Thursday, September 11, 2008

Thoughts on educational attainment and voting patterns at the state level

++Addition++Razib notes we're just looking at statewide averages, not specific locales within states. The same analysis could be done on the city or county level as well, as the Census data are there. Half Sigma also responds, pointing out that even considering the high school dropout rate, blue states are still clearly more educated on average than red states are.

---

Based in part on her spending five years to graduate with a journalism degree from a state university and her marrying a man without a degree at all, Half Sigma estimates that Sarah Palin's IQ is in the 110-115 range. He refutes the argument that lacking a college education is a special attribute of the Alaskan frontier:
I just checked the 2000 U.S. Census, and discovered that 24.6% of Alaskans 25 or older have at least a Bachelor’s degree. This is in comparison to 24.4% for the country as a whole.

This contradicts a group of comments left here telling me that in Alaska people don’t go to college because of all the high paying blue collar work, and that I can’t judge Alaska by mainland U.S. standards. These comments turned out to be wrong. Educational attainment in Alaska is nearly identical to the rest of the United States.
It'd be interesting to see a state-level analysis of the average income gap between college graduates and the rest of the population (if anyone is aware of this having been done, please make it known in the comments). As the median income in Alaska is among the highest in the country, this might provide those commenters a little more ammo. The more lucrative it is to go to work right out of high school relative to spending four years in a university, the more enterprising the person choosing to forgo college will be. Conversely, the kids who go to a university for a non-technical degree in these states are probably more likely to be unmotivated and directionless, seeing college as a party without parents extending for four years, than are students who go to school in places where the degree premium is higher.

HS goes on to make this assertion:
Virginia, 29.5%, is the most highly educated red state (voted Republican in last four presidential elections). For this reason, it’s the red state most likely to flip to a blue state.
Virginia may become a Democratic state like Maryland is. But to presume it is vigorously tied to the percentage of the population with a bachelor's or better is tenuous at best. Utah, the most pro-Bush in '04, is the second-best educated red state. It would literally require a 50 state-plus-DC landslide for Utah to become blue for an election cycle. Kansas, the third-best educated red state, would hold out for almost as long as Utah would.

As flushed out a bit below, I'm more inclined to think that Virginia's lack of educational parity (as measured by the percentage of the population that has a high school education but not a bachelor's) is a better explanation as to why the state may lose its red credentials. The five most educationally unequal states, measured in this way, are California, Massachusetts, New York, Virginia, and New Jersey. Virginia is in solidly blue company.

Considering only the bachelor-plus percentage of the population isn't the optimal way to determine the best-educated state, and not because it makes blue states look especially good or the sparsely populated northern central states look especially bad. It presumes the left side of the educational spectrum is identical across states.

That's not the case. As shown in one of this blog's first posts, red states have greater educational (and to a lesser extent economic) parity than blue states do. Red state educational attainment is like female intelligence--its distribution is narrower than that of blue states (and male intelligence). The majority of the population (56%) with at least a high school diploma but not a bachelor's degree are more likely to be found in red states than they are in blue states. The correlation between the percentage of a state's population that falls in this category and Bush's share of the vote in '04 is .65 (p=0).

The percentage of a state's population with a bachelor's-plus does not correlate nearly as strongly with estimated IQ scores based on NAEP data (.46) as the percentage of a state's population with less than a high school education (.72) does. The prolish behavior HS detests might be better indicated by the less than high school percentage than by the bachelor-plus number, depending on where the cutoff is*. Both measures include similar proportions of the population (19.6% of the population over 25 years of age has less than a high school diploma, while 24.4% has a bachelor's-plus).

Taking the bachelor-plus percentage by state, subtracting the less than high school education percentage from it, and multiplying by 100 for ease of viewing, creates a simple educational attainment index by state that more fully accounts for not just the right side of a state's educational distribution, but the entire thing:

1. Colorado -- 19.6
2. Massachusetts -- 18.0
3. New Hampshire -- 16.1
4. Vermont -- 15.8
5. Connecticut -- 15.4
6. Minnesota -- 15.3
7. Maryland -- 15.2
8. Washington -- 14.8
9. Utah -- 13.8
10. Alaska -- 13.0
11. New Jersey -- 11.9
12. Kansas -- 11.8
13. Montana -- 11.6
14. Virginia -- 11.0
15. Hawaii -- 10.8
16. Nebraska -- 10.3
17. Oregon -- 10.2
18. Wyoming -- 9.8
19. Maine -- 8.3
20. Delaware -- 7.6
21. Illinois -- 7.5
21. Wisconsin -- 7.5
23. Iowa -- 7.3
24. New York -- 6.5
25. Idaho -- 6.4
26. South Dakota -- 6.1
27. North Dakota -- 5.9
28. Michigan -- 5.2
29. Arizona -- 4.5
30. Pennsylvania -- 4.3
31. Ohio -- 4.1
32. Rhode Island -- 3.6
33. California -- 3.4
34. Missouri -- 2.9
34. Georgia -- 2.9
36. New Mexico -- 2.4
37. Florida -- 2.2
38. Indiana -- 1.5
39. Oklahoma -- 0.9
40. North Carolina -- 0.6
41. Texas -- (1.1)
41. Nevada -- (1.1)
43. South Carolina -- (3.3)
44. Tennessee -- (4.5)
45. Alabama -- (5.7)
46. Louisiana -- (6.5)
47. Arkansas -- (8.0)
48. Kentucky -- (8.8)
49. West Virginia -- (10.0)
50. Mississippi -- (10.2)

Utah comes out on top among solidly red states, followed closely by Alaska. While the relationship between a state's estimated average IQ and its electoral preferences is weak and well outside statistical significance (r=.12, p=.41), states where people spend more time in academic institutions are moderately more likely to vote Democratic (r=.33, p=.02) than people from other places are.

I suspect people attending college who receive little benefit from it but who are otherwise capable of being productive and prosperous (in the 100-110 IQ range) is a bad thing for Republicans. They'll rack up debt without having obtained the attributes to easily pay it off, while unnecessarily delaying wealth accumulation and family formation, in the process becoming more likely to vote Democratic than they otherwise would have been. The trend toward aptitudinal mediocrity in college probably isn't politically favorable for Republicans, either.

Tangentially, in the discussion thread on HS' post, Razib challenged another commenter:
"College education is a proxy for whites."

did someone look into this? (here or inductivist?) or are you just talking out of your ass?
He then took a quick glance at degree rates and demographics by state and concluded the latter. He's correct. In fact, to the extent that a very marginal relationship exists, it is an inverse one. There is no statistically significant relationship (r=.12, p=.40) between the percentage of a state that is non-Hispanic white and its bachelor-plus percentage.

However, there is a moderately inverse correlation of .39 (p=.005) between the white population percentage and the percentage of the population that has not graduated from high school. Whiter whites in Massachusetts go to college. The wrong kind of whites in West Virginia do not. But both at least graduate from high school. Hispanics and especially blacks are the ones who often don't make it that far.

The data, via Swivel, are here.

* HS' seems to demand upper-middle class company rather than just wanting to avoid the underclass, so that might not be correct. It could be even better to look at the percentage of the population having attained a graduate or doctoral degree for his purposes.

18 comments:

Matt said...

Wow. This is pretty interesting to me. I think it would be extremely interesting if I understood it all. I have been following Half Sigma's blog closely lately.
Re: Utah and education; you might be interested in googling the Mormon Church's Perpetual Education Fund. It would help illustrate why Utahns are fairly well educated.

Razib said...

re: alaska. the returns on labor are pretty high there because of the structural economic conditions (lots of land and resources, not so much labor, etc.). but it needs to normalized to higher cost of living. that won't eliminate the discrepancy, but mitigate somewhat (though i suppose of the cost of living stuff is on discretionary goods).

John S. Bolton said...

I don't know if the data for this is readily available or not, but % graduate degree-holders times % HS dropouts, with the middle ranges more or less exterminated locally, is a leitmotif of a liberal jurisdiction. If DC issues comparable numbers, it will be a far standout on that index. I suggested this when Sailer gave out his challenge to GNXP readers to guess what correlated highest with the 2004 presidentail results. He gave us the clue that DC was included along with the 50 states. The white women's fertility turned out the highest, and this relates strongly to the time that women spend in graduate school, with DC off the charts for having professional women ministering to HS dropouts and their children whose presence drives forth the middle class. Years of female education have been found throughout the world to be the strong predictor of number of children. But how to assign the arrow of causality, by strength and clarity of the correlation signal? At least it should be recognized that female years of education is intimately connected in these high correlations. Not that you haven't, but there could always be another way of drawing out these relations, that might just break through the noise and emit the clearer signal.

agnostic said...

Neat idea -- not just the "smart fraction" from La Griffe du Lion, but the "custodial fraction" that will sap the resources of the smarties.

Looks like the same climate pattern as with White IQ. I'm definitely a convert to the Mountain Time Zone, wacky as some of the local customs are.

What if you defined the index as (SMART - DUMB) * SMART? Where the smarties and dumbies are about as common, the original and new index are about 0. But what about two states where there are many more smarties than dumbies, like Washington vs. Utah -- wouldn't you rather live where the fraction of smarties is larger, all else equal? (Which it probably isn't -- more expensive to live in WA than UT.)

Audacious Epigone said...

Matt,

The fund is a like a federal low interest student loan. Are you pointing it out to illustrate a general emphasis on education, or does it substantially increase the number of young missionaries who go to school after completing their work?

Razib,

Steve attempted a monetary standard of living comparison by state awhile back, but he used the state of MO's numbers from ACCRA to come up with the cost of living, which weren't adjusted for population size of the municipalities comprising the state averages, and so aren't that useful. It'd be nice to have a better comparison done. Other states with a higher median income than Alaska are also expensive places to live (like CT, MD, MA).

A personal anecdote: When I lived in Seattle, my dad's district territory covered Alaska. I remember basic grocery items being 10% or so more expensive than in Redmond, mostly due to logistical costs as I understood it.

JSB,

If DC is counted as its own state, it would top the list as the most educationally unequal in the country, but that is primarily due to so much of the population having a bachelor's degree (39.1%). It's on the high end (22.2%) among high school dropouts, but doesn't 'top' that list--there are several southern states that fare even more poorly in that regard.

DC would also be tops on the "educational attainment" index from the post, increasing the inverse correlation with Bush's share of the vote from .33 to .37.

Half Sigma said...

Being Mormon is a factor, indepentent of anything else, that makes one vote Republican.

Nearly everyone in Utah is Mormon, so that explains why it's a red state.

Audacious Epigone said...

Agnostic,

Honestly, I just have an aversion to the behaviors of the hard underclass. Minimizing that is most important for immediate quality of life. I spend some free time in mixed working class areas and actually enjoy it. But that's being short-sighted on my part, I don't live there, and I'm not raising a family, and anyway, the question was rhetorical so I shouldn't flatter myself in thinking it was directed specifically at me! I suspect most people probably agree with you.

Doing what you suggest only changes the numbers at the margins, and the two indices correlate at .97, but it's worth looking at. I'll also do something similar but include the graduate+ percentage on the educated side, to give more weight to the amount of smarties than to the amount of dumbies, splitting the difference between myself and HS.

agnostic said...

Nearly everyone in Utah is Mormon, so that explains why it's a red state.

"Only" 70% of Utah is Mormon:

Mormonism by state

Fat Knowledge said...

AE,

Good analytical stuff as usual. A couple of thoughts:

1) I was just looking at data on education from the Census on voting and they show 14.5% (32 mil of 220 mil) of over 18 Americans with less than a HS degree (.xls). This is a bit lower than your 19.6% estimate, enough so that it made me wonder why there was a difference. For those with bachelor-plus they have 26.4% vs. 24.4% in your numbers.

2) I was going to ask to see a modified version of your table with extra columns for bachelor-plus and HS dropout, but I was able to find that information in Swivel. Glad that you uploaded it. It is interesting to see that HS dropout rate and bachelor-plus don't appear to be negatively correlated very hightly. I was surprised to see that Alaska has the lowest dropout rate of all states. Any idea why that is?

3) I also wonder how much the chart would differ if you were to do it for state of birth rather than current residency. I was surprised at first that DC had the highest level of college graduates, but then I realized that most of them are imports.

4) Does the median income for Alaskans include their oil revenue check from the state government? For a family of 4 with a $2,000 check + an additional $1,200 that Palin just pushed through that comes to $12,800, which would lower that median income down a bit.

Steve Setzer said...

@ae
I don't know what Matt was looking at, but I would take the fund as illustrating a general emphasis on education. From the lds.org website:
"The Perpetual Education Fund is currently available for Latin America, Caribbean, Philippines and Africa Southeast."
US Mormons can get Stafford loans, so they don't need the PEF, and the PEF won't affect state-by-state educational attainment.

Audacious Epigone said...

FK,

That difference is curious. It's not a math error on my part, either--I'm taking that figure 19.6% figure directly from the Census. A few possible reasons for the discrepancy:

1) The data I use are from 2000, the one you provided is from 2006. The 2000 Census is much more thorough--it's a physical count of everyone, whereas the interim years are just estimates based on representative surveys.

2) My numbers use the entire population, whereas yours includes citizens only. In absolute numbers, we're talking about 11 million more HS dropouts from my data. Well, 9 million of the 27 million foreign-born over 25 are high school dropouts, which gives us (forcing different years and minimum age data a little, I know) 220+27 for a total population of 247 million and 32+9 for a total dropout population of 41 million. That gets us 16.6% with a HS or less. This would do a better job accounting for the bachelor+ discrepancy of 2% of the total population.

3) Graduation rates have been increasing. The 2000 data I used are for 25+, yours for 18+. According to this from the Census, the rate has increased nearly 5% since 2000, which would show up disproportionately strong in the 18-24 age range.

Re: the modest inverse relationship between less than high school and bachelor+, that's what makes using bachelor's alone seem lacking to me. If the two correlated strongly, you'd just be measuring the same thing twice by looking at each, but the correlation between them is only .43, meaning less than 20% of the bachelor+ percentage is 'explained' by the less than high school percentage, and viceversa.

Looking at the numbers by place of birth instead of settled residence would be neat. I suspect it would make flyover country look more advanced educationally. Not sure how to go about doing it, though.

I assume income figures include the petroleum dividend. It is taxable income (although Alaska has no income tax, it still gets caught federally). In terms of percentage, that alone reduces the gap between college grads and the rest, albeit just a little.

Steve,

Thanks. I just wanted to be sure I wasn't missing something.

Audacious Epigone said...

JSB,

I was off a bit when I said DC would've topped the list using the index employed for the other states. It would've been third, with 16.9, behind Colorado and Massachusetts. Sorry about that.

Clark Goble said...

While the Mormon emphasis on education probably counts a bit looking at how the surrounding states do it may be more of a western thing. The western states that don't fit into the trend (Arizona, New Mexico) also have other issues such as a large proportion of Latinos who sometimes aren't well served by the school systems and communities there.

Fat Knowledge said...

Hmm, strange that the numbers should be that different given they are both from the same source.

Actually, the numbers I looked at were for total population rather than total citizens, and it looks like there are 8 mil of 19.5 foreign born that have less than a HS degree for my numbers.

So the only differences are 2000 vs 2006 and age 25 vs 18. That still seems like a big difference to be explained by just those 2 factors, but I guess it must explain it.

Looking at the numbers by place of birth instead of settled residence would be neat. I suspect it would make flyover country look more advanced educationally. Not sure how to go about doing it, though.

Yeah, I have no idea if anyone has collected such information. More than the flyover country, I think it would impact the rust belt states. In fact, I think that is the rust belt states biggest issue is that their smart people move away because their are no opportunities. So the only people left in those states are the ones who want to go back to their good old days of the 50-60s. Pisses me off that they are swing states once again so they get undue political influence and they want dumb policies that are unlikely to stop the manufacturing jobs from leaving anyway.

Tangentially, I would really like to know where the birth place of all that died on 9/11 and compare that with the birth place of all that have died in the Iraq war.

I have thought one of the big political ironies of the Bush era is that if you believe that fighting in Iraq has saved the US from another terrorist attack, and that the terrorists were likely to only hit big cities in blue states like NY, DC, or LA, what Bush has really done is allow Red state military personnel sacrifice their lives to save Blue state citizens. Funny then how the Blue staters hate Bush and the Red staters love him. But, then I wondered what % of the people that would die in a terrorist attack in a big city grew up in a small town in a Red state. Anyways, this really has nothing to do with your post, but your comment triggered the thoughts. :)

Audacious Epigone said...

FK,

Hehe, interesting chain of thought. I don't have an answer for you, sorry.

Audacious Epigone said...

FK,

Also, re: the discrepancies, it shouldn't have any impact on the ordering of the index, since the figures used are all by state for the same year.

Stopped Clock said...

Colorado is likely to go blue this year:

http://www.realclearpolitics.com/epolls/2008/president/co/colorado_mccain_vs_obama-546.html

Steve Sailer said...

One thing to keep in mind is that younger populations tend to have higher college graduation rates -- Utah has younger white people than North Dakota, so it has higher college graduate percentages. I don't know what a more apples to apples comparison would show.