# How went the 2008 election, looking only at the votes of rich, middle-income, and poor?

I posted the pretty maps at 538. (I’d post them here, but Jeff was complaining that I was crossposting too much.) But one thing that people do like here is R code, so here’s some:

```
> M1 <- glmer (rvote ~ z.inc2*z.state.income.full + (1|inc2) + (1 + z.inc2 | stnum), family=binomial(link="logit"))
> display (M1)
glmer(formula = rvote ~ z.inc2 * z.state.income.full + (1 | inc2) +
(1 + z.inc2 | stnum), family = binomial(link = "logit"))
coef.est coef.se
(Intercept)                -0.06     0.05
z.inc2                      0.52     0.07
z.state.income.full        -0.49     0.07
z.inc2:z.state.income.full -0.27     0.11

Error terms:
Groups   Name        Std.Dev. Corr
stnum    (Intercept) 0.20
z.inc2      0.30     0.62
inc2     (Intercept) 0.06
Residual             1.00
---
number of obs: 20510, groups: stnum, 49; inc2, 5
AIC = 27668.9, DIC = 27652.9
deviance = 27652.9
```

I used the estimates from this model to extract McCain’s estimated two-party vote share for each income category within each state. I then took a weighted average for each state (weighting by the number of respondents in each income category) and did one final adjustment by shifting the estimates for each category so that the average came out to the same as the actual vote outcome within each state.

I won’t show you this R code because it’s too damn ugly; I’m embarrassed.

OK, OK, here are the graphs:

Sorry, Jeff.

Update: My original maps were slightly wrong–they were based on unadjusted poll numbers. I reposted the corrected maps 6.28pm.

## 15 thoughts on “How went the 2008 election, looking only at the votes of rich, middle-income, and poor?”

1. How did you decide on who is rich, MI and low income? Is it 33% each? And within state or across the whole country?

2. How did Kansas go for McCain if only rich voters liked him? Are there that many rich Kansans, proportionally?

3. What are the electoral college splits in those maps? Ok, the third one is easy (538 – 0 or maybe 535 – 0 since you didn't include Alaska and I'm pretty sure that Hawaii goes to Obama in all three scenarios)

And the result for Kansas doesn't make any sense to me. Obama lost Kansas. So, if your figures are correct, he lost rich people by such a wide margin that overwhelmed his win among middle and lower class voters.

4. Ugly r code is better than no r code. Those of us with less programming chops would likely benefit from seeing it.

5. Peter: See the 538 post for info on the income categories.

Tim, William: Here are the estimates for McCain's share of the 2-party vote in Kansas for each of the 5 income categories:

0.37 0.45 0.49 0.54 0.57

And here are the number of survey respondents in each category:

35 75 75 80 14

I agree that this doesn't quite work out. This might have something to do with the survey weighting. I'll take a look.

Thanks for the beta-testing! The funny thing is, I did look at a few states to make sure they all made sense, but I hadn't checked Kansas. McCain got 57% of the vote there, so I did something wrong…!

6. Hey, sorry about Kansas! The comments here and at the other blog served as useful beta-testing and led me back to my code to fix that. It turns out that I'd presented the unadjusted numbers. Once adjusted for the actual election outcome, you see the new maps above. I don't really know what to say about New Mexico, but there is such a thing as sampling variability!

7. Cool! thanks for the post… would be interesting to see how many new poor voters voted this time in comparison with the previous election… is it true that in this election many new poor voters decided to vote? I guess that's a hard number to estimate though.. would it be possible to estimate that number? With these kind of estimates we would be able to test the hypothesis that the electoral pool has changed and that Obama is far more popular than any other recent president (in terms of bringing more masses to the picture)…

8. Could race be a confounding variable here? It would be interesting to see how poor whites voted compared to how poor blacks did.

9. Good work! It's great to see you doing election analysis again.

Any chance you could shade by either probability that Obama was ahead in each of the categories, or by estimated margin?

The math wouldn't be too hard, though maybe, the graphics might be.

If not, even a spreadsheet of results and regression standard errors would be fun.

10. OI think you might find that the people that emigrated to NM from the East have a higher proportion of Blue State population than AZ has gotten. Just a guess based on the few people I know that moved to the southwest. NM is a lot more attractive to the liberal leaning people I have known for several disparate reasons.

11. New Mexico is potentially explained by the overwhelming number of Hispanics in the "wealthy" categories, coupled with the unusually high percentage of Hispanics in the voting population as a whole. I suspect if you cross-correlate for racial factors as well as income, New Mexico becomes much more reasonable.

Also, what was the sample size for NM? It may just be noise.

12. I'd be interested to see the data adjusted for some cost-of-living index. A household earning \$150K in Mississippi is in a different social category than a household earning \$150K in California or New York.

And, yeah, I'd be interested in the electoral count among the affluent. I can't tell who would win.

13. I'd really like to see cost-of-living adjustments as well. I live in fort lauderdale, fl the worst major city in America if you compare average income to cost-of-living.

14. Andrew,
Thanks for the interesting data. I know there are some kinks in it and some relativism that still needs to be worked through, but I was hoping that you can (whenever you are satisfied with your modeling) provide the type of detail for all states that you showed for Kansas above.

Something on the order of:

State, Tier 1, Tier 2, Tier 3, Tier 4, Tier 5
Kansas, 37%, 45%, 49%, 54%, 57%
(pct) (30%), (20%), (20%), (25%), (5%)

With the bottom (pct) representing the % of the population represented in each income category.

Final note to those who asked: mid-income voters, had their votes been the only ones counted in '08, would have still elected Obama with 290 electoral votes. Rich voters would have elected McCain, 311-227.