Awhile ago I posted some maps based on the Pew pre-election polls to estimate how Obama and McCain did among different income groups, for all voters and for non-Hispanic whites alone. The next day the blogger and political activist Kos posted some criticisms. I disagree with one of Kos's suggestions--he wanted me to rely on exit polls, but I don't actually see them as more reliable than the Pew pre-election polls--but he pointed out some serious problems with my maps. I realized that some fixes were in order. Most importantly:
- My maps would be improved by replacing solid red and blue with continuous shading to distinguish between landslides and narrow margins.
- I needed a more flexible model that would allow the nonlinear pattern of voting and income to vary by state. (In the previous model, I fit a nonlinear pattern (by including a separate logistic regression coefficient for each of the five income categories) but allowed the states to vary only with intercepts and slopes. In the new model, we're letting all five coefficients vary by state.)
During the past couple of months, I've been working on this when I've had a spare hour or two, and now I think we have something reasonable to share. Here it is:
States colored deep red and deep blue indicate clear McCain and Obama wins; pink and light blue represent wins by narrower margins, with a continuous range of shades going to pure white for states estimated at exactly 50/50.
I know that the 538 audience is more interested in results than in technical details, so I've put the detailed state-by-state discussion at my statistical modeling blog (along with a graph showing data and fitted model for each state).
All this work is joint with Yair Ghitza.
P.S. We did not include Alaska and Hawaii on the maps because they were not included in the Pew Research surveys that we analyzed.
48 comments
first post! at any rate this was an interesting analysis as always. I'm a huge fan.
In your post about Huckabee, you say "Obama also, however, won a majority of voters making $200,000 or more, ...". How does this square with your map of State Winners with incomes over $150,000 which appears to show a majority of states, and presumably voters, going for McCain?
Excellent maps! One suggestion--could you post a key? Do the dark colors mean a 20 point difference, or a 30, for example? Thanks!
Man, this post makes me homesick for New England.
One question I have is how all this interacted with age. Most particularly, young people have lower incomes (from what I surmise although I assume there is an income drop at retirement also) and they voted in greater percentages for Obama. Is the income effect confounded to some extent with age? It might be worth checking out.
Those old cotton producing states are gonna be a tough sell for Obama, no matter how you look at it.
Repubs are becoming a small regional party dominated by whites. Should make Kos happy.
That definitely looks more like it, by the 'New Hampshire makes sense now' metric.
@ Bob: My guess is that the answer to your question involves the fact that the states vary widely in population. A solid majority for Obama in California and New York probably cancels out most, if not all, of those pink and red states in between.
I would be curious to see a total of the electoral votes results of each map, I don't see that anywhere.
From eyeballing it seems Obama is close in the 75k overall vote, I can't tell if he wins or loses but it seems close. He starts the cleanup at the 40-75k bracket.
Going to the right side maps, Obama lost even the 20-40k vote? I wouldn't have expected that.
-
Earlier poster:
As far as "total 200k+" votes vs 150k+ map, I would think one being a national total and the other dividing by states would make a difference.
Probably has to do with NY and CA having such a large amount of 200k people vs most red states, if both those statements are in fact simultaneously true.
This appears to be a much-improved analysis. Kudos for being willing to accept criticism and improve.
Can we get a study like this, with all the fancy maps, for "How Black People Voted"?
With regard to the 150 K vs. 200 K issue, I suspect there are many more people in the 150K to 200K range than in the 200K+ range so the former essentially dominate numerically over the latter in terms of statistical effects. It may be similar to the college graduate versus advanced degree comparison, where college graduates as a group tend historically to vote Republican whereas those with a graduate degree are more apt to vote for Democrats. When you combine the two you end up with the college graduate, non-graduate degree crowd numerically overwhelming the graduate school crowd.
A lot of Americans live in Alaska and Hawaii. Perhaps those two states should not have been eliminated from analysis because they don't share a border with any other state, that doesn't seem to be a logical reason for ignoring them (but including, say, Rhode Island and Idaho).
If you ever redo this post again, could you include percent of voters for each map. I could probably find this information, but having it right next to each map would be very interesting (and I wouldn't be wasting my work day looking up statistics...something I spend too much time on as is...). Now if you wanted this to be really sexy, hovering over a state on each map would pop up a thing showing the voter percentage for that map in that state. We could see exactly which maps are the most meaningful for each state.
I wish someone would write a book/blog about Obama's Indiana strategy.
Based on these maps its amazing he won.
North Carolina and to a lesser extent Ohio both have similar map profiles but also have large AA minorities to explain the win but Indian isn't a high minority state. There must have been some very elaborate micro-targeting occuring.
Unrelated anyone else having issues with site loading times.
Interesting to see that the tighty whities in my home state of New Jersey are still quite conservative. NJ seems blue, but it still has a hell of a lot of stalwarts. We are like 45% registered independents, myself included.
In response to Orin's comment, these are the electoral vote totals for each map:
All races and ethnicities:
Over 150k: McCain 353, Obama 185
75k-150k: Obama 284, McCain 254
40k-75k: Obama 349, McCain 189
20k-40k: Obama 462, McCain 76
Under 20k: Obama 523, McCain 15
Non-Hispanic whites only:
Over 150k: McCain 387, Obama 151
75k-150k: McCain 386, Obama 152
40k-75k: McCain 325, Obama 213
20k-40k: McCain 282, Obama 256
Under 20k: Obama 364, McCain 174
A few things to keep in mind:
1. There's no data for Alaska and Hawaii on Andrew's maps. I left Alaska Republican and Hawaii Democratic on each of my maps for simplicity's sake.
2. I don't have great eyesight, and I can't really distinguish Rhode Island, Delaware and D.C on the maps. I left D.C Democratic at all times, shifted Delaware with Maryland, and shifted Rhode Island with Connecticut and Massachusetts.
When you look back at the Carter/Ford election The Republicans took all the west including California, most of the Rust Belt including Illinois and chunks of the Northeast and this was in a losing year. The Democrats took all of the confederacy and some of the northeast.
When the Republicans decided to take the south and appeal to all those white folks in the former confederacy they made a Faustian pact: give us those socially conservative votes, not realising they were giving up their soul.
When will a Republocan win again in New England or the Pacific Coast? Even the mid-west is vulnerable once big cities like Vegas or Denver appear (Phoenix if McCain hadn't been on the ticket). Appalachia, the Deep South and the Prairie Belt just doesn't have enough (white) folks.
Whoa this makes for some extremely interesting analysis.
1) The only two states where whites voted for McCain regardless of income are MS and LA. Obama's problems in WV and Kentucky were overblown.
2)WI was the only state in the Midwest (including IL) where Obama won the rich whites.
3) The progressiveness of the vote proportional to decreasing incomes is shared between whites and Latinos even though Latinos are decidedly more democratic than whites.
4) The voting behavior of whites in NH is particularly peculiar (small statistical sample?)
5) Missouri and then Montana and Georgia were really close and may be the next to fall if Obama's efforts succeed.
If there were maps about men and women voting habits per state that would be extremely interesting. Or is their behavior uniform? What do these maps say about the 2010 election?
Orin: Obama wins the lower 4/5 income brackets on the left side electorally. On the right side, he loses the upper 4/5 electorally. 20-40k (at least 281-257, if RI, DE, and NJ are blue...Due to the loss of NY), loses 40-75k (despite winning CO, especially due to losing MI, PA, and MD).
How is it that if only white people who made over $150,000 voted, California goes blue, but if only the whites from the $75,000-150,000 bracket voted, it goes red? That conclusion looks a little fishy to me.
Interesting. Nice to see that not nearly as many people vote against their financial self-interests as press coverage would indicate (at first glance anyway).
I agree with Dan *, that would be useful.
The data I'd like to see would be, under each map, the percent of all voters that it includes, and the electoral vote count that would result for that group. Wouldn't be too hard, hopefully.
I was struck by the increasing polarization between left and right columns as you move down the page to lower incomes. After thinking about it, though, I suspect that it's just that whites make up a greater portion of the overall sample in the top maps, so there's very little evident polarization. Toward the bottom of the page, the whites are almost certainly a minority, so the polarization shows up.
Also, you should break this apart by age brackets. While poor whites in the South voted for McCain (not surprising), this could be generational-the old Dixiecrats that defected after the Civil Rights Act was signed into law may be dying off in favor of a more liberal South. At least, I hope they are.
Also, if you could get a map that breaks things down to the county/parish level, that would be great. That way we can see what the rural vs urban population breakdown is like.
Lastly, I'd like to see what role gender plays in this, since the Clinton vs Obama tension was a factor in this election.
Adding dimensions to this allows us to see what's going on. Currently, you explore state and income level, and that produces a two dimensional image of the situation, which clarifies a few things but obscures others. Poor whites includes both college students and rednecks- and I have a feeling these two groups probably didn't vote the same way in November. Call it a hunch.
WTF is up with Idaho, Wyoming and Utah? Is the Mormon brainwash that powerful?
Seeing as average income varies greatly throughout the several states, which is to say $200k in NYC means something very different than does $200k in, say, Witchita, KS, perhaps a more useful study would operationalize income as deviation from the mean or median, rather than as an absolute number.
terrydarc said:
"Repubs are becoming a small regional party dominated by whites."
Which makes no sense. What have Dems done for blacks for the past 50 years? They've created welfare states in ghettos from LA to Detroit to New York and everywhere in between. The black inter city in America is a national tragedy, but blacks continue to elect Democrat leadership who continue to lead them down the road of government inuduced poverty with little chance of ever escaping.
Dems have also somehow convinced the union worker that they have their backs. They don't. Look at the state of unions today. Union plants are closing across the country because they have priced themselves out of the marketplace.
Republicans need to hold their ground and wait this administration out as I believe the Obama policies will fail. They will then have their chance to sell their platform and win voters back.
chilcoot, I don't think the reason Alaska and Hawaii are excluded is "because they don't share a border with any other state"; Alaska and Hawaii are not included in most election polling for a number of reasons, so they were probably not included in the data set Andrew was using.
I really wouldn't bet money on Charlie Crist, not after this movie hits the big screen this summer
Grog,
Actually, the social welfare programs Democrats supported from LBJ through Carter did do alot for lower-income folks. It was only after Reagan slashed funding for the job training programs and housing funds included in the Great Society that the program collapsed.
The GOP is to blame for the failure of the programs they slashed in 1980, them and Blue Dogs like Clinton that went along with that crap. The Liberals were doing just fine until you came along. And we'll do even better once you're gone.
The more you try to obstruct progress, the more the electorate sees that. A hungry person is only made hungrier by the promise of food, and angrier when he sees your hand blocking the people who want to feed him. And a whole lot of angry, hungry people are gonna come howling for your blood in November 2010 unless you get the fuck out of the way.
t.dr.k.e.s. said...
A solid majority for Obama in California and New York probably cancels out most, if not all, of those pink and red states in between.
A quick look at the top 101 of the Forbes 500 richest Americans seems to back you up.
Of the top 101 that Forbes lists as richest, 23 live in New York state, 22 live in California or just under 45%. Texas is next at 8, then followed by 5 in Illinois and 4 in Washington state.
By my count, more than 50% of the 101 richest Americans live in just 4 states that all voted for Obama by large majorities - 54 live in NY, CA, IL, and WA.
This is not to say that all the richest Americans voted for Obama, but it does show that there is a great disparity of wealth across the US, with great pockets of wealth in New York and California, with smaller pockets spread in the rest of the US.
Mike in Maryland
My Blogger ID is http://www.blogger.com/profile/02848893412251095965
A tiny bit off topic but relevant to today.
I would really like to see a breakdown of preference for single payer Health care vs Insurance only and blended coverage with a Medicare opt in.
There is a lot of fog out there mostly generated by Insurance companies, but the only refutations are also partisan. The data about what actually works is well documented, but the actual support numbers seem very skewed by whay I am seeing.
Poor Republican whites seem to correlate very strongly with religion. Red is confined to the Mormon belt in the west, and the Bible Belt in the south. New York is less blue than its neighbors-- perhaps left over from its early Mormon roots!
So poor, black people vote for Democrats. Pathetic.
@Andrew: a fine fine improvement on the original -- thank to Yair as well.
I agree with those who propose putting the EV totals at the bottom of each map.
I also agree with those who are interested in some of the interactions (e.g., the "poorest" group in your maps is probably also the youngest).
Question: I'd like to see maps by generation (age cohort), regardless of incomes. Where do the "generations" divide (cf. Nate's discussion of party ID by age)? 18-29 vs. 30-39 vs. 40-49 vs. 50-59 vs. 60+?
Also, do you have enough Hispanics to break out the ages for that group, if only in the aggregate (not by state, though perhaps by region -- CA, SWest, FL, NY-NJ)?
Thanks for continuing to post interesting original analyses.
So, poor, white Southerners vote for Republicans. Pathetic.
@John
Also poor whites and Hispanics, as well as many whites making over $200,000, and as well as most other non-whites in this country at all income levels. And this is "voted for Obama", not "vote for democrats".
Being New Mexican, I found the breakdown fascinating - NM went blue in a big way this year, and that was only possible because of Hispanic votes. Being a minority-majority state, the interests of whites as a collective block doesn't work, except on occasion to hold small offices. Yet having a slim Hispanic majority means that cross-racial coalitions are what's needed. In NM in 2008, the relative moderateness of NM whites (who tilt slightly republican, and only rarely go strong that way) combined with the general center-left bent of catholic voters (which is, by and large, how Hispanics tend to vote), and the state went overwhelmingly blue.
Also, I'm willing to bet NM's population is much poorer than most states, so the bottom two income brackets are, relative to other states, disproportionately strong in elections.
John, As a white woman with a family income of 150K who votes for Democrats, I consider your comment pathetic.
Statler said
“WTF is up with Idaho, Wyoming and Utah? Is the Mormon brainwash that powerful?”
Aside from the Mormon population and one-party rule, I think whiteness accounts for the vote distribution in Idaho, Wyoming, and Utah. According to 2000 census
Idaho 97% white 0.65% black
Wyoming 96% white 1.0% black
Utah 95% white 1.1% black
US 74% white 13% black
Hispanic is included in the white statistics. Excluded, Hispanic runs 7.5/6/8.6%, US 15%. Large (not very) minorities are American Indian and Asian.
My point is that these three states could not have seen a large bloc turnout of motivated black voters intent on striking at the race barrier, because a large bloc of black voters simply doesn‘t exist in these states.
OT Fun Fact: Alaska is 69.3% white. Todd Palin, his children, and his grandchildren are not counted in this figure. They would be counted in the 7% mixed-race category and are thus eligible for government health service. Socialist? You betcha.
As H Jay indicates, the age factor may overwhelm certain categories. My understanding is that Obama/Biden won the youth vote, especially the college vote by a large amount and these folks are mostly low (to no) income.
John said...
So poor, black people vote for Democrats. Pathetic.
You are correct. I mean, why wouldn't they vote for the party of cross-burners, sheet-wearing, angry southern white men with tokens (Steele)???
Pathetic.
John said, "..so poor black people voted for Democrats. Pathetic."
I'd like to know from this person what is pathetic about poor black people voting for one of their own. a black man who apparently cares for poor people. If you are not a black person please tell me what is pathetic about what you do not understand and if you are a black person tell me why it's less pathetic to be ignorant!
great stuff...it amazes me how working class southern whites overwhelmingly vote against their economic interests...and big props to amanda @ 3:34 PM for all the math....
ps..not to diss hi and ak, but their combined populations are less than 1% of the u.s. total...plus hawaii votes blue by a big margin in all groups and alaska does the opposite..no surprises worth teasing out of the date, i suspect...
@Russell - Working class southern whites are NOT overwhelmingly voting against their best interests. The ratio between federal tax receipts and federal tax expenditures has been radically skewed in favor of the so-called "conservative" states for quite some time now, particularly when Republicans are in office. Southern states (except Florida and Texas) overwhelmingly pay less tax and receive more from the federal government than any other region in the country except the Mountain West.
Out of the eighteen states that are net contributors to the federal treasury, only one voted for McCain in the 2008 election (Texas). Every other state that voted for McCain is a net drain on the national treasury.
It is exceptionally clear what working-class southern whites are voting for: they're voting for their gubmint handouts. I guess that's why they call them "Red States."
It appears the only useful statistical analysis on this post is the minority swing in vote. If whites vote McCain wins. If minorities vote then Obama wins. Does not appear income has too much to do with it.
"WTF is up with Idaho, Wyoming and Utah? Is the Mormon brainwash that powerful?"
The Mormon factor only really comes into effect in Utah, as only eastern Idaho (the portion immediately north of Utah) is heavily Mormon, and Mormons barely make up a double-digit percentage in Wyoming.
What's going on in Wyoming and Idaho is that their populations are overwhelmingly rural and white. A lot of the other states with low overall population densities are either more racially diverse, or they have an uneven population distribution internally, with the rural voters partially offset by voters in a moderate-to-large city or two.
What's happened in Utah — which used to be a swing state — is that when the Reagan Coalition started using social issues to win elections, the Utah Republican Party jumped on the bandwagon and actively branded themselves as the superior choice for Mormons. The DNC, meanwhile, abandoned the state to funnel resources into states that were still swing states, which left the Utah Democratic Party without the resources to properly counter the Republican PR machine.
I think the question as "would white people vote for a black president?" is fundamentally flawed.
Try "would white democrats vote for a black democrat" and the answer is, overwhelmingly, yes.
Will white republicans vote for a black republican? If state elections are any measure then the answer is yes.
But you can still be white, racist and republican or democrat while you're pulling the lever for your black candidate. I wouldn't put too much stock in how people vote for candidates in carefully choreographed pageants.
Post a Comment