Skip to main content

View Diary: 538 on turnout and demographics (15 comments)

Comment Preferences

  •  I'm not so sure he's wrong (0+ / 0-)

    What's the problem with dividing the groups the way he did?  It's a pretty common way to do it, and is probably the way most exit polls did it.

    I'll have to reread his article, but on first reading I'm not seeing a big problem with what he did.

    •  See ecological fallacy (1+ / 0-)
      Recommended by:
      tardis10

      If you ask individual voters, you find that all the difference was  more D no-shows.
      Let me try to explain the ecological fallacy by describing an extreme case. Say that you divided voters into two groups: those who could curl their tongues and those who couldn't. Then you try to compute what changed in the vote by seeing what happened to the turnout of those groups, and how their votes went. You'd find that turnout of the groups changed pretty much the same and the votes of the groups were similar. All that changed was the within-group votes.

      Those would be ridiculous groups, you say. Yes, but any broad grouping shares that problem to a significant degree. That's why the results of this type of analysis are extremely sensitive to what sort of groups you choose. The results don't answer the question you want answered: was it vote switching or who turned out that changed? That question is answered by tabulating the actual voting units: individual voters.

      Michael Weissman UID 197542

      by docmidwest on Thu May 08, 2014 at 03:22:12 PM PDT

      [ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site