Tag Archives: election 2016

Garrison Keillor is not a philosophical genius

Garrison Keillor is some guy who’s mildly famous for an NPR radio show that I think I might have heard half an episode of when I was eight. From what little of it I recall, it was approximately the most Midwestern thing in the world: half of the jokes sounded like Mr. Rogers calmly explaining, to a live audience of kindergarteners, Jonathan Edwards’ theology as developed in Sinners in the Hands of an Angry God, and the other half had something to do with polka.

Needless to say, Garrison Keillor does not like Donald Trump. The reason, presumably, is that Garrison Keillor—who, incidentally, I never would have pictured as looking like a mad scientist’s first attempt at merging Boris Johnson into Tricky Dick, but here we are—is from the Midwest, and Donald Trump is from Manhattan. These two cultures are as different as the Finns, who I’m reliably informed speak an average of three words, all of which are “perkele”, in their lives, and the culture of the fellow who came up to me while I was buying a donut one morning last week and excitedly explained for half an hour that I ought to grow dreadlocks.

After spending two whole paragraphs casting stones about Donald Trump’s hair from his glass house somewhere on the shore of one of those lakes, Boris Nixon—sorry, Garrison Keillor—cuts straight to the heart of his criticism of the East Coast—sorry, Republican—nominee: he had, apparently, the wrong response to the small incident in Orlando where some loser from Cousinfuckistan pledged allegiance to the Islamic State and went Adam Lanza Jihad on a gay club.

After the worst mass shooting in American history on Sunday, 50 persons dead in Orlando, the bodies still being carted from the building, the faces of horror-stricken cops and EMTs on TV, the gentleman issued a statement on Twitter thanking his followers for their congratulations, that the tragedy showed that he had been “right” in calling for America to get “tough.” …

His response to the Orlando tragedy is one more clue that this election is different from any other. If Mitt Romney or John McCain had been elected president, you might be disappointed but you wouldn’t fear for the fate of the Republic. This time, the Republican Party is nominating a man who resides in the dark depths. He is a thug and he doesn’t bother to hide it.

Maybe this obsession with appearance is attractive to whatever audience out there it is that’s kept a show about polka on NPR for thirty years, but I, born and raised where maintaining a good appearance begins and ends with trying not to sweat one’s clothes into a dripping formless heap immediately upon stepping outside into the malarial atmosphere of a land that God never intended for man to inhabit, can’t see the point. If we must have politicians, isn’t it better to have politicians who can solve problems (on the rare occasion that the problems are real and can be solved) than politicians who can act all sad about them for the cameras before going off and doing nothing? The president is the Big Celebrity of the West, the one person in the world who can get more airtime than Kim Kardashian; but the president is also the man responsible for appointing people to the parts of government that appoint people who actually do things, and therefore the man responsible (albeit at a distance) for determining whether it’s legal to, for example, have borders—and this seems more important (not that I or any of you have any way to possibly influence the outcome of any election ever) than the former role, especially since the government could just crown Kim Kardashian Queen of America, and then, so as not to violate the human rights of the American people, hand out free cyanide pills to anyone who would want one in such a scenario, i.e. 99% of the population.

But—he also says that Trump is unduly obsessed with appearance.

We had a dozen or so ducktails in my high school class and they were all about looks.

What gives? Is Donald Trump, a “ducktail” according to world-renowned hairstylist Brick Jixon—sorry, Garrison Keillor—all about looks, or is he not about looks enough?

One can only conclude that people who became mildly famous for running radio shows about polka are not guaranteed to be philosophical geniuses.

Oh well, at least he offended some Unitarians once.

 

The mystery model’s predictions

On May 6, I predicted that Clinton would win New Jersey, California, New Mexico, and Kentucky and Sanders would win West Virginia, Oregon, North Dakota, South Dakota, and Montana.

On May 10, Sanders won West Virginia.

On May 17, Clinton won Kentucky and Sanders won Oregon.

The other six primaries were held yesterday. Sanders won North Dakota and Montana. Clinton won New Jersey, California, New Mexico, South Dakota, and New Mexico.

Eight out of nine isn’t bad.

Now that all states have voted, we can complete the chart.

State % black Winner
Mississippi 37.3% Clinton
Louisiana 32.4% Clinton
Georgia 31.4% Clinton
Maryland 30.1% Clinton
South Carolina 28.5% Clinton
Alabama 26.4% Clinton
North Carolina 21.6% Clinton
Delaware 20.1% Clinton
Virginia 19.9% Clinton
Tennessee 16.8% Clinton
Florida 15.9% Clinton
Arkansas 15.8% Clinton
New York 15.2% Clinton
Illinois 14.9% Clinton
New Jersey 14.5% Clinton
Michigan 14.2% Sanders
Ohio 12.0% Clinton
Texas 11.9% Clinton
Missouri 11.5% Clinton
Pennsylvania 10.8% Clinton
Connecticut 10.3% Clinton
Indiana 9.1% Sanders
Nevada 9.0% Clinton
Kentucky 8.2% Clinton
Massachusetts 8.1% Clinton
Oklahoma 8.0% Sanders
Rhode Island 7.5% Sanders
California 6.7% Clinton
Kansas 6.2% Sanders
Wisconsin 6.1% Sanders
Minnesota 4.6% Sanders
Nebraska 4.5% Sanders
Colorado 4.3% Sanders
Alaska 4.3% Sanders
Arizona 4.2% Clinton
Washington 3.7% Sanders
West Virginia 3.6% Sanders
Hawaii 3.1% Sanders
New Mexico 3.0% Clinton
Iowa 2.7% Clinton
Oregon 2.0% Sanders
Wyoming 1.3% Sanders
Utah 1.3% Sanders
New Hampshire 1.2% Sanders
South Dakota 1.1% Clinton
North Dakota 1.1% Sanders
Maine 1.0% Sanders
Idaho 1.0% Sanders
Vermont 0.9% Sanders
Montana 0.7% Sanders

The four states in orange are marked based on this map.

This very simple model—Clinton wins states with plurality Mexican ancestry and states that are over 8% black, and Sanders wins all other states—predicted the outcome of every state but Michigan, Indiana, Iowa, and South Dakota.

In Michigan, Clinton won the Detroit-Flint area and Sanders won most other counties. In Indiana, Clinton won the southern border of the state and the counties containing Indianapolis and Gary and Sanders won most other counties. In South Dakota, Clinton won the eastern half of the state and Sanders won the western half. Only in Iowa does the distribution of Sanders and Clinton victories look random.

There are obvious demographic factors in Michigan. What happened in Indiana and South Dakota?

The mystery model: 2016 is not 2008

Previously: 1 2 3.

I expected that the mystery model would apply to the 2008 primary, but it doesn’t.

Mississippi 37.3% Obama
Louisiana 32.4% Obama
Georgia 31.4% Obama
Maryland 30.1% Obama
South Carolina 28.5% Obama
Alabama 26.4% Obama
North Carolina 21.6% Obama
Delaware 20.1% Obama
Virginia 19.9% Obama
Tennessee 16.8% Clinton
Florida 15.9% N/A
Arkansas 15.8% Clinton
New York 15.2% Clinton
Illinois 14.9% Obama
New Jersey 14.5% Clinton
Michigan 14.2% N/A
Ohio 12.0% Clinton
Texas 11.9% Clinton
Missouri 11.5% Obama
Pennsylvania 10.8% Clinton
Connecticut 10.3% Obama
Indiana 9.1% Clinton
Nevada 9.0% Clinton
Kentucky 8.2% Clinton
Massachusetts 8.1% Clinton
Oklahoma 8.0% Clinton
Rhode Island 7.5% Clinton
California 6.7% Clinton
Kansas 6.2% Obama
Wisconsin 6.1% Obama
Minnesota 4.6% Obama
Nebraska 4.5% Obama
Colorado 4.3% Obama
Alaska 4.3% Obama
Arizona 4.2% Clinton
Washington 3.7% Obama
West Virginia 3.6% Clinton
Hawaii 3.1% Obama
New Mexico 3.0% Clinton
Iowa 2.7% Obama
Oregon 2.0% Obama
Wyoming 1.3% Obama
Utah 1.3% Obama
New Hampshire 1.2% Clinton
South Dakota 1.1% Clinton
North Dakota 1.1% Obama
Maine 1.0% Obama
Idaho 1.0% Obama
Vermont 0.9% Obama
Montana 0.7% Obama

To recap: there’s one statistic that almost perfectly predicts the results of the 2016 Democratic primary. States that come in at above 8.00% (Oklahoma is rounded up) go to Clinton, with two exceptions, Michigan and Indiana, out of 24; states that come in below 8.00% go to Sanders, with two exceptions, Arizona and Iowa, out of 20. In my original post on this model, I called New Jersey, California, New Mexico, and Kentucky for Clinton, and West Virginia, Oregon, South Dakota, North Dakota, and Montana for Sanders. The West Virginia, Kentucky, and Oregon primaries have since been held. Clinton won Kentucky and Sanders won West Virginia and Oregon.

In 2008, of the 23 eligible states above the 8.00% cutoff, Clinton won eleven and Obama won twelve; and of the 25 states below the 8.00% cutoff, Clinton won eight and Obama won 17.

Has any election in recent history other than the 2016 Democratic primary been decided by a single statistic like this? Eyeballing the maps of both parties’ primaries back to 1980, it doesn’t look like it.

The closest precedent that I know of is the general election in 1968.

Does the mystery model generalize?

I recently found a simple statistic that distinguishes with near-perfect accuracy between states that voted for Clinton and states that voted for Sanders. It would be interesting to see whether it generalizes to the Republican primary, or to the 2012 general, though I doubt it will.

State ??? 2016 Dem primary 2016 GOP Primary 2012 General
Mississippi 37.3% Clinton Trump Romney
Louisiana 32.4% Clinton Trump Romney
Georgia 31.4% Clinton Trump Romney
Maryland 30.1% Clinton Trump Obama
South Carolina 28.5% Clinton Trump Romney
Alabama 26.4% Clinton Trump Romney
North Carolina 21.6% Clinton Trump Romney
Delaware 20.1% Clinton Trump Obama
Virginia 19.9% Clinton Trump Obama
Tennessee 16.8% Clinton Trump Romney
Florida 15.9% Clinton Trump Romney
Arkansas 15.8% Clinton Trump Romney
New York 15.2% Clinton Trump Obama
Illinois 14.9% Clinton Trump Obama
New Jersey 14.5% N/A N/A Obama
Michigan 14.2% Sanders Trump Obama
Ohio 12.0% Clinton Kasich Obama
Texas 11.9% Clinton Cruz Romney
Missouri 11.5% Clinton Trump Romney
Pennsylvania 10.8% Clinton Trump Obama
Connecticut 10.3% Clinton Trump Obama
Indiana 9.1% Sanders Trump Romney
Nevada 9.0% Clinton Trump Obama
Kentucky 8.2% N/A Trump Romney
Massachusetts 8.1% Clinton Trump Obama
Oklahoma 8.0% Sanders Cruz Romney
Rhode Island 7.5% Sanders Trump Obama
California 6.7% N/A N/A Obama
Kansas 6.2% Sanders Cruz Romney
Wisconsin 6.1% Sanders Cruz Obama
Minnesota 4.6% Sanders Rubio Obama
Nebraska 4.5% Sanders N/A Romney
Colorado 4.3% Sanders Cruz Obama
Alaska 4.3% Sanders Cruz Romney
Arizona 4.2% Clinton Trump Romney
Washington 3.7% Sanders N/A Obama
West Virginia 3.6% Sanders N/A Romney
Hawaii 3.1% Sanders Trump Obama
New Mexico 3.0% N/A N/A Obama
Iowa 2.7% Clinton Cruz Obama
Oregon 2.0% N/A  N/A Obama
Wyoming 1.3% Sanders Cruz Romney
Utah 1.3% Sanders Cruz Romney
New Hampshire 1.2% Sanders Trump Obama
South Dakota 1.1% N/A N/A Romney
North Dakota 1.1% N/A N/A Romney
Maine 1.0% Sanders Cruz Obama
Idaho 1.0% Sanders Cruz Romney
Vermont 0.9% Sanders Trump Obama
Montana 0.7% N/A N/A Romney

I’ve left out Nebraska and West Virginia from the GOP side, because those races had only one candidate.

The general election is clearly regional; our mystery statistic doesn’t appear to be relevant. The GOP primary is more interesting: the only states Trump lost above our 8.0% cutoff are Ohio (Kasich’s home state) and Texas (Cruz’s), but he won seven states below the cutoff: Rhode Island, Nebraska, Arizona, West Virginia, Hawaii, New Hampshire, and Vermont. Of these, Rhode Island, West Virginia, New Hampshire, and Vermont are East Coast states, and Trump has won every East Coast state except Maine; of the rest, Arizona is an exception for the Democrats as well, for obvious reasons that would be taken into account in a slightly more complex version of this model.

Another way to put this is: above our 8.0% cutoff, 10 Trump states voted for Obama in 2012 and 12 voted for Romney, but below it, every Trump state but Arizona went to Obama. (Below the cutoff, 6 non-Trump states went to Romney and 5 went to Obama.)

So: the mystery model appears to hold for the GOP primary, but there’s a regional effect on top of it.

The Democratic primary, part 2

Data on white % for Sanders is from CNN’s exit polls.

This is profoundly annoying — there’s exit poll data for only five states below the line, one of which is Sanders’s home state and one of which is an unexplained outlier. How is anyone supposed to run any statistical analysis with this garbage? CBS and NYT are no better.

Is there a way to estimate white % for Sanders in states without exit polling? I don’t know statistics.

A naïve computation of Pearson’s coefficient gives = -0.6653, p = 0.00021. Removing the outliers of Texas, Nevada, and Vermont gives = -0.6971, = 0.00022. Removing Vermont alone gives = -0.6643, p = 0.0003.

Perhaps there are regional effects beyond our mystery coefficient. Looking at states with substantial Southern Baptist populations alone (that is: Mississippi, Georgia, South Carolina, Alabama, North Carolina, Virginia, Tennessee, Florida, Arkansas, Texas, Missouri, and Oklahoma), = -0.5729, = 0.051. This is not significant at p < 0.05, but only barely. If we add Maryland, which was founded as a refuge for Catholics and is plurality Catholic today, to make this a measure of the whole South, we get = -0.562, = 0.046.

Before looking at the non-Southern states, remember that we’re missing most of the low end. Pearson’s coefficient is woefully unsuited to this. Southern states don’t have this problem: all but Louisiana have exit poll data, and all but Kentucky (and West Virginia, which would be included by the Southern Baptist metric, and Delaware, which wouldn’t be) have voted. It’s pointless to give Pearson’s coefficient here; I include it in the hope that it motivates someone who does statistics to do a better analysis, or at least find more exit poll data. Non-Southern states: = -0.5124, = 0.061. Non-Southern states minus Texas, Nevada, and Vermont, = -0.1683, = 0.62.

State ??? White % for Sanders Winner
Mississippi 37.3% 31% Clinton
Louisiana 32.4% ??? Clinton
Georgia 31.4% 41% Clinton
Maryland 30.1% 42% Clinton
South Carolina 28.5% 46% Clinton
Alabama 26.4% 38% Clinton
North Carolina 21.6% 52% Clinton
Delaware 20.1% ??? Clinton
Virginia 19.9% 42% Clinton
Tennessee 16.8% 42% Clinton
Florida 15.9% 43% Clinton
Arkansas 15.8% 35% Clinton
New York 15.2% 50% Clinton
Illinois 14.9% 57% Clinton
New Jersey 14.5% N/A N/A
Michigan 14.2% 56% Sanders
Ohio 12.0% 47% Clinton
Texas 11.9% 41% Clinton
Missouri 11.5% 54% Clinton
Pennsylvania 10.8% 47% Clinton
Connecticut 10.3% 50% Clinton
Indiana 9.1% 59% Sanders
Nevada 9.0% 49% Clinton
Kentucky 8.2% N/A N/A
Massachusetts 8.1% 50% Clinton
Oklahoma 8.0% 56% Sanders
Rhode Island 7.5% ??? Sanders
California 6.7% N/A N/A
Kansas 6.2% ??? Sanders
Wisconsin 6.1% 59% Sanders
Minnesota 4.6% ??? Sanders
Nebraska 4.5% ??? Sanders
Colorado 4.3% ??? Sanders
Alaska 4.3% ??? Sanders
Arizona 4.2% ??? Clinton
Washington 3.7% ??? Sanders
West Virginia 3.6% N/A N/A
Hawaii 3.1% ??? Sanders
New Mexico 3.0% N/A N/A
Iowa 2.7% 46% Clinton
Oregon 2.0% N/A N/A
Wyoming 1.3% ??? Sanders
Utah 1.3% ??? Sanders
New Hampshire 1.2% 61% Sanders
South Dakota 1.1% N/A N/A
North Dakota 1.1% N/A N/A
Maine 1.0% ??? Sanders
Idaho 1.0% ??? Sanders
Vermont 0.9% 86% Sanders
Montana 0.7% N/A N/A

The Democratic primary in one chart

There’s one statistic that almost perfectly predicts the results of the Democratic primary. Can you guess what it is?

Unexpected results are highlighted in red. Arizona is expected under a slightly more complicated model. Indiana may be expected; I’ve drawn the line between Clinton and Sanders at 8.00% exactly. (Oklahoma is rounded up from 7.96%.) If Indiana is expected, Massachusetts is unexpected, but Nevada is still expected under the slightly more complicated model. I can’t explain Iowa. I could possibly explain Michigan if I looked at county-level data, but I don’t know how looking at county-level data would affect the other results from the model; it may be better to also take into account Scandinavian ancestry.

I hope Sanders doesn’t drop out before Kentucky. Kentucky could determine where the line should be drawn. If Sanders wins, the line should be at 10% and Massachusetts is unexpected; if Clinton wins, the line should remain where it is. Of course, it may be that the states closest to the line can be determined by factors not easily amenable to raw statistical analysis, like the competence of a candidate’s campaign.

Needless to say, I predict that, if the primary continues long enough, Clinton will win New Jersey, and Sanders will win West Virginia, Oregon, South Dakota, North Dakota, and Montana. I also predict that Clinton will win California and New Mexico. I haven’t looked at any polls except those for Kentucky, which I predict (with less confidence) that Clinton will win.

It would be interesting to look at exit poll data with this statistic in mind.

State ??? Winner
Mississippi 37.3% Clinton
Louisiana 32.4% Clinton
Georgia 31.4% Clinton
Maryland 30.1% Clinton
South Carolina 28.5% Clinton
Alabama 26.38% Clinton
North Carolina 21.6% Clinton
Delaware 20.1% Clinton
Virginia 19.9% Clinton
Tennessee 16.8% Clinton
Florida 15.9% Clinton
Arkansas 15.8% Clinton
New York 15.2% Clinton
Illinois 14.9% Clinton
New Jersey 14.5% N/A
Michigan 14.2% Sanders
Ohio 12.0% Clinton
Texas 11.9% Clinton
Missouri 11.5% Clinton
Pennsylvania 10.8% Clinton
Connecticut 10.3% Clinton
Indiana 9.1% Sanders
Nevada 9.0% Clinton
Kentucky 8.2% N/A
Massachusetts 8.1% Clinton
Oklahoma 8.0% Sanders
Rhode Island 7.5% Sanders
California 6.7% N/A
Kansas 6.2% Sanders
Wisconsin 6.1% Sanders
Minnesota 4.6% Sanders
Nebraska 4.5% Sanders
Colorado 4.3% Sanders
Alaska 4.3% Sanders
Arizona 4.2% Clinton
Washington 3.7% Sanders
West Virginia 3.6% N/A
Hawaii 3.1% Sanders
New Mexico 3.0% N/A
Iowa 2.7% Clinton
Oregon 2.0% N/A
Wyoming 1.3% Sanders
Utah 1.3% Sanders
New Hampshire 1.2% Sanders
South Dakota 1.1% N/A
North Dakota 1.1% N/A
Maine 1.0% Sanders
Idaho 1.0% Sanders
Vermont 0.9% Sanders
Montana 0.7% N/A