We all know to take polls with a shaker of salt this early in the presidential race. Well, make that two shakers, because there’s a little thing called margin of error.
As of October 6, here are the margins of error of six of the nine polls used by RealClearPolitics to calculate the RCP Average for the 2016 Republican nomination, along with the sample size of Republican-leaning voters surveyed for each:
As of October 6, here are the margins of error of six of the nine polls used by RealClearPolitics to calculate the RCP Average for the 2016 Republican nomination, along with the sample size of Republican-leaning voters surveyed for each:
These six polls have an average margin of error of +/-4.75%. The other three polls, FOX, Suffolk, and Pew, did not give a margin of error for Republican-leaning voters only, but only for all voters surveyed. Based on the sample size of Republican-leaning voters, we can take a rough guess at the size of the margin of error:
The FOX and Suffolk polls have a sample size of Republicans extremely close to that of Bloomberg and IBD polls, which both have a margin of error of +/-5%. The Pew poll is closest to the CNN poll (margin of error +/-4.5%) in sample size. The average margin of error in the nine polls RCP uses is probably around +/-4.5-5%. Let’s have a look at the latest data from the 2016 race for the Republican nomination:
Taking the +/-4.5-5% average margin of error into account, it becomes hard to say with confidence that Rubio leads Bush, or that Bush leads Cruz. Cruz’s lead over Kasich, Huckabee, Christie, and Paul might be smaller than it looks. With Fiorina and Rubio neck and neck, the margin of error makes it plausible that one may have a lead over the other. While this margin of error gives us another reason not to trust to polls, remember that polling determines a candidate’s eligibility for and placement in debates.
No comments:
Post a Comment