Rich Voters, Poor Voters

Over the past few months, Bryan Palmer and I have been debating the merits of using aggregate-level data to learn about individual-level behaviour. He’s been gracious enough to post a comment of mine on his recent tabulations.

Bryan also has his regular betting markets roundup, which I ought to link to more regularly. The markets have Labor about a 62-63% chance, which is approximately where the Coalition was in early-2005.

This entry was posted in Australian Politics. Bookmark the permalink.

6 Responses to Rich Voters, Poor Voters

  1. Rough Liberal primary versus high SES for Victorian CCDs in 2001:

    I don’t think anyone doubts that high-income, high-education voters are the bedrock of the Liberal vote. But of course plenty of other segments also skew their way.

  2. conrad says:

    I just looked at your comment about the voting stuff. I don’t think you do need to look at the individual to learn the pattern, but you also shouldn’t be looking at the overall mean.

    I think it its a clear cut case of where a two level hierarchical model with people nested under State would describe the data correctly. You can do that analysis with SPSS 14 now, and there is a free package (HLM) which is even better (the interface on SPSS is a disaster). You might be interested in the NATSEM data given your research — there is a book that examines this data (as an example) in terms of amount of homework children do, and you find to distinct clusters — one where it helps and one where it is negative and how this relates to variables like SES. If you a really enthusiastic, you can now do this sort of analysis using constructs as well, so you end up with something like a cross between a SEM and a HLM model (although you’ll have to use S-plus)

  3. Andrew Leigh says:

    Conrad, I wrote a paper a couple of years ago, looking independently at the effects of own income and neighbourhood income. Since the neighbourhood income data was drawn from a different source (rather than being aggregated up from the same survey), I didn’t use HLM.

  4. Sacha says:

    Andrew, could you give a precis of your discussions about “using aggregate-level data to learn about individual-level behaviour”?


  5. Fred Argy says:

    Andrew, while the national betting favours Labor, I understand that the individual seat betting favours Howard. Can you please clarify this point?

  6. Andrew Leigh says:

    Sacha, my two main posts on the issue are here and here.

    Fred, I’m confused too.

Comments are closed.