Comments on FiveThirtyEight: Politics Done Right: Rich and Poor Still Vote Differently in Red and Bl...

I am the only one it's too much work to get half t...

2009-02-26T10:11:04.365-05:00

I am the only one it's too much work to get half through this? Maybe get Nate to translate the numbers and pictures into English, please. Shouldn't assume the conclusions are apparent. JMHO. Thanks,

@Juris: thanks for the acknowledgment; also I appr...

2009-02-24T21:07:55.180-05:00

@Juris: thanks for the acknowledgment; also I appreciate your higher knowledge of statistical methods.

I think you are saying that Andrew erred by trying to use "republicanness" of a state to predict the republicanness of its voters. I agree with you, leaving us wondering what the real underlying factors are.

I wonder what we could learn by doing a factor analysis, going district-by-district, using variables such as the following to try to predict voting:
1. average (or median) income
2. population density ("urbanness")
3. region of country
4. state
5. race (%non-white, etc.)
6. geographic latitude (just to see)
7. geographic longitude (why not?)
8. Some measure of agricultural importance
9. distance to nearest ocean
10. average age
11. distance to nearest state capital
x. (more that I havent thought of)

My hypothesis is that "state" would not matter much in predicting the behavior of a district. On the other hand, you should of course be able to predict a state given enough knowledge about its distrcts.

Put more simply, I don't expect to be able to understand the blueness of a state unless I know the blueness of its districts. Unless that state is in the deep south!

Thank you, Professor Gelman! Still, if Red states ...

2009-02-24T21:00:41.086-05:00

Thank you, Professor Gelman! Still, if Red states are poorer than Blue states, and voting patterns among the poor do not differ as greatly voting patterns among the rich between Red and blue states, shouldn't we have seen some more red states flip for Obama when the economy is the most prevalent issue. He did win some, but not as many as Clinton in 1994 or 1996. Racism may have stunted states from voting Blue this year, regardless of their economic condition. Perhaps if Obama proves himself by 2012 we'll see more states vote Blue. Thank you for the post. Solidifies voting axioms among the rich according to ideology.

@xaffeine: I agree with your conclusion -- and rec...

2009-02-24T18:07:22.951-05:00

@xaffeine: I agree with your conclusion -- and recommendation that the analysis drill down at least one more layer, perhaps congressional district?

But I think you're also endorsing an interpretation by Gelman that borders on tautology (that republicanness predicts republicanness).

Also, because there is a very high overlap between wealth of a state's population and it's aggregate republicanness (Red, Purple, Blue), because of collinearity it is probably questionable to assume that the state's republicanness -- rather than its wealth -- is the appropriate explanatory factor in the analysis.

In a 2-level hierarchical model it would make sense to use "state wealth (median family income)" as a level 2 predictor, along with "individual family income" as a level 1 predictor. But I think it's questionable (tautological) -- even implicitly -- to use "republicanness" as an explanatory factor even at level 2, given that the dependent variable of interest is "probability of voting Republican" at level 1 (the individual respondent level).

Once the context is specified as wealth or median family income, then what Gelman does is reasonable. And further, it makes sense to explore the explanatory contributions of other contextual (level 2) variables (such as % African American, % urban) and other individual (level 1) variables (such as race of the voter, urban vs. rural residence).

To simplify, the data merely says the following:1....

2009-02-24T16:06:48.305-05:00

To simplify, the data merely says the following:

1. For individuals, democraticity is strongly correlated with income.

2. For states, the correlation is attenuated, toward the high end of the income spectrum, for the most democratic states, which also tend to be the wealthier states.

This only make us wonder what other factors affect state-by-state voting patterns.

I assert that we need to look at other degrees of geographical granularity. I think you could do state-by-state clustering to identify regions of the country that vote differently and this would "explain" much of the discrepancy.

But we might also want to go down to the level of counties (or congressional districts) and do some analysis.

My own conjecture is that there are cultural differences that can be mapped, if not well understood, that play the greatest role.

andrew said..."We need to dig deeper. Numbers alon...

2009-02-24T14:48:28.818-05:00

andrew said...
"We need to dig deeper. Numbers alone don't paint the picture -- they only stencil the outline. My suggestion to this site is to add some, dare I say, non-statistical context.... In other words, you're going to have to add a non-stat-head to the lineup..."

Nate said...
"In the case of something like [x]... where the ratio of subjective/qualitative to objective/quantitative information is relatively high, I'm pretty certain that the limitations of hewing to a rule-based approach (like a computer program) outweigh the advantages... I'm also pretty certain that the gap can be closed with better model-building..."

If one is (like andrew) interested in looking at context and subjective/qualitative information in order to (like Nate) build better models, it seems to me that several comments are already on the right track:

ttorack said...
"The discussion of this post should be why do urbanites care less about wealth as their income increases... My guess is that urbanites have a lot more contact with people outside of their income bracket, hence more empathy once they reach an income level that supplies them all of their needs..."

Michael (mbw) said...
"If... then almost all R votes are irrational... Perhaps the big difference is that in culturally bleak regions a lot of rich people don't see any way to be happier other than to amass more of the sorts of material things which are useful smaller amounts."

Aranfell said...
"Is RRR due to (lack of) education? Perhaps referencing against statistics on education level (relative to income, if that data is available) would help clear this up."

By analyzing the data in a series of two variables...

2009-02-24T12:43:16.595-05:00

By analyzing the data in a series of two variables at a time, it is hard to see the interaction effects. The world is not black and white and it is also not bivariate.

I wonder if the effects would be clearer using one of the tools like CHAID?

Welcome, Andrew, and thanks for the cool charts. ...

2009-02-24T12:41:54.340-05:00

Welcome, Andrew, and thanks for the cool charts. They are very interesting, but I'm not sure I'm seeing that any deep conclusions can be drawn from them. Rich people in Republican states are more likely to vote Republican, sure, but apparently from the charts, so are poor people in these states. The gap in Republican voting between rich and poor is wider in these states, but these states are also poorer on the average so perhaps there are fewer rich people. Does your definition of rich vs poor change from state to state so as to capture the top/bottom 1/3 of voters by income in each state? Or could we be looking at a smaller group of "elites" with more exclusive big-business ties in these states vs wealthier states where "regular people" regularly make $75K a year? These questions make it hard for me to conclude that I've learned anything definitive from looking at this data.

Hope this is helpful.

A very nice first post. Great job!

2009-02-24T12:34:39.409-05:00

A very nice first post. Great job!

Welcome Andrew,Insightful post. Useful tool for th...

2009-02-24T12:34:09.369-05:00

Welcome Andrew,

Insightful post. Useful tool for those who are already itching and getting ready for midterm elections.

@ttorack- You're description sounds good , but peo...

2009-02-24T11:37:36.583-05:00

@ttorack- You're description sounds good , but people might be misled by taking this too literally: "The discussion of this post should be why do urbanites care less about wealth as their income increases, not dissecting the red states, which appear to be acting rationally?"

That makes sense only if you define 'rational' as 'short term maximization of nominal income' or perhaps as 'maximization of income relative to the median', since except for the top percentile incomes have grown faster under Democratic administrations. If be 'rational' you mean something about maximizing personal well-being, as objectively measured, or maximizing social capital, etc., then almost all R votes are irrational.
Perhaps the big difference is that in culturally bleak regions a lot of rich people don't see any way to be happier other than to amass more of the sorts of material things which are useful smaller amounts.

Question:In general, is income less correlated wit...

2009-02-24T11:06:32.225-05:00

Question:
In general, is income less correlated with education in poorer states?

I ask because the Republican strategy for the past few cycles seems to have been to market themselves as the everyman, bumbling fools, while decrying any and all educated "fancy-talk" as elitist. Which seems self defeating if their best path to victory to try to get the rich in swing states to vote for them in larger numbers.

Does this play well amongst their wealthy-in-poor-states base, or have they just misjudged and think that higher support in poor states means higher support amongst the poor?

I would also be interested in seeing education lev...

2009-02-24T10:30:08.347-05:00

I would also be interested in seeing education level and voting trends compared. Unless 2008 was unusual, the GOP vote is usually better educated. This seems to be news to those of you committed to self-congratulation and religious bigotry.

Welcome Andrew

2009-02-24T10:20:19.509-05:00

Welcome Andrew

Oh, so now I get it -- GOP stands for Greed Old Pa...

2009-02-24T10:17:36.930-05:00

Oh, so now I get it -- GOP stands for Greed Old Party. That explains alot -- especially why their political strategy is focused on appealing to folks who put Personal Greed above all else.

Congratulations on a pretty phenomenal entrance. I...

2009-02-24T09:48:09.628-05:00

Congratulations on a pretty phenomenal entrance. I can see why Nate was so eager to let us know you're here! Thanks for giving us all a pretty enthralling lesson there, Professor. I'm looking forward to more posts.

So final analysis, an equal amount of rich middle ...

2009-02-24T09:41:50.136-05:00

So final analysis, an equal amount of rich middle income and poor form red purple and blue states shifted from the R column to the D column.

Thats a very broad statement. So you are saying turnout didn't play any part at all. There were equal shifts across all economic levels in red purple and blue states

So basically, all the microtargeting by Obama was wasted. I find that hard to believe.

Welcome to 538, Andrew. One thing that occurred t...

2009-02-24T09:35:56.949-05:00

Welcome to 538, Andrew. One thing that occurred to me when looking at your rich-poor graphs is that the division of rich and poor subsets was based on fixed criteria ($75k and $40k, respectively) while you were graphing against a variable axis of average income. Someone who makes $75k is far richer in a state where the median is $20k than in one where the median is $35k. I wonder if the results would change any if you based your delineations on some percentage of the state average income (rich >= 250% of average income, e.g.).

Thanks.

It is a pleasure to see you debuting here, Dr. Gel...

2009-02-24T09:31:04.749-05:00

It is a pleasure to see you debuting here, Dr. Gelman!

--

@Juris: It does indeed appear as if there was a higher reward for voting republican in the red states.

So may I proudly present to you my rather ingenuous theory:
I am a stranger to the complex US tax system, but travelling the US I learned that red states tend to have much lower tax revenue per capita and I was told that some have even abandoned the state income tax at all - in many ways, the economic and fiscal order in some Red States reminded me of the order in second world countries such as Russia, Chile, Mexico or early 1990s Brazil & Argentina.
Therefore the conclusion:
The more any particular US state resembles a second world country, the more likely its population will vote accordingly (i.e. the elite will vote for the Manchesterist party, the rest will vote for the populist party).
Because the GOP today is both more manchesterist and populist than the Democratic party, Red States will inevitably be Red States until the military seizes political power.

Nice try petekent,Obama is taking over in an econo...

2009-02-24T09:27:30.087-05:00

Nice try petekent,

Obama is taking over in an economic disaster, only Reagan was in a similar situation, but really you have to go back to FDR. Compare Obama's 63 to Reagan's 55.

False equivalence, another of the mainstays in the trolletariat arsenal.

As has been pointed out on several "liberal" blogs, and I am assuming you have read them, Obama's loss in approval rating is strictly in the repub camp, which doesn't really count; but we can all pretend it does, just to make you feel wanted.

By the way, what is your point?

Oh, I get it, Obama really isn't that popular and the Cheney/Bush agenda of the last 8 years is what America really wants, if you can just stomach giving it another chance.

Riiiiiiight.

Correlation does not imply causation, but maybe ex...

2009-02-24T09:14:48.932-05:00

Correlation does not imply causation, but maybe experiments could identify potential causation -- as some commenters have questioned.

Here are some possibilities. I'll abbreviate as "RRR" the phrase "red state rich vote more Republican."

1) Is RRR due to there being more poor people? Perhaps graphing against median incomes might help reveal this, or perhaps a study that shows whether the rich-->Republican curve changes over time with changing median income might help.

2)Is RRR due to (lack of) education? Perhaps referencing against statistics on education level (relative to income, if that data is available) would help clear this up.

3)Is RRR due to religion? Asking "Christian, Jew or Moslem" probably wouldn't be helpful -- too much variance within each group. I once saw statistics (on approval of torture actually) that were related to number of times one attended a worship service per week.

I'm sure that there are lots of other examples. I think it would be quite amusing to learn (for example) that the better off the poor are, the more Democratic the rich vote. That would mean that helping the poor in red states is not just the right thing for Democrats to do, but also the smart thing. (Smart or dumb politically, it's still the right thing to do.)

An interesting analysis of Obama's popularity: It...

2009-02-24T09:12:12.174-05:00

An interesting analysis of Obama's popularity: It seems he is less popular than Carter but more so than Reagan and as pupular as Dubya at this point in his term.

Obama started out with 68 percent job approval, one of the highest initial approval ratings for a new president since Gallup started taking the measurement under newly-elected Dwight Eisenhower in 1953.

Since then, Obama's approval rating has dropped to 63 percent, with 24 percent disapproving and 13 percent having no opinion. The poll was taken Feb. 19-21, among 1,614 adults, with a 3-point error margin.

According to Gallup, this is a "typical" rating for a president at this point, better than Ronald Reagan's 55 percent but not as high as Jimmy Carter's 71 percent. Reagan's low approval rating and Carter's high approval rating at this stage make one wonder about the importance of what a high-rated start really means. So, it's uncertain what being in the middle means. Obama's 63 percent is exactly where President George H.W. Bush was at this stage in 1989. It's also 4 points better than Bill Clinton's was in 1993, and just 1 point better than George W. Bush in 2001.

There's a long way to 2012 . . . .

Im fairly well off, better than most people, but I...

2009-02-24T08:36:46.341-05:00

Im fairly well off, better than most people, but I still wouldnt be caught dead voting Republican, I dont care how many tax cuts you offer me, your not getting my vote.

Welcome Andrew!

2009-02-24T08:12:40.704-05:00

Welcome Andrew!

This post doesn't go far enough in explaining its ...

2009-02-24T08:05:55.517-05:00

This post doesn't go far enough in explaining its conclusions. Ok, we know that wealthy people in the South vote republican while wealthy urbanites on the West Coast vote democratic. Duh. But we need some context. Is it age, education, amount of foreign travel that is making the difference? Obviously a 30 year old Google wiz kid is going to hold more progressive views than an Oklahoma rancher. We need to dig deeper. Numbers alone don't paint the picture -- they only stencil the outline. My suggestion to this site is to add some, dare I say, non-statistical context. Tell me about the religious culture in Oklahoma, or the history of a racial voting-divide between white and black voters. In other words, you're going to have to add a non-stat-head to the lineup.

This is what is missing with this site. For example, the 50 state series was a great idea in theory, but didn't really tell you anything that surprising. I don't want to read about the percentage of burger kings to espresso stands; I want to hear about the state's less-quantifiable social, religious, and cultural currents, with some numbers thrown in for a framework. Numbers are incomplete without context, and I'm sorry to all the "hard" scientists out there, but there's some things that numbers can't tell you.