Friday, November 30, 2018

Senator, heal thyself

My last post was suggested by a column about Senator Ben Sasse's new book  “Them: Why We Hate Each Other — and How to Heal.”  The argument of the book is apparently (I haven't read it) that America is suffering from a loss of community, and that people try to fill their need for connection by joining political "tribes."  To quote the publisher's description "contrary to conventional wisdom, our crisis isn't really about politics. It's that we're so lonely we can't see straight—and it bubbles out as anger.  Local communities are collapsing. . . . As traditional tribes of place evaporate, we rally against common enemies so we can feel part of a team."  In my last post, I pointed out that there is no evidence that loneliness is increasing.   In this post I will consider the other part of the argument:  that the lack of community increases political "tribalism."  This is not a new claim:  "mass society theory," which was popular in the 1950s, said essentially the same thing.

For the strength of community, I use the index of "social capital" compiled under the leadership of Sasse's colleague Mike Lee.  It is measured at the state level--Lee's home state of Utah is highest, Sasse's state of Nebraska is eighth.  I defined people who "rally against common enemies," as those who rank the Democratic or Republican party at 0-4 in the 2016 American National Election Studies "feeling thermometer."  The hypothesis is that people in places with less "social capital" will be more likely to think that one (or both) of the parties is completely bad.  This produces four groups:  neither party is completely bad, just the Republicans are, just the Democrats are, and both are.  This can be modeled with three logistic regressions:  both vs. none, Democrats vs. none, and Republicans vs. none.  I control for the overall political tendency of the state by the share of the vote won by Mitt Romney in 2012.  The results:

                                   Romney                Social Capital
Both                                -1.51                 -.232
                                       (1.18)                 (.142)

Democratic                      2.99                 -.158
                                       (0.55)                (.058)

Republican                     -1.59                  -0.73
                                       (0.52)                 (.058)

The estimated effect of social capital is statistically significant only for ratings of the Democratic party, but it's negative for all ratings, and I'm pretty sure that the differences among them would not be statistically significant.  That is, it seems that people in states with less "social capital" are more likely to really detest political opponents.  I haven't tried to control for anything else, but the results are intriguing. 

However, even if the relationship is real, it's unlikely that it explains the rise of political "tribalism."  Very low ratings of the parties stayed about the same or rose slowly between the 1970s and about 2000, and then rose sharply in the 21st century.  There's not much evidence on changes in loneliness, but there is some data on satisfaction with your local community, and it hasn't shown any trend since the 1990s (and is higher than in the 1970s). 

So maybe the conventional wisdom is right, and political polarization is about politics.  People dislike political conflict, and seem to especially dislike conflict that goes on without a clear resolution.  There has been a lot of that in recent years.  Why?  Ben Sasse helps to provide the answer, not in his book, but in his statement when announcing his candidacy for Senate in 2014.  A few selections: 

"This glorious idea of freedom and of the creative self-sufficiency of local communities and extended families is under attack – both by intentional opponents and from our lazy national neglect in recent decades.

Our current President was re-elected in a campaign that had as its centerpiece a vision of cradle-to-grave dependency. He has been selling a fundamentally different vision of America's history, and a redefined relationship between government and the people. As Obama’s vision of government wraps its tentacles around more and more aspects of American life, initiative is discouraged, achievement is disparaged, and we grow closer to a permanent dependency class.

Nowhere has this been stated this more clearly than in President Obama’s 'You didn’t build that' speech. This speech angered us, but even more, it should sadden us. . . . And the greatest single insinuation of government into every aspect of our life is his signature initiative, Obamacare. If it lives, America as we know it will die. If the idea of America is to live, it must be stopped."


"The Obamacare worldview holds that Government can successfully take over the largest sector of the economy and orchestrating it better with its allegedly 'all-knowing' central planners.

This worldview says that false promises can somehow become true if only we had even more government; or that they aren’t simply lies because they are founded on good intentions."

So why has political polarization increased?  A major reason is that the Republican party has come to be dominated by dogmatists who imagine that ordinary policy disagreements are issues of high principle on which no compromise is possible.  That was what Thomas Mann and Norman Ornstein argued in 2012, and the history since that time has supported their case. 

Saturday, November 24, 2018

So alone?

About six months ago, I had a post about the claim that there has been a dramatic increase in loneliness in recent years.  I concluded that (1) there was no trend in feelings of loneliness among people in general between 1963 and 2001, and no useful data on changes since 2001 (2) there was some evidence that feelings of loneliness had declined among high school and college students since the 1970s.   But yesterday Arthur Brooks had a piece in the NY Times which began: "America is suffering an epidemic of loneliness."  He cited "a recent large-scale survey from the health care provider Cigna [which] shows that loneliness is worse in each successive generation."   That sounds pretty definitive, doesn't it?  But the survey he referred to is a cross-section--it was conducted only once, in February 2018.  Therefore, it can tell us about differences among people today, but it can't tell us anything about historical changes.  The report compares generations, and finds loneliness is higher in more recent ones, which could be a sign that loneliness is on the rise.  However, it could be an age effect, not a lasting difference between generations.   That is, people might get less lonely as they get older.  That seems plausible--older people are more likely to be married, have children and eventually grandchildren, and more likely to be settled in their jobs and communities.    So there is still no evidence that loneliness is on the rise. 

Brooks went on to say that loneliness is responsible for political rancor, an idea that he attributes to Senator Ben Sasse.  In order to "fill the hole of belonging in their lives," lonely people "turn to angry politics."  This doesn't mean that people who engage in "angry politics" will be lonelier than other people: someone who gets involved with the "polarized tribes forming on the left and the right in America," may make friends and find a sense of purpose.  Rather, the idea is that if they had been part of a "healthy" community, they would not have become involved.  It's hard to evaluate this idea, because it's hard to define, let alone measure, a healthy community.  However, I will make an attempt in my next post. 

Sunday, November 18, 2018

Odds and ends

The pollster Stanley Greenberg had a piece in the New York Times today making pretty much the same point I made in my last post:  that in the 2018 the Democrats didn't just gain among special groups (like college-educated suburban women), they gained  among almost all kinds of people.  At one point, he said that their biggest gains were in rural areas.  That reminded me, that I had wanted to include something on that, but the CNN exit polls I used didn't seem to include it.  I checked again, and discovered I just hadn't looked far enough.  Here is the Democratic share of the vote in 2016 and 2018:
                     2016       2018
Cities            61%        65%       +4%
Suburbs        45%        49%       +4%
Rural            35%        42%       +7%

Greenberg referred to another exit poll that showed an even bigger swing to the Democrats in rural areas.  So it seems that Democratic gains were at least as large in rural areas as in cities and suburbs, and maybe larger. 

That wasn't much of a post, so here is a bonus.  I had a post a few months ago about the decline of educational differences on a question about whether the government should reduce income differences between the rich and poor.  The GSS has a similar question, "Some people think that the

government in Washington is trying to do too many things that should be left to individuals and private business; they are at point 5 on this card. Others disagree and think that the government should do even more to solve our country's problems; they are at point 1. Where would you place yourself on this scale?"  Here is the gap between people with a college degree and those without. 

Positive numbers mean college graduates are more towards the "doing too many things" end.  The gap has pretty steadily become smaller, and in 2016 was near zero. 

Thursday, November 15, 2018

Coalitions and votes

Both academics and journalists often talk about elections in terms of "coalitions." A coalition is when two groups agree to work together.  In a multiparty system, you often have coalitions, where the representatives from two parties agree to vote the same way.  A party can withdraw from the coalition and switch to voting with someone else.  So the language of coalitions suggests discrete groups that can suddenly switch sides.  That leads to several tendencies that are usually mistakes:
1.  The idea that only certain groups shift from one election to the next, or at least that they move a lot more than all the others.  Usually most groups move in the same direction:  e. g., if the Democrats do 5% better than last election among men, they will also do about 5% better among women.  There are cases when two groups move differently, like people with and without college degrees  in the 2016 election, but they are unusual.
2.  A focus on groups defined by a combination of characteristics.  For example, the New York Times had a story entitled "As suburban women turn to Democrats, many suburban men stand with Trump."  It started talking about suburban men and women, but then turned to talking about college-educated men and women.  But either way the idea was that that it wasn't just men and women, but something more complex. 
3,  The idea that there is a "winning coalition"--e. g., some of Trump's supporters say that because he won with a "working-class coalition," they don't have to worry about suburban college graduates who used to vote Republican.  You can win or lose with any kind of "coalition"--what matters is how many votes you get, not which groups your votes come from.  There may be a germ of truth in the idea that some "coalitions" are better than others--people often tend to think of politics in group terms and may support or oppose a party because they associate it with particular groups--but that's a more subtle point. 

This is an introduction to a comparison that is not all that interesting:  how Democrats did among different groups in the 2016 and 2018 vote for House of Representatives, according to CNN exit polls. 

                              2016               2018         Change
Men                        43%                49%          +6%
Women                   54%                59%          +5%

White                     38%               44%           +6%
Black                      88%               90%           +2%
Latino/a                  67%               69%           +2%
Asian-Am.              65%               77%           +12%
Other                       56%               54%           -2%

White Men              33%               39%            +6%
White Women         43%               49%            +6%

W College Women 49%                59%           +10%
W non-C Women    35%                42%           +7%
W Coll Men            38%                47%           +9%
W non-C Men         27%                32%           +5

White Coll.             44%                53%             +9%
White non-C           31%                37%             +6
Non-W C                71%                77%             +6%
Non-W non-C         77%                76%             -1%

under $30,000         56%                63%             +7%
$30-49,999              55%                57%             +2%
$50-99,999              47%                52%             +5%
$100-199,999          46%                47%             +1%
$200,000+               44%                47%             +3%

Married Men           37%                48%             +11%
Married Women      48%                54%             +6%
unmarried M           49%                54%             +5%
unmarried W          63%                66%             +3%

Veterans                  36%                41%             +5%
non-Vets                  51%                56%             +5%

The Democrats did better among almost all groups, and the gains are usually similar (despite sampling error).  Note that the gains are about the same among men and women--it's easy to think of reasons why women might have swung more strongly against the Republicans than men, but apparently it didn't happen.  There was a somewhat stronger shift among college graduates than non-graduates--that is, the education gap grew.   There are a few other cases where the movements were different--married and unmarried people (maybe*), Asian-Americans and Latinos.  But basically, the story of the election was that the Democrats gained among all sorts of people. 

*The numbers don't fit:  a shift of +5 among unmarried men, +11 among married men, but +6 among men as a whole.  About 60% of men were married in both years.

Monday, November 12, 2018

The empty quarter

Last week Ross Douthat had a column proposing that "that a real center-right majority could be built on economic populism and an approach to national identity that rejects both wokeness and white nationalism."   He asks us to "imagine that [Trump] followed through on Steve Bannon’s boasts about a big infrastructure bill instead of trying for Obamacare repeal . . . tilted his tax cut more toward middle-class families . . . bullying Silicon Valley into inshoring factory jobs . . . made lower Medicare drug prices a signature issue rather than a last-minute pre-election gambit."  His terminology is misleading:  what he's talking about would not be "center-right" but left of center on economics and right of center on "social issues."  However, he raises an interesting question.   In the general public, opinions on economic and social issues are pretty much uncorrelated.  That is, you have a lot of people who are to the left on economics and to the right on social issues, or to the left on social issues and right on economics.  But electoral politics is dominated by a right/right to left/left dimension:  there are no prominent politicians who offer the combination that Douthat wants (and many ordinary people seem to want).  Why do things stay that way?  A left/right vs. right/left alignment seems just as reasonable in principle--maybe even more reasonable (because fewer people would be "cross-pressured" by education and income).

It seems that there are two possible answers.  I offered one last year--that once a pattern is established, it's safer to play to your "base" rather than trying to change things.  If Trump tried to do the things that Douthat talks about, a lot of congressional Republicans would revolt.  Maybe he would win some Democrats over, or the proposals would be so popular that reluctant Republicans would have to go along, but maybe he would fail, lose the trust of Republicans, and look weak.  That's probably more likely, since Democrats wouldn't want to help him.  So why risk it?

 Another possibility is that the left/left vs. right/right dimension has some psychological basis--that the combination of supporting both legal abortion and redistribution toward the working class (for example) is more natural than the combination of opposing legal abortion and supporting redistribution to the working class.  There have been a number of arguments along these lines, all saying that the line of division is something like sympathy for marginalized people vs. support for rules and authority (an example from a few days ago).  It might seem like these are refuted by the lack of correlation between economic and social attitudes.  But you can say that it would be present only among people who think about politics.  The problem is that people who pay attention to politics are likely to pick up the conventions:  if I'm a liberal, I'm supposed to favor legal abortion.  So in order to evaluate the idea that there is a "natural" pattern, you need to consider attitudes that aren't part of normal political debate.

The General Social Survey has a couple of possibilities.  One is "What is your opinion about a married person having sexual  relations with someone other than the marriage partner? [Always wrong, almost always, only sometimes, or not wrong at all].  This isn't a political issue in the sense of something either party proposes legislation on, and there are no obvious partisan or ideological differences in the behavior of prominent politicians.  Another one is closer to a political issue, but still not part of standard partisan debate:  "Do you strongly agree, agree, disagree, or strongly
disagree that it is sometimes necessary to discipline a child with a good, hard, spanking?"

The correlations of opinions about extramarital sex with opinions on some standard political issues, comparing people with and without a college degree (since the GSS doesn't have a good measure of political knowledge or interest).

                                                                                                       Not Grad      Coll Grad
Redistribute from rich to poor (EQWLTH)                                      .003                    .108
Gov't help people with medical costs (HELPSICK)                        .076                    .140
Allow gay man teach in college (COLHOMO)                               .125                    .125
Prayer in schools (PRAYER)                                                           .094                    .192
Abortion if doesn't want more children (ABNOMORE)                 .200                    .297

It has substantially higher correlations among all items (except COLHOMO, where it's the same) among college graduates

The correlations of opinions about spanking with opinions on the same issues, comparing people with and without a college degree:
                                                                                                      Not Grad        Coll Grad
Redistribute from rich to poor (EQWLTH)                                     -.012                   .167
Gov't help people with medical costs (HELPSICK)                        .054                    .189
Allow gay man teach in college (COLHOMO)                               .100                    .120
Prayer in schools (PRAYER)                                                           .139                    .234
Abortion if doesn't want more children (ABNOMORE)                 .050                    .202

For all questions, it's higher among college graduates. 

Note that redistribution goes from essentially zero to comparable to the other questions in both cases.  My interpretation is that there is something to the idea of a psychological affinity among positions on economic and social issues.  Of course, a left/right or right/left combination is not logically inconsistent, but it's less likely to appeal to people who are interested in politics--that is, the people who might be leaders or advocates for a position.  

Saturday, November 10, 2018

Which working class?

I am going to have a post on a column by Ross Douthat about the possibility of "a real center-right majority" based on the working class.  However, it's a little complicated, so here's some background.

Just before the 2010 election, Barack Obama's approval rating had fallen to about 45%, only a little higher than Donald Trump's was just before the 2018 election.  Obama's rating was higher among college graduates (48%) than among non-graduates (44%).  But it was also higher among people with low incomes than among people with high incomes (50% among people who got less than $2,000 a month, 45% among 2-5,000, 45% among 5,000-7,500, and 42% among over 7,500).   Donald Trump got higher approval among people without college degrees (45% vs. 38% among college graduates), but also among people with higher incomes (31%, 40%, 50%, 48% for the income categories given above).

So who was more popular with the "working class"?   It depends on how you define it.  (Most sociologists would prefer to define class by some combination of occupation and self-employed vs. works for others, but few surveys ask those questions now).  Or you could bypass the question of defining the working class and just say that Trump is more popular among people with high incomes and low education, and Obama was the other way round.

For Obama, the gap between rating among more and less educated people grew over the course of his presidency:  when he started it was only a couple of percent, and when he ended it was 64% to 57%.  The income gap may have a little.  For Trump, the education gap has stayed about the same, and the income gap may have grown.   But in both cases, they were pretty stable--Obama's approval rating rose and fell, but when it did, it was by about the same amount among all groups.  (Trump's approval rating has been very steady among all groups). 

[Data from the Gallup Presidential Job Approval Center]

Tuesday, November 6, 2018

The forgotten men and women

There's been a lot of discussion of people who shifted from voting from Obama in 2012 to Trump in 2016 and some discussion of people who switched from Romney to Clinton.  But there's another group who hasn't gotten much attention--those who switched from Romney or Obama to a minor candidate or write-in.  The share of the vote in 2012 and 2016:

                           2012           2016
Republican         47.2%         45.9%
Democrat           51.0%         47.2%

Libertarian         1.0%           3.3%
Green                 0,4%           1.1%
McMullin           0                 0.5%
Write-in             0.2%           0.8%
Others                0.3%           0.3%

I expect that most of the gain for the Libertarian (Gary Johnson in both years), Evan McMullin, and the write-ins came from people who would normally vote Republican.  Their combined total went from 1.2% in 2012 to 4.6% in 2016, for a gain of 3.4%.   It seems likely that those people are committed voters--otherwise they wouldn't have gone to the polls or would have left the Presidential race blank.  So what they do in 2018 could make a significant difference.  The question is whether they are mostly people who simply would not vote for a Democrat, or people who might vote for a Democrat but not Hillary Clinton, and I haven't seen anything that sheds any light on that.  However, my impression from other data is that negative partisanship has become a strong force.   That suggests that some Republicans who now say that they will vote Democratic this time will switch back at the last minute, and the Democratic gain will be on the smaller side of what has been predicted, giving the Democrats a majority of maybe 225-210 in the House.  But given that Trump has been so prominent in the election, that he's focused exclusively on his "base," and that the base was not that big to start with (a lower share of the vote than Mitt Romney), I'll put the probability of a Democratic majority at over 90%. 

Saturday, November 3, 2018

Good news or bad news?

Thomas Edsall has a piece in the New York Times in which he discusses research on the relationship between values and political views.  By "values," I don't mean things that get called "values issues," but general views of life as measured by questions that don't have any obvious connection to the political issues of the day.   Edsall talks about two measurements of values, but I will focus on the analysis by Marc Hetherington and Jonathan Weiler, who compare people with "fixed" and "fluid" worldviews.  (Some people have used "authoritarian" for the same thing, but I think that "fixed" and "fluid" are better terms).  They are measured by four questions about which qualities it was more important for a child to have:  "Independence or respect for elders", "curiosity or good manners", "obedience or self-reliance", and "being considerate or well-behaved".  When the questions were first asked in the American National Election Study (1992), answers had little association with party identification, but in 2016, they had a strong association. 

What I found most interesting was that Edsall assumed that this split by values was bad news for the Democrats.  He says the work raises "a warning flag for the Democratic Party — that the rightward movement in contemporary politics is neither evanescent nor trivial."   But there is a movement towards "fluid" values.  I show the percent who chose the "fluid" side on each question in 2000 and 2016:

                               2000             2016
Indep                     22.5%           26.3%
Curious                 37.6%           35.9%
Self                       40.4%           52.6%
Considerate          63.4%           66.9%

This movement is almost certain to continue, because people with more education are more likely to choose the "fluid" side and average educational levels will continue to increase because of generational replacement.  So like a lot of other splits (ethnicity, urban/rural, education), the tide is running in favor of the Democrats.  But what about now?  Over the four issues, the average is about 40% for the "fluid" answers and 60% for the "fixed."*  Isn't it a problem to be the choice of the minority group rather than the majority group?  Not really.  Suppose we start out with a small group (20%) and a large group (80%).  At first, both are evenly split between parties A and B.  Then something happens and the small group aligns with party A while the large aligns with party B.  Here is one possible result:

                     A                          B
Small          14 (70%)               6   (30%)
Large          28 (35%)             52   (65%)
Total           42                        58

The change has been bad for party A.  Here is another: 

                    A                          B
Small          16 (80%)               4   (20%)
Large          36 (45%)             44   (55%)
Total           52                        48

The change has been good for party A.   It's possible to get a majority even when you're aligned with the minority group and the other side is aligned with the majority group.  In fact, when you think about it, that happens all the time. 

So the growing split by values helps to explain which people voted for Trump, but doesn't explain (at least not directly) why Trump did as well as he did (or as poorly, depending on how you look at it).  It also doesn't explain why he appeared in 2016, rather than at some other time.  In fact, it suggests that the potential support for someone like him is lower than it used to be. 

So why did we get Trump in 2016, rather than some other time?  I proposed an explanation in April 2016, and I still think it's about right. 

*The relative sizes may depend on the exact way the questions are asked.  For example, if the first question had given the "fixed" alternative as "respect for parents," or "respect for authority," it would have measured the same general attitude, but the numbers choosing those options might be different.  But I'll assume that "fluid" values are a minority, since the general point applies even when one group is definitely smaller than the other.