and Others (all the rest lumped into one group) from each state. Based on previous turnout history information, calculate the winner of each state and who would ultimately win the election. (These numbers could collected from each state’s website.)

My reply:

If they’re going to do this, they should do senate or governor’s races where the data are more sparse and the results will be more interesting. Different students can do different states.

{ 2 comments }

Marty September 26, 2012 at 11:22 am

Watch out. At least 16 states don’t have party registration.

Marty September 26, 2012 at 11:23 am

So what the polls say about party ID may be drastically different from the other data you’ll collect.

Plus party ID isn’t necessarily connected to turnout. It’s well-known that Republicans have higher turnout than Democrats in general.

Comments on this entry are closed.

Previous post:

Next post: