Saturday, March 28, 2020

He's just not that into us

John Sides had a piece in the Washington Post called "Surprisingly few voters think Trump cares about 'people like me.'" I have been harping on this issue for several years--I think this was the first post and here is another.  But Sides found some surveys that I hadn't discussed, and made an important point that I hadn't stated explicitly:  the idea that Democrats care more about ordinary people is pretty much a fixed part of popular images of the parties--it would take more than a few promises to protect social security and close tax loopholes for a Republican to convince people that he's the one who cares more.  What was unusual about 2016 was not that Trump did well in this respect, but that Clinton did poorly for a Democrat.

Sides's column inspired me to look for more data on the subject, and I found a survey from 2011 that asked people whether Barack Obama:

"Can manage the government effectively"
"Is in touch with the problems ordinary Americans face in their daily lives"
"Is honest and trustworthy"
"Shares your values"
"Is not a typical politician"
"Is a person you admire"
"Is intelligent"
"Can get the economy moving"
"Is likeable"
It also asked the same questions about Donald Trump, who I guess was making noises about running for president then.  The comparison, arranged from those on which Trump did best relative to Obama to those on which he did worst:

                        Obama    Trump      Difference
Not typical politician    54%     68%         +14%
Tough enough              55%     57%          +2%
Economy moving            45%     47%          +2%
Intelligent               88%     81%          -7%
Manage                    47%     36%         -11%
In touch                  50%     28%         -22%
Honest                    57%     35%         -22%
Admire                    53%     29%         -24%
Shares values             50%     25%         -25%
Likeable                  83%     47%         -36%

Putting it another way, an overwhelming majority saw Trump as intelligent, a majority as not a typical politician, about half saw him as tough, able to get the economy moving, and likeable, but only a minority saw him as able to manage the government, in touch with the problems of ordinary people, or as sharing their values.  That is, the perception of caring was one of Trump's weaker points, not one of his stronger ones.

The survey also asked if you approved of the job Obama was doing as president, and if you had a favorable or an unfavorable view of Trump.  I did logistic regressions of each on the qualities.  I omitted "get the economy moving" because that could depend on your view of policies rather than personal qualities, and "is a person you admire" because that's a general judgment not a specific quality. The results (standard errors in parentheses):

                  Obama        Trump
Manage            2.47         1.51
                 (0.32)       (0.29)

In touch          1.34         0.04
                 (0.32)       (0.30)

Honest            0.94         1.15
                 (0.39)       (0.27)

Values            1.70          1.38
                 (0.34)        (0.29)

Not typical       0.13         -0.28
                 (0.30)        (0.28)

Tough             1.39          0.87
                 (0.35)        (0.34)

Intelligent       0.05          0.00
                 (0.69)        (0.53)

Likeable          0.59          1.71
                 (0.62)        (0.27)

The qualities of being a good manager, in touch, honest, sharing your values, and being tough enough made a difference for both.  There was no evidence that being seen as intelligent or not a typical politician mattered (controlling for the other qualities).  Being in touch with the problems of ordinary people mattered for Obama, but not for Trump (or at least not as much).*  So not only did people not see Trump as caring, but they didn't seem to care about whether he cared.  I once offered the hypothesis that "people [who supported Trump] saw him as an s.o.b., but thought that was what the country needed," and I think these results support that.

*Likeability seemed to matter more for Trump, although the evidence is not that strong.  That might have to do with the specific qualities that some people saw as likeable in each.  I think the people who liked Trump would mention qualities like "a fighter," or "not politically correct," while many people who liked Obama would think of less politically relevant qualities like being a good husband and father.   

[Data from the Roper Center for Public Opinion Research]

Sunday, March 22, 2020


An idea for a constitutional amendment that occasionally comes up is to make all citizens (not just "natural-born" citizens) eligible to be president.  In November 2004, a Gallup/CNN/USA Today poll asked about it:  "Would you favor or oppose an amendment to the U.S. Constitution that would allow a U.S.citizen who was born in a foreign country to be elected president?"   The most obvious argument in favor of the hypothetical amendment is the principle of equal treatment, and the most obvious one against it is established tradition: the founders thought the restriction was important enough to put in the Constitution and we've done it that way for more than 200 years.   That is, support or opposition is related to what you could regard as the basic principles of liberal and conservative philosophies, even though the amendment was not a subject of current political controversies. 

The survey also asked "Would you favor or oppose an amendment to the U.S. Constitution that would allow a U.S. citizen, such as Arnold Schwarzenegger, who was born in a foreign country to be
elected president?"  Schwarzenegger had been elected Governor of California in 2003, and was sometimes mentioned as a possible presidential candidate.  The comparison between answers to the two questions (which were asked to different randomly selected halves of the sample) seems interesting.  Answers to the second form presumably depend on two things:  your feelings about the general idea, and your views of Schwarzenegger.  Although the survey didn't ask people for their views of Schwarzenegger, other surveys taken at about the same time showed Republicans, conservatives, less educated people, and men had more positive views of him. 

I looked at race, gender, and age differences, and found nothing of note.  Support for the amendment by self-rated ideology

                                  no name           AS named
Conservative                34%                34%
Moderate                      27%                48%
Liberal                          38%                40%

The strongest difference was an unexpected one:  moderates were considerably more likely to support the proposed amendment when Arnold Schwarzenegger was mentioned.

Support by education:

not college grad             28%             43%
college grad                   43%             33%

When no name was mentioned, college graduates were more likely to support the idea (as I expected).  The mention of Schwarzenegger helped among non-graduates, but hurt among graduates.   I had thought that college graduates might be more influenced by principle, and less by their views of Schwarzenegger--there was some sign of this (10% difference between the two forms vs. 15%), but it wasn't clear. 

I wondered if the strong influence of naming Schwarzenegger among moderates might be because they tended to be less interested in politics, and therefore more influenced by the person than the principle.  The survey didn't have a question about interest in politics, but it did have one about whether you had voted in the presidential election, which had been held about two weeks earlier.

                                 no name          AS named
voted                            33%                  37%
didn't vote                    25%                  62%

Naming Schwarzenegger made a difference among non-voters.  The numbers in the category are small, so the size of the effect is probably exaggerated by chance, but it seems safe to say that it's real (p=.0002).  

The combination suggests that many non-voters and less educated voters were attracted to someone who was a political outsider.   There's obvious possible relevance to the 2016 election.

[Data from the Roper Center for Public Opinion Research]

Sunday, March 15, 2020

The Era of Bad Feelings

Since the early 1950s, the Gallup Poll has asked people to rate various public figures on a scale of +5 to -5.  During most Presidential elections (all except 1988, 1996, and 2000) since 1952, it has asked about the major party candidates (plus George Wallace in 1968, John Anderson in 1980, and Ross Perot in 1992).  I wrote about the balance of positive and negative ratings in a previous post, and mentioned that there was also a shift in extreme ratings.  Here I look at that shift in more detail.  The percent giving the major party candidates the highest possible rating:

Basically a downward trend--looking more closely, it seems that the trend was stronger earlier on and has leveled off since about 1980. 

Next, the percent giving the lowest possible rating:

The most striking thing here is the big jump in 2016, when both Hillary Clinton and Donald Trump had more -5 ratings than any previous major-party candidate.  But it seems that there was some upward trend before that time. 

Finally, I plot -5 against +5 ratings, distinguishing between three "eras"--1952-68, 1972-92, and 2004-16.  I include George Wallace and Ross Perot in this one (I forgot to record the data for Anderson, but he stood out by not having many strong reactions either way--5% plus 5 and 13% minus 5. 

In the first period, there was a lot of variation in +5 ratings, but only Goldwater and Wallace had large numbers of -5 ratings.  In the second period, there were fewer +5 ratings and somewhat more -5.  The third period is marked by large numbers of -5 ratings for everyone.  Some (Trump and Clinton) do worse than others, but no one does well by historical standards.  In 2004, 20% rated John Kerry at -5, breaking the previous record for a Democratic candidate (16% for McGovern in 1972).  In 2008 and 2018, Obama improved on Kerry in +5 ratings, but was almost the same in -5 (18% in 2008, 20% in 2012). . The omission of 1996 and 2000 makes it hard to identify exactly when the upward movement in negative feelings began, but in the ANES "feeling thermometer" for parties, the rise in very low ratings didn't seem to start until after 2000.  

[Data from the Roper Center for Public Opinion Research]

Sunday, March 8, 2020

Trump and Sanders

A little more than a week ago, I read a number of articles saying that Bernie Sanders's run this year was like Trump's in 2016--a populist outsider pushing aside a divided and demoralized opposition.  In the last few days, the story has changed to the Democratic establishment doing what the Republican establishment couldn't do--uniting to stop the progress of a populist outsider. 

These stories led me to wonder whether Sanders and Trump had any common appeal in 2016.  There were some anecdotal accounts of people who liked both, but I hadn't seen any systematic evidence.  One reason is that during the primaries, most surveys ask Republicans about the Republican candidates and Democrats about the Democratic candidates. However, I found a CNN/ORC survey from Feb 24-7, 2016 that asked everyone about both Democratic and Republican contenders, plus a few other political figures.  Specifically, it asked whether they had a favorable or unfavorable opinion about:  Michael Bloomberg (who was being mentioned as a possible independent candidate), Ben Carson, Bill Clinton, Hillary Clinton, John Kasich, Marco Rubio, Bernie Sanders, and Donald Trump (also Melania Trump, but I left her out). 

I computed correlations among the ratings of all of these people.  The correlations of the ratings of each Republican candidate with Bill Clinton, Hillary Clinton, and Sanders:

                       Bill             Hillary             Sanders
Carson            -.32              -.38                  -.27
Kasich            -.05              -.13                  -.06
Rubio             -.27              -.32                  -.19
Cruz               -.30              -.33                  -.29
Trump            -.30               -.39                 -.28

There is no sign of affinity between views of Trump and views of Sanders.  The negative correlation between views of Sanders and Trump was smaller than the negative correlation between views of Hillary Clinton and Trump, but the same pattern held for all Republican candidates.  The most likely reason Hillary had a stronger negative association than Sanders was simply that she was more familiar and was the front-runner for the nomination.  Kasich stands out as having smaller negative correlations; presumably that's because he was regarded as more of a centrist. 

I did a few factor analyses and found one major factor that could be interpreted as partisanship.  There was a second factor that could be interpreted as Trump vs. everyone else, or Trump vs. all other Republicans, but nothing suggesting common liking for Trump and Sanders.

The survey also asked which of the candidates could handle various issues the best.  I looked at the two where a Trump/Sanders affinity seemed most likely:  economics and immigration.  That is, to the extent that people saw both as trying to "protect American workers," they might have liked Sanders on the economy and Trump on immigration (or vice versa).  That was not the case--once you took out the tendency to like the same candidate on both issues, the only substantial pattern was partisan:  Democrats tended to like the other Democrat on the other issue and Republicans liked another Republican. 

Of course, it's possible to think of ways in which Trump and Sanders are similar (and ways in which they are different), but it seems that by late February, people were not treating them as having anything in common. 

[Data from the Roper Center for Public Opinion Research]

Sunday, March 1, 2020

Unanswered and mostly unasked

I have a book on public opinion coming out in September 2020.  Although there is a lot of good research on public opinion, some of which I discuss in the book, I've come to think that there are a number of basic questions that are not only unanswered, but which are mostly not even asked:
1.  Why have the effects of higher education on opinions changed?  Most people who follow politics are aware that college-educated voters have shifted towards parties of the left, and less educated voters have shifted towards parties of the right.  This is usually attributed to the growing importance of "social issues" (understanding "issues" in a broad sense), on which education goes with more liberal opinions.  But educated people have also moved to the left on at least some economic issues (see this post for an example).  Why?  I haven't seen any efforts to explain that movement, or even a clear recognition that it's happened.
2.  Why are their no major parties or politicians that are conservative on economic issues and liberal on social issues, or liberal on economic issues and conservative on social issues?  In the general public, opinions on social and economic issues are almost uncorrelated:  there are a lot of people who are to the left on economics and to the right on social issues, or vice-versa.  So you'd expect some politicians to appeal to these groups.  There have been lots of predictions that this will happen, or claims that it is already happening, going back to the 1970s, but so far it hasn't.   I've written about this in this blog, and Paul Krugman has discussed it in his newspaper column, but I haven't seen any discussion in the academic literature.
3.  When do the effects of different influences on opinion simply add together, and when do they combine in more complicated ways (interact)?  It's not unusual to find interactions.  For example, here are the percentages agreeing that people should have to get a police permit before buying a gun, broken down by race and gender:

                        Men          Women
White                67%            81%
Black                77%            85%

The combination of race and gender is more than the sum of its parts.  Gender differences are larger among whites than among blacks, and race differences are larger among men than among women.     There are lots of studies that look at interaction effects on specific opinions, but I haven't found any efforts to generalize about when interactions do or don't occur.
4.  Why are there long-term liberal trends in some (but not all) opinions?  There are a lot of questions on which opinions show a liberal trend over fifty or more years, but very few which show a conservative trend.  Looking  back over history, some "liberal" trends seem to have been going on for centuries:  for example, increasing support for the principle of gender equality.  In everyday discussion, people are well aware of this: a popular definition of a conservative is someone who wants to stop change or "turn back the clock."   But among intellectuals, claims that there is any tendency for moral progress have come to be regarded as naive or worse.  So there are few attempts to specify or explain the trends.  Those attempts tend to be by people who have a "grand theory" into which they try to force everything, so it's easy for specialists to ignore or dismiss them rather than trying to refine or improve on them.* 

*After reading Steven Pinker's The Better Angels of Our Nature a few years ago, I wondered how it had been received by sociologists, so I checked Contemporary Sociology, the American Journal of Sociology, and Social Forces, and found that none of them had reviewed it.  There was a review symposium in Sociology (the journal of the British Sociological Association), but only one of the four really discussed the book--the others were just expressions of indignation.  Claude Fischer gave it a serious review, but that was in the Boston Review, not a professional journal.