Skip to main content

View Diary: Renters make good Democrats, and other demographic observations (129 comments)

Comment Preferences

  •  About predictability .... (0+ / 0-)

    The correlation coeeficient doesn't speak directly to predictability.  There is always, except for perfect correlations of 1 or -1, some variability.  For a sense of predictability, then, what you want to know is how much of that variability you can account for by your measures.  In other words, how much does the % of renters account for whether a district votes Democratic or Republican.

    The measure to use to answer that kind of question is the square of the correlation.  So, if the correlation for % renters is 0.59, then you can account for 35% of the variance in Dem voting districts by knowing the % renters (i.e., .59 X .59).  For age, the highest correlation was .43 so age accounts for 18% of the variance (variability) in which party control a district.  When looked at in this way, the relationships are not as impressive.  However, that doesn't mean that targeting certain populations couldn't affect things on the margins, but just that those margins might be quite small.

    And it's worth repeating that correlations do not speak to causation.  Is there a higher % of renters in Dem districts because Dems tend to rent or because Dems are attracted to places with lots of rental property?

    The trouble with the world is that the stupid are cocksure and the intelligent are full of doubt. Bertrand Russell

    by accumbens on Sun Mar 17, 2013 at 03:21:33 PM PDT

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site