Skip to main content

View Diary: Electing a president: The native son effect (16 comments)

Comment Preferences

  •  interesting start... (2+ / 0-)
    Recommended by:
    d to the f, plf515

    I would think that regional variations are a major factor that aren't simply driven by the demographic variables in your model.  Have you thought about including region dummy variables -- at least for the Northeast and South.  

    I'd also be somewhat concerned that the state dummy variables based on specific  candidates may mask other effects -- perhaps these regional effects.  

    I might also expect some interaction effects -- income may matter differently in the South than the North (as Gelman has shown).  

    I think these types of models can be sort of fun to mess around with, but you can spend endless time playing with the model specifications to support a wide variety of conclusions.

    Any link to your dataset?  Any luck with the larger dataset issue (other than what I saw on the stats geeks of daily kos list)?

    Good luck

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site