Daily Poll Watch: Measuring how certain polls are skewed one way or the other

This seems like as a good time as any to explore the subject of what people in the polling business call “house effects,” which are systemic factors that tend to skew poll results in favor of one political party or candidate or the other.

These house effects aren’t really intentional efforts to come up with biased results. Some of them are just by-products of methodology. For example, a poll that excludes cell phones usually will have a house effect unfavorable to Democrats.

Both of the studies cited here include charts. Here’s the one that accompanies THIS PIECE by Simon Jackman, which was published just today:


And here’s the chart that goes with THIS ARTICLE by Nate Silver from this past June:



The differences between the two charts in terms of rating the house effects of certain polling firms presumably is attributable, at least in part, to the time lapse between the two studies.

Notice, too, that polls commissioned by Fox News don’t show a lot of house effects, especially in the first of the charts above. That doesn’t surprise me. I’ve argued here on numerous occasions that Fox News polls generally are as reliable as any other (even though the reporting of the results on Fox News Channel is sometimes skewed).


Leave a Reply

Your email address will not be published. Required fields are marked *