Thursday, January 20, 2022

Who's being suppressed?

A lot of the commentary on the voting laws that Republican legislatures are passing or considering holds that they will not only reduce turnout, but that reduce it more among blacks and other minorities.  The Current Population Survey has a module on registration and voting (not who you voted for, just whether you voted) that includes a large enough sample to give state-level estimates of turnout by race and ethnicity.  I looked at the relationship between voting rates in 2020 and the "cost of voting" index that I discussed in my last post.  The regression results

                        White            Black        Hispanic        Asian

Intercept           70.9            63.1            53.7                 60.5

                        (0.6)            (1.3)            (1.1)                (1.5)


COV                -2.6            -1.3                -1.2                -1.3

                        (1.0)            (1.9)            (1.4)                (2.1)

N                        50                35                34                25

The cost of voting index wasn't computed for the District of Columbia, and group estimates were not reported for some states because of small numbers (less than about 50 people).   Values of the cost of voting index range from -1.69 (Oregon) to 1.29 (Texas). 

The estimated relationship is negative--less turnout where laws are more restrictive--but it's larger for whites than for any of the other groups.  Given the standard errors, you can't be confident that there are  really any differences among groups, and the relationship isn't necessarily causal.  Still, these results don't support the idea that restrictive voting laws have more effect on blacks or other minorities.*

However, there are lots of laws and regulations that affect voting, and this index is just one attempt to reduce them to a single value.  Another approach is to look at the differences in black and white voting rates by state.  The largest and smallest ratios of white./black turnout rates:

Mass.     1.9

Wisc.      1.7

Okla.      1.4

Oregon   1.4

Iowa       1.3

Wash      1.3

Ark.       1.3


Del          1.0

NC           1.0

Missouri   1.0

Penna.      1.0

Tenn.        1.0

Miss.        1.0

Maryland  0.9

Sampling error has a substantial effect on some of these estimates, so you can't put much weight on the position of any single state.  The important point is that there is no pattern--I can't see anything that either group has in common, even if you allow for a couple of exceptions.  

 This connects to an issue that I discussed a few weeks ago:  the nature of the contemporary Republican party.  There are a number of historical examples of "how democracies die":  fascism, the end of Reconstruction in the South, caudillismo.  Most critics of the Republicans see them in these terms--e. g., they are disenfranchising black voters, just like the "redeemers" did in the 1890s.  Other observers argue (correctly, in my view) that none of the examples really fit, and conclude (incorrectly) that there's not too much to worry about: it's just "one of many percolating dangers in the United States today."  What's happened, I think, is that the Republican party has discovered a new way to be anti-democratic. 

Sunday, January 16, 2022


 Traditionally the United States had low rates of voting compared to most democracies, with especially low rates among blacks and the working class.  Some of the biggest barriers, like poll taxes and literacy tests, have been abolished, and registration and voting are easier than they used to be.  However, there are still things that state and local governments can do to make voting easier or more difficult.  Some observers say that groups which favor the Democrats, especially blacks, still face more obstacles.  There's a lot of information on group differences in voting rates, but they are affected by interest, motivation, and other factors.  So I looked for questions about experiences while voting or trying to vote and found a poll conducted by the Marist College Institute for Public Opinion in January 2020.  It asked whether you had experienced some things while trying to vote:  one was "having to wait on long lines at your polling place".  Answers were a lot, sometimes, hardly ever, or never, which I coded 1-4 with higher numbers meaning more often.  I looked at various demographic factors and found two that made a clear difference:  ethnicity and type of place:

White             1.70

Black              2.06

Hispanic         2.09

Asian              2.77*

Big city          1.95

Small city       1.89

Suburb            1.85

Small town     1.70

Rural               1.44

Since non-whites and people who live in urban areas tend to vote Democratic, one might expect a difference by party, but none was visible:

Democrats      1.74

Republicans    1.76

Independents   1.78

 Why not?  There were also differences by state:

 People were more likely to report having to wait in long lines in "red states".**  That is, in partisan terms the effects of the state differences offset the ethnic and urban differences.  Why would this be?  My guess is that the Republican states generally don't spend as much on government, so they tend to have fewer polling places and fewer voting machines, relative to the people who want to vote.  I wondered if anyone had attempted to make an index of the general difficulty of voting by state, and found this, by Scot Schraufnagel of Northern Illinois University,  Michael J. Pomante II of Jacksonville University and Quan Li of Wuhan University. Southern states had the highest scores, and some Midwestern states were also pretty high, while states in the Northeast and Pacific coast had low scores.   The index was correlated with reported waiting time, somewhat more strongly than Biden vote was.   I didn't have time to look into exactly how it was constructed, but I think it's impressive that a correlation with state differences in reported waiting time is visible, since sampling error has a substantial effect on state-level estimates because of small numbers of cases. 


*The numbers are small, so it's not clear if the mean is really larger for Asian-Americans than for blacks and Hispanics, but it's pretty safe to say that it's larger than the mean for whites. 

**I limited it to whites since I thought that state effects might differ by race.  Only states with more 10 or more respondents are shown. 

Tuesday, January 11, 2022


  This figure shows a summary of answers to three survey questions:  

1.  "Hard work offers little guarantee of success"  [agree/disagree]

2.  [choose between] "First statement: Most people who want to get ahead can make it if they're willing to work hard. Second statement: Hard work and determination are no guarantee of success for most people."

3.  " Hard work and determination are no guarantee of success for most people" [agree/disagree]

The vertical axis gives the log of the ratio "optimistic" (disagree on 1, first statement on 2, disagree on 3) to "pessimistic" answers.  


I thought that #2 would get more optimistic responses than #1 and #3, because it explicitly presented the optimistic alternative.   It did get more optimistic responses than #2 in the early years, but the difference was pretty much gone by 2010--there was no trend in answers to #1, but a pessimistic trend in answers to #2.  #3 was only asked between 2012 and 2016, so you can't say anything about trends, but it clearly gets more pessimistic responses than the others. 

In an attempt to get a better sense of what was going on, I looked at breakdowns by education (college graduates vs. everyone else) for early and late examples of #1 (1987 and 2012), and #2 (1999 and 2020).  The percent giving optimistic answers:

1.  "hard work offers little guarantee of success"

            not grad      college grad

1987     65%                82%

2012     60%                75%

2.  most can get ahead vs. no guarantee

1999      75%                80%

2020      60%                55%

On #1, more educated people are more optimistic and the educationaldifferences are about the same in both years.  On #2 the educational differences are smaller, and they change direction:  more educated people are more optimistic in 1999 but less optimistic in 2020 (and that change is statistically significant).  

My interpretation:  When people are asked question #1 they think mostly of themselves, and because most of us believe in ourselves, we are more optimistic when answering about ourselves than about people in general.  The educational difference on that question is understandable if people are thinking of their own experience ("I worked hard and I.....").  Question 2 refers to "most people," so answers are more about perceived fairness of society.   The change in the educational differences supports something I've said a number of times, that more educated people are getting more egalitarian (despite what critics of "meritocracy" say).   But there was some shift towards pessimism among less educated people as well--that's consistent with the shift on questions on the causes of poverty (lack of effort vs. circumstances).    

[Data from the Roper Center for Public Opinion Research]

Thursday, January 6, 2022

What could happen here, part 2

A couple of weeks ago, I suggested that Republican elites took advantage of the complexity of the electoral system to make objections to Joe Biden's election--they knew that they weren't going to be successful, but figured that they would appeal to "the base."  One additional point that I didn't mention is that some of the complexity comes from the Constitution, which left the conduct of elections to the states, or more exactly, the state legislatures:  "Each State shall appoint, in such Manner as the Legislature thereof may direct, a Number of Electors."   As many observers have remarked, Americans have a reverence for the Constitution--not the philosophy behind it, but the words.  If something is in the Constitution, many people will think that it represents timeless wisdom.   I think this is part of the reason that Republican elites have been so reluctant to accept that Joe Biden won.  The objection that some decisions about voting rules (e. g. absentee ballots) were made by courts or by election officials may be a technicality to the average person, but if you believe we should follow the original text of the constitution, it seems like an important principle.  Moreover, it raises the possibility that a state legislature could decide to appoint the electors in a different way--for example, if the Democrats win the vote in 2024, a Republican legislature could decide that it's going to go ahead and appoint Republican electors anyway.  

This means that the Republican party has been able to drift away from support for democracy without really becoming authoritarian. 

Friday, December 31, 2021

The wisdom of the people

 A few years ago, I had a post on questions on "how much trust and confidence do you have in the wisdom of the American people" from 1964 to 2016.  It's been asked a few times since then (as recently as last month), so I'll give the updated figures:

There's clearly been a decline since the 1960s, or even the early 21st century.  My original post just considered the means, but this one will look at educational differences in the 1964 survey and the two latest surveys for which individual-level data are available, August 2015 and March 2016 (all of them asked about "the wisdom of the American people when it comes to making political decisions."  I excluded blacks because I thought that their opinions might follow a different pattern, although it didn't turn out to make much difference.  The means, on a scale of 1-4 with higher numbers indicating more trust and confidence:

                         Not college grad            College Grad

9/1964                       2.93                           2.74

8/2015                       2.29                           2.33

3/2016                       2.29                           2.23

In 1964, people with a college degree had less trust and confidence in the wisdom of the people.  In 2015 and 2016, the differences by education were small and not statistically significant.  The t-value for the drop among college graduates between 2015 and 2016 was about 2.5, so there's some sign that their confidence declined during that time--the obvious reason would be the success of Donald Trump in the Republican primaries (by March 2016 he was the clear favorite to get the nomination).  Still, the major point is that there was a clear educational difference in 1964, but not in 2015-6.  

As it happened, there were some important parallels between 1964 and 2016--the Republicans had nominated (or were about to nominate) someone who was opposed by the party establishment and was thought to have some extreme ideas.  Journalistic accounts of Goldwater and Trump supporters were similar--they suggested that they were fanatical, new to politics, and of relatively low social standing.  So maybe the 1964 difference reflected an elite* reaction against Goldwater?  Apparently not:  support for Goldwater was higher among college graduates (45% vs. 27% among non-graduates).  There were also questions about how much you agreed with what the candidates stood for, and college graduates were more likely to say that they agreed with Goldwater.  Finally, there was a question about why you intended to vote for a candidate--you were enthusiastic, you were not enthusiastic but OK with him, or you didn't like the other candidate.  College graduates who supported Goldwater were more enthusiastic than non-graduates who did (and more enthusiastic than college graduates who supported Johnson).  So while Goldwater was unpopular, he wasn't especially unpopular with college graduates.  That is, if there was any loss of confidence because of the nomination of Goldwater it should have applied at least as strongly to people without a college degree. 

Why did the gap between college graduates and other people diminish or disappear?  I'll go back to a point I've made before:  despite what is often said, people of higher social standing have become more sensitive about seeming elitist, and less willing to say that they know better than other people.  Of course, the question I've talked about here isn't definitive, but all the evidence I've seen points that way. 

*And in 1964, only about 10% of American adults had graduated from college, so it was a more elite status than it is today.

[Data from the Roper Center for Public Opinion Research]

Thursday, December 23, 2021

What could happen here?

 This is a post I've been meaning to write for a while.  The impetus was a column by Ross Douthat on October 5 called "The Once and Future Threat of Trump," which argued that he had been basically right in a column from 2020 called "There Will be No Trump Coup":  that even if Trump had been more organized and competent, he didn't have the institutional support he would need to pull off a coup.  Douthat said that this continued to be true, so Trump's "threat to constitutional norms is one of many percolating dangers in the United States today, not a singular danger that should organize all other political choices and suspend all other disagreements."   I think he is right in saying that the probability of a coup was, and continues to be, very low.  The reasons are a strong military tradition of political neutrality and an independent judiciary that operates on the basis of a large accumulation of legal precedent.  

Douthat also said that Trump didn't get much support from Republican elected officials:  "No statehouse leader proposed setting aside the popular vote, no state legislature put such a measure on the floor, no Republican governor threatened to block certification."   But he doesn't mention that Texas filed suit which asked (to quote Wikipedia):  "to temporarily withhold the certified vote count from these four states [Pennsylvania, Arizona, Wisconsin, and Georgia]  prior to the Electoral College vote on December 14."  Seventeen other state attorneys general signed on to a brief supporting the suit, and a majority of the Republican members of the House of Representative (about 125) signed on to another one.  After the Supreme Court denied the suit and the Electoral College cast its votes, most Republican representatives, and several Senators, voted against "certifying" them on January 6.  So some Republican officials didn't go along with Trump, but others did.

The American electoral system (especially for Presidents) is very complex and provides many opportunities for grandstanding--making a statement that you can be pretty sure won't make a difference.  For example, the attorneys general undoubtedly knew that the Supreme Court would reject the suit, and the Republican representatives knew that the Democrats had enough votes to get the certification approved.   

The complexity of the system also means that there's no definite stopping point--if they put their minds to it, people can think of new ways to grandstand and pass things along to someone else (usually the courts).  So last time no state legislature tried to set aside the popular vote, but if the Democrats win narrowly in 2024, I'd be willing to bet that one will.  And after seeing how much heat Doug Ducey and Brian Kemp have taken, some Republican governor may decline to certify a Democratic win.  In either case, they could say (and maybe believe) that they're not really trying to overrule the vote, they're just raising questions.  The courts would probably reject these efforts, but at some point you might an unpredictable decision (especially in the state courts), so things might remain unsettled until almost the day of inauguration, or even after.  So elections could start becoming like votes to raise the debt ceiling, where we've gotten used to crises that are averted at the last minute.  I'm surprised that Douthat doesn't anticipate something like this, since it fits with his idea of decadence--a political system that is increasingly unable to do even basic things. 

Sunday, December 12, 2021


Following up on a couple of points from recent posts:

1.  I noted that confidence in newspapers held steady or rose during the Trump years.  Confidence in TV news rose from 2016 to 2017, and then declined, but it's been declining pretty steadily since the 1990s.  Overall, nothing particularly unusual happened to confidence in the media during the Trump years, although " many observers (especially but not exclusively on the right) claim that people lost confidence in the media during the Trump years as many journalists moved away from their traditional efforts to appear neutral."  A few days ago, Ross Douthat had a column making exactly that claim:  "from his shocking November victory onward, much of the press adopted exactly the self-understanding that its critics are still urging as the Only Way to Stop Trump . . . the public’s trust in the national press declined during the Trump era."  My post used data from Gallup, and he links to data from Pew.  Do they point in different directions?  The Pew question is "how much, if at all, do you trust the information that comes from each of the following?"--answers are a lot, some, not too much, and not at all.  The Gallup questions are asked once a year, in May or June, and the Pew questions are asked at irregular intervals starting in 2016.  The Pew survey asks about national and local "news organizations"--here is a graph of the averages (higher numbers mean more trust):

No change from February 2016 to July 2019, and then a drop between July and November 2019.  The next survey was in June 2021, and it showed a drop from November 2019.  The timing doesn't fit Douthat's argument, which implies that there should have been an immediate drop starting when Trump took office (the second survey was in March 2017).  Also, Douthat specified the "national press," but trust in national and local news organizations also followed a very similar pattern (a correlation of .95).  That suggests that variations reflected a general trust factor (either real temporal variation or sampling variation), not a reaction to changes in national news coverage.  

2.  In recent years, the national Democratic party has united against any restrictions on abortion.  In my last post, I observed that this position is not very popular among the public:  in a recent survey, 15% said it should be legal in all cases during the second trimester and 35% said that it should be illegal in all; in the third trimester, the figures were 8% and 54%.  But when asked to choose between the Democrats and Republicans on abortion, more people favor the Democrats.  A 2019 poll by Marist College asked "Do you think the Democratic Party or the Republican Party would do a better job of dealing with the issue of abortion?":  46% said the Democrats and 33% the Republicans (the rest said they would be the same or that they were unsure).  Why?  Some of it may be the focus on Roe v. Wade:  many people seem to be under the impression that overturning it would mean a national ban on abortion.  The Marist poll had a question about general view of abortion, with the options of available any time (17%), only in the first six months (11%), only the first three months (22%), only rape, incest, or to save the life of the woman (28%), only to save the life (9%), or never (9%).  They also had a question about a candidate who "would appoint justices to the Supreme court to limit or overturn Roe vs. Wade":  options were would vote for, would vote for but with reservations, and would definitely not vote for.  A cross tabulation:

                         Would     With Reservations        Definitely not    Not sure

Always                8%                   7%                          79%                   6%

Six months          3%                 10%                          84%                   3%

Three months      5%                 18%                          63%                  13%

Rape, Incest       24%                 26%                          36%                  14%

Life only            47%                 14%                         19%                   20%

Never                 65%                   7%                         19%                     9%

Roe vs. Wade basically found that abortions should be allowed in the first six months, but most of the people who say that they should be limited to the first three months, and 36% of those who say they should be allowed only in the cases of rape, incest, or to save the life say that they would definitely not vote for a candidate who would appoint justices to overturn the decision.  But this raises the question of why so many people have a mistaken impression about a long-standing issue. 

[Data from the Roper Center for Public Opinion Research]