Friday, April 27, 2018

Them vs. us

In 2013, the ISSP survey on National Identity asked people how they felt about the statement:  "International organizations are taking away too much power from the [name of nation] government" (1-5, higher numbers indicating disagreement).  I regressed responses on education separately for each nation and computed predicted values for lowest and highest levels of education (no formal schooling and graduate degree).  A plot of the predicted values for the highest level vs. GDP shows a strong relationship:  the higher the GDP, the more likely people are to disagree:

 For the opinions of people at the lowest educational level vs. GDP, there is no clear relationship:  the correlation is negative (about -.3) but not statistically significant.  Controlling for GDP, there is a national effect that applies to both levels.  General disagreement is highest in the Philippines, Iceland, United States, Germany, and Japan; general agreement is highest in Spain, Portugal, Latvia, the Czech Republic, India, and Denmark.  Although it's not possible to measure exactly, the rankings seem to have some connection to the actual power of international organizations over the nation in question.  For example, economic policy in Spain and Portugal was strongly constrained by the EU after the recession of 2008. 

However, the main conclusion is that there is a common pattern in which educated people are more likely to disagree (that is the case in all 33 nations), but the size of the gap increases with GDP (a correlation of over 0.7). 

Note:  I used GDP from 1995, on the grounds that general national vs. international orientation is likely to be set early in life, and 1995 is roughly the youth of the average voter. 

Note 2:  In an article published in 2003, I found that for many opinions, education had the same direction of effect in almost all nations, but that the magnitude grew with GDP.  This is another case of that kind of pattern. 

Sunday, April 22, 2018

Confidence in Government, 1952-2016

This was the post I was going to do last week--an update of a overall measure of confidence in government that I wrote about in 2013.  I found that in addition to the six variables I used last time, there was one that asks "How much do you feel that having elections makes the government pay attention to what the people think, a good deal, some or not much?"  I included that and added the 2016 data, giving this general index:

I don't think many people will be surprised that it reached its lowest level in 2016; the intriguing thing is that it was almost as low in 1994 and then rebounded.  I wanted to update this partly to see how well it tracked the growth negative feelings about the parties that I wrote about last month.  Not that closely:  there is a correlation, but that's almost entirely because of the influence of 2012 and 2016.  

While looking back at the 2013 post, I see that it was inspired by claims that Mitt Romney's loss in 2012 was because of low turnout among "downscale, Northern, rural whites."  Some people have suggested that in 2016, Trump appealed to those voters and they turned out in large numbers.  According to the ANES, among non-Hispanic whites who said they voted in 2012, Trump got 53.4% and Clinton got 39.9%; among non-Hispanic whites who said they didn't vote in 2012, Trump got 53.5% and Clinton got 37.4%.  That difference is not statistically significant, or even close.  Given the small numbers of reported non-voters, there's a lot of uncertainty, but the data doesn't suggest that Trump had a particularly strong appeal to "alienated" white voters.  There was one interesting difference:  Clinton did substantially worse among blacks who didn't vote in 2012 than among those who did (68.5% vs. 92.5%); Trump, Johnson, and Stein all did substantially better.  Although the numbers are very small, the difference is statistically significant.  

Wednesday, April 18, 2018

All the lonely people

A couple of days ago, David Brooks had a column in which he wrote "In the 1980s, 20 percent of Americans said they were often lonely. Now it’s 40 percent."    I've seen research on changes in the number and type of ties among people, but I didn't know of anything on feelings of loneliness, so I tried to investigate further.  Brooks didn't provide a link to his source, but a Google search showed other articles making the same claim.  The source of the 40% figure seems to be a survey of people aged 45 and over sponsored by the AARP in 2010.  However, the report of that survey didn't say anything about changes in loneliness.

There is a question that has been asked in a number of surveys asking people if they had felt "very lonely or remote from other people" in the past few weeks.  The percent saying they had:

Nov 1963    28%
June 1965   26%
Jan 1981     17%
May 1990   19%
Sept 2001   26%
Dec 2001    24%

That doesn't look like any kind of trend.  The numbers in the 1981 and 1990 are lower, but they were in surveys taken by Gallup, and the others were by NORC, so that may be a factor.  Unfortunately, the question hasn't been asked since 2001.

I searched Google scholar for papers about trends in loneliness, and found one from 2014 entitled "Declining Loneliness Over Time:  Evidence From American Colleges and High Schools" .  It was based on surveys at various colleges and universities and on the Monitoring the Future Survey, a representative survey of high school students that has been conducted since the 1970.  It mentioned that other literature claimed that loneliness had increased, but I checked the sources they cited and they didn't provide any evidence--they just said it had, or cited research that wasn't really relevant.

It's possible that I missed something, but I doubt that there is any actual evidence that feelings of loneliness have doubled since the 1980s.  My guess is that the claim is based on a widely cited paper published in 2006, "Social Isolation in America: Changes in Core Discussion Networks over Two Decades," which found that the percentage of people saying that in the last six months they had not "discussed matters important to you" with anyone went from 10% in 1985 to 25% in 2004.  They called this "social isolation," which sounds more or less equivalent to "loneliness," so you can see how one would turn into the other.  The change in discussion networks for "important matters" is interesting, if it happened (as the authors acknowledge, it might be at least partly an artifact of survey procedures), but it's not necessarily the same as a change in feelings of loneliness.

Two observations:
1.  It's remarkable that online editions of newspapers and magazines haven't developed reasonable conventions about when to include links to a source.  I checked five or six articles, all in well-regarded publications, which included the claim that levels of loneliness had doubled.  Only one provided a link:  that was to the AARP survey report, which didn't support the claim.
2.  There are cases when you can't say much about trends because there are recent survey questions, but no older ones.  This isn't one of them:  in addition to the "very lonely or remote," there was a 1964 survey asking people to agree or disagree with the statement "I often feel quite lonely" (27% did), and a 1990 Gallup Poll asking "How often do you ever feel lonely?" (10% frequently, 26% sometimes, 40% seldom, and 23% never) and a number of related questions.  There is also a Gallup question from 1950:  "When you have personal problems, do you like to discuss them with anyone to help clear them up, or not?" and a follow-up about who you discuss them with.

[Data from the Roper Center for Public Opinion Research]

Sunday, April 15, 2018

Is 80% a lot?

I was going to post on another subject, but things turned out to be more complicated than I thought, so here is a stopgap.  The Washington Post has a story about how Republican senate candidates are emphasizing their closeness to Donald Trump.  Although it offers some qualifications, it says that a major reason is that "Trump enjoys enormous popularity among Republican primary voters."  According to the most recent Gallup Polls, 85% of Republicans approve of the job Trump is doing.  How impressive is that?  Between June 1, 2017 and today, Trump's approval rating among Republicans has ranged from 79% to 87%.  Between June 1, 2009 and April 2010, Obama's approval ratings among Democrats ranged from 81% to 92%; between June 1, 2005 and April 2006, George Bush's approval rating among Republicans ranged from 75% to 89%.  That is, it's normal for a President to have "enormous" approval among people who identify with his own party. 

Only 7% of Democrats say that they approve of the job Trump is doing as President, but that's also normal.  Things get more interesting among Independents:  Trump has ranged from 30% to 33% approval (currently at 33%).  Obama ranged from 43% to 60% (in early April 2010 he was at 43%), and Bush from 23% to 45% (26% in early April 2006)

Overall, 41% approve of Trump's performance in the latest Gallup Polls, 47% approved of Obama at the equivalent time in his first term, and 37% approved of Bush at the equivalent in his second term.  The difference in overall popularity was concentrated among Independents.  Probably some of that is people switching from saying that they support a party to saying that they are independent.  It could be my memory, but I don't recall that Democrats were making efforts to tie themselves to Obama in spring 2010 or Republicans making efforts to tie themselves to Bush in 2006--in fact, they seemed to be going the other way and emphasizing their independence and commitment to do what's best for their state, which is a sensible strategy when your president is not especially popular. 

Why are candidates trying to move closer to an unpopular president?  I think that they are buying into the story that Trump has a particularly strong connection to "the base."  In my view, what actually happened was that once Trump got the nomination he benefited from party loyalty (and even more important, from dislike of the Democrats).    I've had several posts (especially this one) noting that there is no evidence that Trump voters were unusually enthusiastic. 

Tuesday, April 10, 2018


In my last post, I wrote about a piece by Thomas Edsall reviewing research that shows a large and increasing connection between "authoritarianism" and Republican voting.  Authoritarianism is measured by "a long-established authoritarian scale — based on four survey questions about which childhood traits parents would like to see in their offspring ..... Those respondents who choose respect for elders, good manners, obedience and being well-behaved are rated more authoritarian."  This scale measures something meaningful, but why call it "authoritarianism" rather than "traditionalism" or maybe even "conservativism"?  The basic idea of the "authoritarian personality," as proposed by Theodor W. Adorno, Else Frenkel-Brunswik, Daniel Levinson, and Nevitt Sanford, was that it was different from ordinary conservatism, and that an authoritarian conservative could appeal to people who didn't normally support conservatives (and repel some people who normally did)*.  Their idea of authoritarianism was complicated, but I would say that the essence is a tendency to say that every problem is the result of evil or contemptible behavior that ought to be punished (and sometimes to imagine problems so that you have an opportunity to blame someone).  This idea seems very relevant to Donald Trump--as Andrew Gelman said "Political scientists used to worry about authoritarianism within the electorate. Mainstream politicians, ranging from Republicans on the far right to lefties such as Sanders, tend not to go there. Trump did." 

The Authoritarian Personality team came up with questions that they thought measured the concept, some of which were used in surveys in the 1950s and 1960s.  One by the National Opinion Research Center included the following agree/disagree items:
"The most important thing to teach children is absolute obedience to their parents"
"Any good leader should be strict with people under him in order to gain their respect"
"Prison is too good for sex criminals.  They should be publicly whipped or worse"
"There are two kinds of people in the world:  the weak and the strong"
"No decent man can respect a woman who has had sex relations before marriage"

The survey also had a question asking if various kinds of people "are taking advantage of present conditions to make money."  The Korean War had started that summer, so I guess that was the "present conditions" they were talking about.  For ten of the twelve groups they asked about, people who said a good leader should be strict were more likely to say that they were taking advantage of conditions; for seven of those ten, the difference was statistically significant.  Those seven were Negroes, grocery store keepers, doctors, Puerto Ricans in the United States, Jews, bankers, and Catholics.  The three on which there was a non-significant difference in that direction were Labor union members, Protestants, and car dealers.  The two for which the difference was in the other direction (in both cases very small and nowhere near statistical significance) were farmers and steel companies.  The "authoritarian" answer was more common among less educated people, but controlling for education didn't change this the basic pattern--in most cases, it increased the t-ratios. 

So it does seem that an "authoritarian" answer on this question went along with a tendency to blame.  Of course, in some cases, it was plausible to say that a group was trying to make money--maybe grocery stores had raised prices.  But it's hard to imagine a reasonable argument that groups like Catholics or blacks were  doing that. 

The other questions had approximately the same pattern of correlations, but it was weaker and generally not statistically significant.  Some of those questions are dated or just don't seem very good in principle.  But maybe the "leader should be strict" question deserves to be revived.  It's not enough by itself, but it seems to be getting at something. 

*There has been some controversy about whether left-wing authoritarians exist; my view is that they do.

[Data from the Roper Center for Public Opinion Research]

Friday, April 6, 2018

Social science repeats itself

Thomas Edsall has a piece in which he cites a variety of work saying that Democratic and Republican voters are increasingly divided by values.  He's particularly concerned with "authoritarianism," which is an interesting issue, but one I'll save for another post.  What I want to talk about here is the idea that the recent rise in political polarization is the result of a rise of "cultural and lifestyle politics" at the expense of economic issues.  The reasoning is that it's easier to compromise on economics, on which you can split the difference, than on cultural issues, which involve principles of right and wrong.  The idea that culture has been displacing economics as the main axis of political conflict been around for about fifty years--it was first proposed in response to the developments of the late 1960s and early 1970s.  I think it has value (although with some qualifications which I discuss in this paper), but I don't see how it can explain the rising polarization of the last decade or so.  In that time, the single most divisive issue in American politics has probably been the Affordable Care Act.  This is basically an economic policy, and a very complicated one involving a lot of technical issues--that is, exactly the kind of issue where it seems you could make deals, offering a concession here in return for getting something  there.  The second most divisive issue has probably been the combination of bailouts, tax changes, and stimulus spending that gave birth to the Tea Party:  another complicated set of economic policies that seemed to offer lots of room for compromise.  Meanwhile, some leading cultural issues have faded.  For example, same-sex marriage is widely accepted--even people who aren't enthusiastic about it have mostly given up the fight.  Another example involves drugs:  a consensus seems to be developing in favor of legalizing and regulating marijuana, and the rise in opioid abuse has been treated as a public health problem rather than producing a "moral panic." 

What I think these examples show is that both economic and cultural issues can be more or less "moralized."  There was a period in the middle of the 20th century when leading politicians of both left and right accepted the basic principles of the welfare state and government intervention to maintain high employment.  But that consensus had not been around before then, and it isn't around now.  Now issues that were once part of what Seymour Martin Lipset called "the politics of collective bargaining" are part of the "culture wars."