Square footage as a predictor of vote and party affiliation?

After reading Steve Sailer’s discussion of unmarried Democrats living in crowded cities and Republicans with large families, we decide that the ultimate predictor of political leanings would be . . . square footage of your residence. It has all the right properties:

– Within any state, people in bigger houses vote more Republican. Check.

– Lower cost-of-living states, where houses are bigger (I assume), are more Republican. Check.

– In crowded coastal states, there is little difference in square footage between the houses of the rich and the poor; in less-crowded, poorer inland states, rich and poor differ more in house size. As a result, the “square footage” model predicts that the rich-poor gap in Republican voting should be larger in poor than in rich states. Check.

I don’t know of any datasets that have voting or party ID along with square footage–although, with a large amount of effort it should be possible to put something together using public voter registration information. Also, I can’t really see anything useful about the hypothesis (that square footage is an excellent predictor of who you vote for), even if it’s true. Nonetheless, the idea amuses me.

P.S. Seeing as I live in a cramped NYC apartment with no understanding of square footage at all, so I’d appreciate others’ input on this. (Also, I have no idea how this would work in other countries.)

11 thoughts on “Square footage as a predictor of vote and party affiliation?

  1. Not really a comment, but just a notation that some of these census variables might be of interest. Not sure what granularity they are available, but I could check.

    RHU00O1 Housing Units 2000 Rooms in Housing Unit (Housing Units):Owner Occupied 1 room
    RHU00O2 Housing Units 2000 Rooms in Housing Unit (Housing Units):Owner Occupied 2 rooms
    RHU00O3 Housing Units 2000 Rooms in Housing Unit (Housing Units):Owner Occupied 3 rooms
    RHU00O4 Housing Units 2000 Rooms in Housing Unit (Housing Units):Owner Occupied 4 rooms
    RHU00O5 Housing Units 2000 Rooms in Housing Unit (Housing Units):Owner Occupied 5 rooms
    RHU00O6 Housing Units 2000 Rooms in Housing Unit (Housing Units):Owner Occupied 6 rooms
    RHU00O7 Housing Units 2000 Rooms in Housing Unit (Housing Units):Owner Occupied 7 rooms
    RHU00O8 Housing Units 2000 Rooms in Housing Unit (Housing Units):Owner Occupied 8 rooms
    RHU00O9 Housing Units 2000 Rooms in Housing Unit (Housing Units):Owner Occupied 9 rooms or more rooms
    RHU00OAGG Housing Units 2000 Rooms in Housing Unit (Housing Units):Owner Occupied Aggregate number of rooms
    RHU00R1 Housing Units 2000 Rooms in Housing Unit (Housing Units):Renter Occupied 1 room
    RHU00R2 Housing Units 2000 Rooms in Housing Unit (Housing Units):Renter Occupied 2 rooms
    RHU00R3 Housing Units 2000 Rooms in Housing Unit (Housing Units):Renter Occupied 3 rooms
    RHU00R4 Housing Units 2000 Rooms in Housing Unit (Housing Units):Renter Occupied 4 rooms
    RHU00R5 Housing Units 2000 Rooms in Housing Unit (Housing Units):Renter Occupied 5 rooms
    RHU00R6 Housing Units 2000 Rooms in Housing Unit (Housing Units):Renter Occupied 6 rooms
    RHU00R7 Housing Units 2000 Rooms in Housing Unit (Housing Units):Renter Occupied 7 rooms
    RHU00R8 Housing Units 2000 Rooms in Housing Unit (Housing Units):Renter Occupied 8 rooms
    RHU00R9 Housing Units 2000 Rooms in Housing Unit (Housing Units):Renter Occupied 9 rooms or more rooms
    RHU00RAGG Housing Units 2000 Rooms in Housing Unit (Housing Units):Renter Occupied Aggregate number of rooms
    RHU001 Housing Units 2000 Rooms in Housing Unit (Housing Units) 1 room
    RHU002 Housing Units 2000 Rooms in Housing Unit (Housing Units) 2 rooms
    RHU003 Housing Units 2000 Rooms in Housing Unit (Housing Units) 3 rooms
    RHU004 Housing Units 2000 Rooms in Housing Unit (Housing Units) 4 rooms
    RHU005 Housing Units 2000 Rooms in Housing Unit (Housing Units) 5 rooms
    RHU006 Housing Units 2000 Rooms in Housing Unit (Housing Units) 6 rooms
    RHU007 Housing Units 2000 Rooms in Housing Unit (Housing Units) 7 rooms
    RHU008 Housing Units 2000 Rooms in Housing Unit (Housing Units) 8 rooms
    RHU009 Housing Units 2000 Rooms in Housing Unit (Housing Units) 9 rooms or more rooms
    RHU00AVE Housing Units 2000 Rooms in Housing Unit (Housing Units) Aggregate number of rooms
    NBD001 Housing Units 2000 Rooms in Housing Unit (HU):Number of Bedrooms 1 bedroom
    NBD002 Housing Units 2000 Rooms in Housing Unit (HU):Number of Bedrooms 2 bedrooms
    NBD003 Housing Units 2000 Rooms in Housing Unit (HU):Number of Bedrooms 3 bedrooms
    NBD004 Housing Units 2000 Rooms in Housing Unit (HU):Number of Bedrooms 4 bedrooms
    NBD005 Housing Units 2000 Rooms in Housing Unit (HU):Number of Bedrooms 5 or more bedrooms
    NBD00NONE Housing Units 2000 Rooms in Housing Unit (HU):Number of Bedrooms No bedroom

  2. "Also, I can't really see anything useful about the hypothesis (that square footage is an excellent predictor of who you vote for), even if it's true."

    It's not interesting which predicts people's voting? That sounds rather strange from you. Am I missing something here?

  3. Does the decline of the popularity of the Republicans track with the decline of the value of the suburbs?

    Atlantic had a good article on this recently.

    See also a recent KLD blog post about responsible real estate investing.

  4. I don't know for sure, but I would be profoundly surprised if campaign dataminers don't use this kind of data, especially since, as you mention, a lot of data about individual housing characteristics is publicly available (from property tax rolls). I know in my county, I can go online and find not only the square footage, but the type of foundation, whether or not it has central air and heat, what the exterior walls are composed of, number of rooms, number of bathrooms, whether it has a garage, the lot size, the size of the slab, etc., etc., of any house in the county.

    As you say, there is probably a pretty direct correlation between square footage and voting patterns. But combined with lots of other behavioral information (club memberships, online and mail order purchasing info, and so on), political data miners can probably build lots of useful, actionable models.

  5. Interesting idea. It seems though that current square footage would be endogenous to current party affiliation or voting preference, and would require some form of identification. Maybe, the square footage of their parents house growing up?

  6. Lionel Berber, the current editor of Financial Times suggested people who live in houses where the trees are taller than the house, vote republican.

    There is another hypothesis related to voting behavior and housing by Charlie Cook that I can't remember at the moment.

  7. Funny — last time the trees in my neighborhood were above the roofline, I was in an 80%-D precinct in Minneapolis. I am pretty sure Potrero Hill, with its piddly little treescape in famously liberal San Francisco, is below 70%. Though I grant, an erstwhile locale, West Oakland, no trees of note, has both beat in the 90s.

    Trees. Whatever happened to gender, age, income, geography and race?

  8. @Allison

    I've found that it isn't always the best outlook to assume that all unexploited opportunities are taken advantage of for THIS particular field.

    In my opinion the real limit of this model is its constrained predictive power. We may show that areas with large sq. fottage homes trend republicans but it is about as helpful as showing an urban/rural divide. What I mean to say is that we already know the outcome: political leanings and determining the spread based on home buying has limited value in discovering the input. that being said, discovering the input probably has limited value as well. Might help for adding as another term to the regression though.

  9. This seems pretty predictive to me.

    In Europe they don't even have square footage (they use meters, squared or cubed) and they don't tend to vote either Republican or Democratic. Those people who live in Europe and can quote their house size in square feet probably do vote in the US.

    How much more predictive can you get? It even works on NA!

Comments are closed.