Saturday, August 13, 2022

The era of (relatively) good feelings

Since the 1970s, the Gallup poll has asked people about how much confidence they have in a variety of institutions.  The 2022 Gallup data has recently come out, and I will write about it soon.  But today I will write about a different series of questions on confidence, from surveys sponsored by the Pew Research Center.  The general form is "How much confidence, if any, do you have in each of the following to act in the best interests of the public?"  [Institution].  Why look at these questions, when the Gallup series covers a much longer period of time?  One reason is that they consider a different group of institutions.  Another is that they are taken at irregular intervals, unlike the Gallup surveys, which are taken every June.  As it turns out, one of the Pew surveys was carried out in April 2020, in the early stages of the Covid pandemic; another in May-June 2020, when George Floyd was murdered and a wave of protests began; and another in late November 2020, just after the Presidential election.  As a result, the Pew surveys make it possible to look at the impact of these events.  

The Pew surveys regularly asked about scientists, medical scientists, elected officials, religious leaders, principals in public schools, journalists, police officers, business leaders, and the military.   Possible answers were "A great deal of confidence, a fair amount of confidence, not too much confidence, no confidence at all": I counted these as 4,3,2,1 and computed the averages.  I show these groups together because people had high confidence in all.  

It turns out that they had a similar pattern of change--rising from 2016 to 2019, then pretty steady despite everything that happened between January 2019 and November 2020, and then lower in the latest survey (December 2021).  

Principals have a similar pattern, although they are higher in April 2020, which may may indicate that people came together to support schools in the early stages of Covid.  Data for police starts only in 2018--there has been a steady decline since then.

Religious leaders have the same pattern of a rise followed by a fall, although it may be weaker than for some of the other groups.  Confidence in business follows a completely different course.  

Finally, two groups with low confidence:  elected officials and journalists.

The question about journalists wasn't asked until January 2019--there has been a steady decline since then.  Elected officials follow the more common pattern--higher in the Trump presidency, at least the later part, but then declining between November 2020 and December 2021.  

In November, I had a post using the Gallup data that also showed a general rise after 2016 (going through June 2021).  The general point that emerges from both analyses is that confidence in a variety of institutions tends to rise and fall together.  There's no way to be sure about what drives it, but I think that a combination of economic conditions and general satisfaction with politics are the most likely sources.  People often suggest that confidence in particular institutions mostly reflects the performance of those institutions.  For example, today's column by Ross Douthat said that confidence in science and public health experts has declined in the last few years because the "expert community" has departed from neutrality and allowed political goals to influence their recommendations.  There are certainly examples of that, but most people don't pay that much attention to the news, so things like the "Open letter advocating for an anti-racist public health response to demonstrations against systemic injustice occurring during the COVID-19 pandemic" don't have much impact.  In contrast, when prominent politicians saying that everything you've been told is wrong, even people who don't follow the news closely are likely to be aware of that.  

[Data from the Roper Center for Public Opinion Research]

Friday, August 5, 2022

Something new

 Most of my recent posts have involved the Supreme Court and/or abortion.  I'm sure I'll return to those topics, but this post will be on Social Security.  It's prompted by a blog post from Mark Palko, who notes that many observers think that the traditional Republican disadvantage on Social Security has disappeared.  The idea is that in 2016 Donald Trump pledged not to cut Social Security and Medicare and the party has followed him, as it has on many issues.  Mark questioned this account, but didn't have any data that directly addressed it, so I looked and found some.  

In most presidential elections starting in 1984, there were questions about which candidate would be better on Social Security.  They were not all by the same organization, so the wording varied.  Most of the variations were minor (e. g.  "handling" vs. "dealing with"), but in 1988 and 1992 they asked about "protecting the Social Security system" and in 2016 they asked about "Social Security and Medicare."   I calculated the difference between the percent naming the Democratic candidate and the percent naming the Republican.  The Democrat was always ahead, which is why I call the figure "Republican disadvantage."

In 2016, Trump trailed Clinton by 50-42%, giving an 8% gap, which was just about average--unfortunately the question wasn't asked in 2020.   Reagan in 1984 stands out as an unusually large gap, which is plausible because in one of his Presidential campaigns (I think it was his first, in 1976) he suggested that maybe Social Security should be privatized and got a lot of negative publicity.     Aside from that, there's no trend, and the ups and downs don't show any obvious pattern and are small enough so that they could be sampling error.  So there's not evidence that Trump changed anything--the Democrats consistently have an advantage on the issue.   This isn't really surprising--even someone who doesn't pay much attention to politics can tell that if forced to make a choice between tax increases and spending cuts, Republicans would be more likely to go for spending cuts and Democrats would be more likely to go for tax increases.  

So why do many observers continue to say that Trump changed perceptions?  I think that it's partly because of surveys showing that in 2016 Trump was regarded as less conservative than other Republican candidates, and partly because he did well among the "working class" (less educated people).  The idea that he positioned himself to the left on economic issues is appealing because it seems to explain both of those points.  However, I don't think that they actually had a common cause--the perception of him as less conservative probably involved social issues, while his appeal to less educated voters involved a combination of immigration and trade and personal style (blunt, not worried about offending people). 

[Data from the Roper Center for Public Opinion Research] 

Monday, August 1, 2022

Can't stop writing about the Supreme Court

 I will move on to other topics soon, but I found one more point about the Supreme Court that I want to mention.  In July 1991, the Gallup Poll asked "From what you know about Clarence Thomas, as a Supreme Court justice, do you think he would be too liberal, too conservative, or just about right?"  7% said too liberal, 20% too conservative, 46% about right, and 27% didn't know.  The same question has been asked about most of the other Supreme Court nominees since then (in all cases, very soon after they were nominated, before the hearings started).  The figure shows the expected bias (absolute value of "too liberal" minus "too conservative").

There's a clear tendency to increase, which can be seen as another aspect of political polarization, but there are a few nominees who depart from the trend.  Perceived bias was relatively low for Merrick Garland (6% too conservative, 25% too liberal) and high for Amy Coney Barrett (43% too conservative, 4% too liberal), and also for Elena Kagan (6% too conservative, 40% too liberal).  I was surprised by the results for Kagan--I didn't remember much controversy about her.  Her "don't knows" were unusually low, only 4%.  That may be because the question followed a number of questions that gave information about he.  The next figure shows the percent saying "about right":

There's no trend here, and Kagan doesn't stand out, but once again Garland and Barrett are exceptions.  The results for Garland aren't surprising, since he was perceived as a moderate.  But why does Barrett stand out, when Neil Gorsuch didn't?  Both were conservatives, both were nominated by Trump, and you might have expected Democrats to be especially negative about Gorsuch since he was filling the seat that had opened under Obama.  

The answer may be the justices they were replacing--Scalia for Gorsuch and Ginsburg for Barrett.  Replacing Scalia with Gorsuch didn't have much impact on the ideology of the court, but replacing Ginsburg by Barrett moved it substantially to the right.   After Scalia died, Gallup asked if you would like a justice "who would make the Supreme Court more liberal than it currently is, more conservative than it currently is, or keep the court as it was?"  32% said more liberal, 29% more conservative, and 37% keep it as it was.  Of course, replacing Scalia with Garland would have made the court more liberal, but he was regarded as more moderate than others Obama might have chosen, so people who wanted to keep things as they were would have been happy.  Unfortunately, the question about how people would like the Court to change was not asked in 2020, but if there was a similar desire to keep things about the same, replacing a liberal justice with a solidly conservative one would be unpopular.  

Like Barrett, Clarence Thomas moved the court well to the right--he replaced Thurgood Marshall, who was among the most liberal justices.  The fact that people were pretty satisfied with his nomination may reflect general trust, or maybe many people wanted the court to move to the right at that time (I'll try to look at that in a future post).  

[Data from the Roper Center for Public Opinion Research]

Thursday, July 28, 2022

More on the Supreme Court

  A comment on my last post noted that the Harris Poll question saying the Supreme Court had decided "that the police could not question a criminal unless he had a lawyer with him" was a mischaracterization--the decision just said that the police had to inform suspects of their right to have a lawyer present--and suggested that support might have been higher if the description had been accurate (saying "suspect" rather than "criminal" might also have made a difference).  Shortly after the decision, the Gallup Poll asked "have you heard or read about the recent US Supreme Court decision about confessions by persons charged with a crime."  If the respondent said yes, they asked "do you think the Supreme Court's ruling is good or bad?"  Combining the two questions, the results were:

Hadn't heard:  58%

Good:              10%

Bad:                 19%

Not sure:          13%

The comparison with the Harris Poll (which found that 30% approved, 56% disapproved, and 14% weren't sure) illustrates some general issues with interpreting surveys.  You could argue that the Gallup results show that most people didn't really have an opinion on the Miranda decision--that most people who said they approved or disapproved in the Harris Poll were just giving an off-the-cuff response to the (inaccurate) description.  On the other hand, you could say that there may also have been people who had heard about it but forgotten, and just needed a hint to jog their memory.   Despite the differences, both surveys showed about the same ratio of negative to positive evaluations (almost 2:1):  that is, they both suggested a widespread perception that the court was "soft on crime."  You could imagine other questions that would tell you something about whether that perception could be changed--for example, what if you used a different description of the decision, or gave some background information, or asked people who said that they had heard of the decision what they thought that it said.  One of the reasons that candidates commission their own surveys is to investigate things like that.  The general point is that there's no definitive question that tells you what people "really think," and when you have several questions on a topic, you don't have to pick one as the best--they can all shed some light.   

My recent posts on the Supreme Court got me wondering about how perceptions of the court have changed.  Starting in 1986, a number of surveys, mostly by Gallup, have asked "do you think the Supreme Court has been too liberal, too conservative, or just about right" (with some minor variations in the introduction).  The figures gives the percent saying too liberal minus the percent saying too conservative:

The different colors indicate Democratic and Republican presidents.  There is a clear pattern in which people perceive the Court as too liberal when a Democrat is president and too conservative when a Republican is.  This "thermostatic" reaction has been noted for other issues, but it's surprising that it appears for the Supreme Court.  The last observation (September 2021) is an exception--despite a Democratic president, the perception of the court as conservative is even stronger than it was under Trump.  The absence of a general trend is also interesting--most informed observers would say that the court has been getting more conservative over the period.

The percent who saw the court as "too liberal" or "too conservative" shows a different pattern--a pretty steady increase, without much difference between Democratic and Republican administrations.  There isn't much change in the percent saying "about right," but the percent of "don't knows" has declined pretty steadily, from 15-20 percent in the 1980s to around 3 percent today.  This change presumably reflects a combination of rising partisan polarization and increased attention to the Supreme Court.

[Data from the Roper Center for Public Opinion Research]

Thursday, July 21, 2022

One thing led to another

 This post started as a postscript to my previous one, on public opinion about the Supreme Court in the 1960s.  I gave figures from a 1966 Harris poll on opinions of six decisions, but they were based on people who had an opinion.   I thought there might be substantial numbers of "don't knows" for some of them, so I looked for the original data, which can be found in the Odum Institute data archive at the University of North Carolina.  So here they are:

"In 1954, the US Supreme Court ruled that it was illegal to require Negro children to go to all-Negro or segregated schools.  Do you personally think that decision of the US Supreme Court was right or wrong?"  right 57%; wrong 32%; DK 11%.

The others were all of the form "Another decision of the US Supreme Court was to ____.  Do you personally ...".  Just giving the key text:

"rule that the State Department could not refuse to allow Communists from travelling abroad"  36%-39%-25%

"rule all Congressional Districts had to have an equal number of people in them so each person's vote would count equally"  56%-24%-20%

"rule that children could not be required to recite a prayer in school" 27%-65%-8%

"rule that the police could not question a criminal unless he had a lawyer with him" 30%-56%-14%

"rule that any refusal by a hotel, motel, or restaurant to serve a person because of their race was illegal" 56%-34%-20%

The survey also asked "all in all, how would you rate the job the US Supreme Court has done over the last 10 to 15 years--excellent, pretty good, only fair, or poor?"  Among those who had an opinion, 8% said excellent, 38% pretty good, 32% only fair, and 22% poor.  

My last post suggested that changes in approval of the Supreme Court might have reflected changes in the importance that people placed on different issues--in  particular, increased focus on crime, an area where the court's decisions were unpopular.  Since the data had an overall rating and measures of the approval of some specific decisions, I decided to use it to investigate this point.  Although it couldn't shed light on change, it could tell us something about how much the different decisions mattered to people.  I did this by regressing ratings on views of the decisions (counting don't know about a decision as an intermediate position).  I did this separately for college graduates and other people, because I suspected that there would be some difference between the groups.  The figure shows regression coefficients for the two groups (actually, the regression coefficients multiplied by 10).

The line represents equal values in the two groups--points above the line mean a decision counted more among college graduates.  Most points were above the line--this is what I expected, because opinions are usually more closely connected to each other among educated people.  The least popular decision--school prayer--was the least important among both college graduates and other people.  The relative importance was similar in both groups, with one major exception--the Miranda decision (right to a lawyer) was the most important among less educated people but only fifth among college graduates.  That decision was unusual in another way--there was little difference between approval among college graduates and less educated people.  On all the others, college graduates were substantially more likely to approve.  

In a previous post, I mentioned that a question in 1994 found overwhelming approval (88%) of the decision on segregated schools.  A survey in 2000 asked  about "the recent decision upholding 'Miranda Rules' requiring police to inform arrested suspects of their rights to remain silent and to have a lawyer present during any questioning? Do you generally agree or disagree with this decision?" and found that 86% approved and only 11% disapproved.   I think that there would also be overwhelming approval of the non-discrimination, equal population in districts, and school prayer decisions (school prayer still is controversial, but the decision referred to in 1966 involved public school teachers leading students in prayer during classroom hours).  Some of this change reflects a widespread liberal trend in opinions on "social issues".  But with the Miranda decision, I think that another factor was that it led to a standard procedure--police gave a "Miranda warning"--which made it easier for people to get used to it.  In contrast, the abortion ruling has led to confusion, as some states have new laws that don't make clear exceptions for cases in which there's overwhelming popular support for legal abortion, like the 10-year old rape victim in Ohio, some have conflicts between local and state governments, and some have controversies about reviving laws that were in force before Roe v. Wade.  So my prediction is that support for legal abortion will grow in the next few years.

Friday, July 15, 2022

The Supreme Court in the 1960s

In my last post, I discussed predictions that the Supreme Court would lose legitimacy because of public disapproval of the decision overturning Roe v. Wade.  I noted that ratings of the Supreme Court were favorable in 1963 and 1967, but then fell sharply from June 1967 to July 1968.  What might have caused this change?  I searched for contemporary survey questions about Supreme Court decisions.  In November 1966, the Harris Poll asked a series of questions--I give the percent approving, among those who had an opinion*:

School desegregation                                     64%

Equal numbers in Congressional districts:    76%

No right to refuse service because of race:    64%

Can't stop Communists from foreign travel:  49%

Right to attorney in police questioning:         35%

school prayer                                                  30%

In all of these cases, the court had ruled in what was considered a liberal direction.  Some of the rulings were popular with the public and others were unpopular.  

There were no questions about Supreme Court rulings between June 1967 and 1968.  Thurgood Marshall had been nominated just a few days before the 1967 survey, and the confirmation hearings took place in the following months.  There were no national survey questions about views of the Marshall nomination, but one from the Minnesota Poll found overwhelming approval in that state.  My guess is that it didn't make much difference--the people who disapproved of him who already had a negative view of the court because of civil rights ruling.  

So maybe the change wasn't because of what the Supreme Court did over that period, but because of other things in the world.  1967 was the "long hot summer" of urban riots--most of them, including the ones in Detroit and Newark, took place after the 1967 survey.  There were more riots after the assassination of Martin Luther King in April 1968.  In July 1968, another Harris Poll asked people if they agreed or disagreed with several statements about the Supreme Court.  For "The present court has been good in the desegregation and one man-one vote decisions,"  41% agreed, 25% disagreed, and 34% weren't sure.  For "The present court made it harder to convict criminals and was wrong to ban prayers from schoolrooms," 78% agreed, 11% disagreed, and 11% weren't sure.  So the events of 1967-68 may have increased the salience of issues about crime, on which most people disapproved of the court decisions.  In June 1968, Gallup asked about whether people would like new members of the Supreme Court to have liberal or conservative views:  51% said conservative and 30% said liberal, although the number of self-described liberals and conservatives was about equal at that time.  As I mentioned last time, views of the Supreme Court eventually recovered--that was partly because the Court became more conservative, and partly because the public came to accept some decisions that had initially been unpopular (most aspects of the Miranda ruling).

*I would have liked to include don't knows, but they weren't given in the source.  

[Data from the Roper Center for Public Opinion Research]

PS:  here's a short piece I wrote with Andrew Gelman that was published earlier this week:

Tuesday, July 5, 2022

Losing legitimacy?

 Since the Dobbs decision, I have seen many claims that it has (or will) cause the Supreme Court to lose legitimacy (here is one example).  It does appear that most people did not want the court to overturn Roe v. Wade, but this is not the first unpopular decision (and almost certainly not the most unpopular).  That raises the question of why this particular decision should have a lasting effect.  It seems like a reasonable model to start with would be that decisions affect views of the court based on a combination of popularity and prominence, but that the impact declines over time as memories fade and people get used to the new situation (which is almost never as bad as critics of the decision predicted it would be).  On the other hand, there may be some decisions that have a lasting impact, maybe because they have unexpected consequences that keep them in the news (for example, if state differences in laws on abortion lead to disputes between states that produce more cases for the Court) or just because they seem so unfair to a substantial part of the public (the case that decided the 2000 election might be an example).  

According to the Gallup Poll, confidence in the Supreme Court has generally declined since the 1970s--see this post and the latest update from Gallup.  But there has been a decline in confidence in many institutions, so it's hard to say what part of that is unique to the Supreme Court.  So I looked for data going farther back, and found one from Gallup that was asked six times from 1963 to 1987:  "In general, what kind of a rating would you give the Supreme Court:  excellent, good, fair, or poor."  You could say that "confidence" gets at something deeper than "rating," but I think that most people don't  make much distinction, so that they are reasonably good substitutes.  What makes this question useful, aside from the general benefit of extending the period for which data are available, is that there were a lot of important decisions in the 1960s.

The average rating was lower in the middle (1968, 1969, and 1973) and then bounced back.  An interesting point is that there was less dispersion in 1986 and 1987 than there had been in 1963 and 1967--fewer "excellents" and fewer "poors".  There was a sharp decline in average confidence  between June 1967 and June 1968.  In my next post, I will try to find evidence, or at least hints, about what might have caused that.

[Data from the Roper Center for Public Opinion Research]