Polls Need More Depth, But Same Clarity
A theme underlying the discussion prompted by David Moore's post is the changing meaning of "undecided" across a campaign timeline. Starting way back in the primaries, an "undecided" person may not be able to choose among candidates seeking her party's nomination who don't vary in their policy positions very much from each other. This is a different circumstance than a person in the last three weeks of the campaign who is weighing the policy positions of the candidates but hasn't arrived at a choice yet.
These differences reflect, among other things, various stages in the process of crystallizing opinions with the passage of time in a political campaign. We don't understand this process or its effects very much because of a lack of empirical information. But George Bishop's proposal for more cognitive interviewing on the topic would be a good start.
If we believe that "undecided" means different things at different stages of a campaign - and we begin to report two forms of the preference distribution - how difficult is it going to be for journalists to describe and interpret polling data in these conditions? And will the public be able to consume these interpretations in a way that adds to their understanding of what is going on in the campaign? We don't want to refine methods in a way that improves and clarifies measurement but complicates providing useful information to citizens.
Leave a comment