Tuesday, September 19, 2017

They did it their way

Since my last post was long and complicated, I thought I should follow with something short and simple.  In 1987, the Roper Organization asked "Thinking about the way your own life has turned out so far, would you say it has been primarily a matter of luck or fate, or has it been more a matter of factors which are within your control?"  The same question was asked in CBS News polls in 1996 and 2016.  The results

                  Luck   Your Control   Both    DK
1986           22%        66%             9%       3%
1996           18%        72%             6%       4%
2016           27%        60%             9%       4%

The differences in the relative frequencies of luck and own control are statistically significant.  It seems possible that opinions on this are affected by economic conditions--when people experience bad things like unemployment or reduced income, they are likely to say it's luck.  However, as I recall economic conditions in 1986 and 1996 were roughly like they were in 2016--pretty good but not outstanding.

[Data from the Roper Center for Public Opinion Research]

Sunday, September 17, 2017

More old news

About six months ago, I saw several stories saying that "Having just one black teacher can keep black kids in school," to quote NPR's summary.  They all noted the magnitude of the effect:  almost 40% reduction in dropout rates for low-income black boys.  I located the paper on which the stories were based and thought about posting on it, but it was a long paper by the time I got around to reading it, the attention seemed to have passed.  However, last week's NY Times magazine had a list of statistics on education, and one of them was "exposure to at least one black teacher in Grades 3 to 5 reduced the probability of low-income black male students dropping out of school by almost 40%."  So that led me back to the paper.

The thing that originally attracted my attention was not the general idea that having a black teacher would help to keep black children in school, which seemed plausible, but that it could reduce dropouts by 40% for any group.  There is a lot of data on basic educational outcomes like finishing school, and by the standards of social science it's high quality data.  Moreover, there are a lot of people who have studied the issue, so it seems that any simple and straightforward way to dramatically reduce dropout rates would have been discovered long ago.

The paper reports that the estimated effect on dropout rates is -.04 for all black students, -.06 for persistently low-income black students, and -.12 for persistently low income black male students.  Since about half of students are boys, that suggests that the estimated effect on persistently low income black female students would be about zero, and indeed they report an estimate of 0.00 for that group.  So the issue was treating only the big estimate as worthy of interest.  If you believe that there are differences in the effects on boys and girls (and the difference appears to be statistically significant), both of the estimates are equally important; if you don't, you should just report the estimate for boys and girls combined.  The differences between persistently low income students and other students don't appear to be statistically significant (it's hard to tell from the tables), so maybe you should just report the estimate for all students.

There's also a more complex issue which relates to the way that they got the estimate.  The simple approach would be to do a regression with dropping out as the dependent variable, and having a black teacher plus some other variables as independent variables.    But the authors say that those estimates "are likely biased by unobserved student characteristics that jointly predict classroom assignments and long-run outcomes, even after conditioning on the basic socio-demographic controls in X and school FE (Rothstein 2010). For example, students with lower achievement (Clotfelter, Ladd & Vigdor, 2006) and greater exposure to school discipline (Lindsay & Hart, 2017) are more likely to be matched to black teachers, and these factors likely affect long-run outcomes as well."  That is, black teachers tend to be given the kind of students who are at higher risk of dropping out.  The authors had an idea on how to eliminate this potential bias.  They had multiple students from each school, which means that they could include a dummy variable for each school.  That's a reasonable thing to do, since it's generally agreed that some schools are more effective than others.  They also had five different classes of students:  those who started third grade in 2000, 2001, 2002, 2003, and 2004.  Because of new hires, departures, and leaves, the percent of the teaching staff that was black could change from year to year.  Those personnel changes would depend on idiosyncratic individual factors--getting pregnant, reaching retirement age, having a spouse get a job offer in another state--so they would be random from the point of view of the students.  So you can use within-school variation in the racial composition of the teaching staff over time as a substitute ("instrument") for the original variable (having a black teacher or not) and get unbiased estimates.

This approach strikes me as clever but not very convincing.  Teachers' decisions to stay or go will depend partly on how rewarding it is to work in a school.  That could depend on student performance (teachers like it when their students do well) or on things that might affect student performance, like discipline problems, or how well teachers get along with the administration.  Things get more complicated because what matters is differential effects on black and white teachers, but I can think of possibilities here too:  for example, black teachers may be particularly interested in how the black students are doing.  I think I might trust the simple results more than the results from their method--at any rate, I'd like to see them, but they aren't reported in the paper.

This isn't a straightforward mistake, but the sort of difference of judgment that often comes up with research, and the authors could probably say more in defense of their approach.  But I will stick with my original feeling that a 40% reduction in dropout rates for anyone is too big to be believed.

Saturday, September 9, 2017

Alternate history

This came up several weeks ago, but I hadn't gotten around to posting on it.  On July 26, Elizabeth Hinton (a professor of History and African-American Studies at Harvard) reviewed three books on race and policing in the New York Times.  She said that one of the books (Chokehold, by Paul Butler), "demonstrates that when citizenship rights are extended to African-Americans, policy makers and officials at all levels of government historically used law and incarceration as proxy to exert social control in black communities. Black Codes, convict leasing and Jim Crow segregation followed Emancipation; overpolicing and mass incarceration followed the civil rights movement."  This reminded me of the figures on crime and prison that I showed in a previous post.  I concentrated on crime rates there, so here is more detail about imprisonment.  First, the figure for 1929-86, which was the period covered in the data source (it kept going up after 1986):

 
The next one focuses on the period during and after the civil rights movement:


The rate of imprisonment didn't start rising until 1973, when the movement had either faded away or become mainstream (that is, not really a "movement").  During the period of peak activity of the civil rights movement, the rate of people in prison declined or stayed about the same, although crime was increasing.  In a literal sense, the rise in imprisonment did follow the civil rights movement, but the suggestion of cause and effect is not very credible.




Saturday, September 2, 2017

Respect

The Pew survey I mentioned in my previous post had a series of questions about "how much respect do you think Donald Trump has" for various groups "a great deal, a fair amount, some, or none at all."  Then the same questions were asked about how much respect Hillary Clinton had for the groups.  The averages, ranked from greatest to least average respect for the group:

                               Trump       Clinton
White people                    3.23       3.03
Men                             3.21       2.77
Veterans                        2.84       2.73
Women                           2.19       3.16
Blue collar workers             2.64       2.66
Black people                    2.31       2.84
Evangelical Christians          2.66       2.42
Hispanic people                 2.16       2.85
Immigrants                      2.00       2.93
Muslims                         1.86       2.93
People who support [opponent]   1.91       1.91

The standard errors are about .03 or .04.  The ratings aren't surprising--Trump is seen as having substantially less respect for women, black and Hispanic people, immigrants, and Muslims, and somewhat more respect for white people, men, veterans, and evangelical Christians.  However, it's noteworthy that Trump and Clinton are rated almost exactly the same in respect for blue collar workers--this is one of many pieces of evidence that contradicts the popular story that working class voters turned to Trump because they thought that liberals were condescending to them.  It's also notable that Clinton was seen as having pretty high respect across the board--her perceived respect for Evangelicals, which was lower than her perceived respect for any group except Trump supporters, was higher than Trump's perceived respect for six of the groups.

What difference did these perceptions make?  I regressed intended vote on each candidate's perceived respect for the groups (one half of the sample was asked about women, men, whites, blacks, Hispanics, and veterans; the other about Muslims, evangelicals, immigrants, blue collar workers, and people who supported the other candidate).  The logistic regression coefficients, with positive values meaning that more perceived respect for the group goes with more support for the candidate (standard errors are typically about .2 or .3, and standard errors of the differences about .3):

                              Trump       Clinton
White people                   -0.31      -0.03
Men                             0.12      -0.09
Veterans                        0.66       1.14
Women                           1.13       0.32
Blue collar workers             1.10       1.19
Black people                    1.17       1.06
Evangelical Christians          0.34       0.91
Hispanic people                 0.74       0.62
Immigrants                      0.57       0.04
Muslims                         1.53       0.72
People who support [opponent]  -0.13       0.77

I don't think that perceptions of respect are necessarily causes of the way that people vote:  to some extent, probably a large extent, people are rationalizing the way they voted.  But the way that people rationalize their actions is still interesting.  

A large coefficient could mean that a group is held in high esteem (that people think it should be respected) or regarded as important in some sense.  But from that point of view, the coefficients for white people and men are puzzling.  Another factor could be whether respect from the candidate in question could be taken for granted.  For example, there wasn't much doubt that Hillary Clinton respected women, and it didn't make much difference in support for her; there was a lot of doubt about whether Trump did, and it made a lot of difference in support for him.  So the fact that perceived respect for men and white people didn't matter could be because most people thought they'd be all right regardless of which candidate won (this contradicts another popular story, about how Trump supporters were motivated by perceived threats against whiteness or masculinity).

The combination of these principles seems to make sense of the coefficients, with one exception:  the difference in the effect of Trump and Clinton's perceived respect for supporters of their opponent.  Clinton gained from being seen as respecting Trump voters; Trump didn't gain from being seen as respecting Clinton voters.  This pattern suggests there's a bit of truth in the "liberal condescension" story--that on the average, people cared more about whether Clinton respected them than whether Trump did.

[Data from the Roper Center for Public Opinion Research]

Saturday, August 26, 2017

What went wrong?

Two years ago, not many people thought that Donald Trump would get the Republican nomination.  In late July, a McClatchy/Marist College polls asked Republicans "Do you think Donald Trump is a serious presidential candidate, or a distraction from the presidential primary process?"  44% said a serious candidate, and 51% said that he was a distraction.  On August 13, 2015, I had a post that offered probabilities of getting the nomination:  40% for Jeb Bush, 20% for Marco Rubio, 15% for John Kasich, 12.5% for other declared candidates, and 12.5% for someone else.  If I'd gone farther, I think I would have put Trump at or near the top of the others, with maybe 5%. I give myself some credit for not being impressed with the field--this was when people were still talking about the "deep bench." But no matter how you slice it, I thought Trump had little chance of winning.  Sometimes unusual things happen--someone who had little chance of winning could make it because of a lot of unpredictable things.  But looking at the campaign, I don't think that Trump got all that many lucky breaks, so I'm not going to try to defend my prediction that way.  I think that a large part of the reason I didn't give him much chance was that I thought he wouldn't stay in very long--that the first time he encountered adversity he would quit, complaining that the rules were rigged against him.  But why was I confident of that?  I was certainly familiar with Donald Trump--I'd been hearing his name since the early 1980s.  But I didn't really know much about him--I hadn't read any of his books, or biographies of him, or watched his media appearances.  So my mistake was in taking general familiarity for real knowledge about him--I should have started by thinking about what would happen if he was serious about the race, and if I had I would have rated him higher.

The same post offered some ideas about the sources of his support at the time. I said that a large part of his appeal was negative--people distrusted politics and politicians, and he was an outsider.  An alternative is that he made a positive connection to a large number of voters.  Reporters who attended his rallies were often struck by the energy, and many people talked about an "enthusiasm gap" in favor of Trump during the general election campaign.  A Pew survey in October 20-5, 2016 asked separate questions about whether Trump and Clinton would be:  great, good, average, poor, or terrible.  The results:

                          Trump             Clinton
Great                    9%                  10%
Good                  17%                  28%
Average              15%                 18%
Poor                   12%                  12%
Terrible              47%                  31%

That's not much enthusiasm for Trump, but it includes both supporters and opponents.  If we limit it to people who said they would support the candidate:

                         Trump             Clinton
Great                  21%                  19%
Good                  39%                  54%
Average              30%                  25 %
Poor                      7%                   2%
Terrible                 2%                   0%

At least in this respect, Clinton supporters were move favorable about their candidate than Trump supporters were about theirs.  In fact, 10% of the people who said they would vote for Trump thought that he would be a poor or terrible president.

On a possibly related note, of the people who had the same expectations of Trump and Clinton (e. g., said both would be average), 75% said they would vote for Trump.  I haven't investigated, but one possibility is that they were Republicans who figured that even if he wouldn't be especially good, Republicans in Congress would get their way if he were president.  So I think my original analysis was correct on that point--Trump got the Republican nomination more because of his opponents' weakness than because of his strength.  After he got the nomination, party loyalty kept him close enough to have a chance.  

[Data from the Roper Center for Public Opinion Research]

Saturday, August 19, 2017

Don't blame the millennials

I have seen a number of articles challenging the "myth" that millennials are less racist than previous generations, and the rally in Charlottesville has inspired more of them.  For example, in the Washington Post, Catherine Rampell writes:

   "If there was one silver lining to President Trump’s election, it was supposed to be this: Those who voted for Trump because of, rather than despite, his demonization of Muslims and Hispanics; who fear a 'majority minority' America; and who wax nostalgic for the Jim Crow era were mostly old white people.
    Which meant they and their abhorrent prejudices would soon pass on — and be replaced by generations of younger, more racially enlightened Americans.
     The white nationalist rally this past weekend in Charlottesville clearly proves this to be a myth."

She points out that many of the participants in the rally, including the man who killed one counter-protester and injured about 20 more by driving his car into a crowd, were young.  But young people, especially young men, are more likely to engage in all kinds of violence, and high-risk behavior more generally. Also, the numbers who participated in the white nationalist rally were small:  according to Wikipedia about 100 on Friday night and 500 on Saturday.  So the only myth that the rally disproves is a myth that no one believes:  that absolutely no young people hold racist views.

But Rampell also offers some more serious evidence:  a story called "white millennials are just about as racist as their parents," which is based on analysis of General Social Survey data from 2010-14.  It considers five issues:  ratings of how intelligent and hardworking blacks and whites are (each based on one question about whites and one about blacks), how you would feel if a relative intended to marry a black person, how you would feel about living in a neighborhood that was 50% black, and that a reason for racial differences in jobs, income, and housing is that "most blacks just don't have the motivation or will power to pull themselves up out of poverty."  It reports that "White millennials (using a definition of being born after 1980) express the least prejudice on 4 out of 5 measures in the survey, but only by a matter of 1 to 3 percentage points, not a meaningful difference."
(By "white" they meant non-Hispanic white).

I redid the analysis, making the following changes:
1.  Including three more variables, whether there should be a law against marriages between blacks and whites, whether racial differences were because blacks had less inborn ability, and whether blacks shouldn't push where they aren't wanted.
2.  Adding data from 2016, and 2000, 2002, 2004, 2006, and 2008.  The 2016 data wasn't available when the article was written.  As far as 2000-2008, some millennials were old enough to be included in the survey in all of the years, and generational differences tend to be enduring.
3.  Using averages for variables that were measured with more than two categories (like ratings of intelligent and hard working."  I'd say that a person who rates blacks at 3 and whites at 4 is different from a person who rates blacks at 1 and whites at 7.
4.  The story just included the "silent generation" (born 1928-45), "baby boomers" (1946-64), "generation X" (1965-1980) and millennials.  I also included people born through 1927.

The percent giving the "racist" response for the three yes/no items:

                          Marriage law   Inborn    Willpower          
Oldest                     27%            25%        69%
Silent                      17%            15%        59%
Boomers                   8%              7%        45%
X                               6%              6%        44%
Millennials                3%              5%        38%

The means for the other items (higher numbers mean more "racist"):

                            Don't push     marry    intelligent  lazy       half    
Oldest                    2.85             3.77         1.01          1.30      3.32
Silent                     2.50             3.45           .60          1.01      3.12
Boomers                2.11             3.03           .35            .65      2.97
X                            1.97             2.77          .32            .53       2.90
Millennials             1.85             2.63          .23            .36       2.87

Millennials are least prejudiced on all eight of the questions:  in fact, each generation is less prejudiced than all previous generations on all eight of the questions.  As far as whether the difference is meaningful,  there's no absolute standard, but one way to judge it is to do a principal components analysis, which gives a score for each generation:



According to this, the difference between millennials and boomers is about half as large as the difference between boomers and the "silent generation."  Although the rate of change has slowed down, racial prejudice is still declining from one generation to the next.

Saturday, August 12, 2017

The way it is

In the New York Times last week, Nate Cohn writes "The polls don’t tell a clear story [about public opinion on affirmative action]. Some polls show that affirmative action is very popular. Others show that it’s not popular at all." I think that they tell a pretty clear story--a large majority of people don't think that race should be considered in college admission. The difference among polls occurs because "affirmative action" covers a lot of things, and some of them are popular--for example, special efforts at outreach to minorities. There's a related issue that hasn't received much attention--how do people think that things actually work? In 2003, 2005, and twice in 2007 the Gallup Poll asked "If two equally qualified students, one white and one black, applied to a major U.S. (United States) college or university, who do you think would have the better chance of being accepted to the college--the white student, the black student--or would they have the same chance?" The distribution of answers was similar on all occasions, so I'll just give the average:

 White     Black     Same      DK
  30%         23%      42%       5%

 Unfortunately the individual-level data aren't available for any of the polls, but even if you make the extreme assumption that every black respondent said that the white student would have the better chance, less than 30% of whites said that the black student would have a better chance.

 [Data from the Roper Center for Public Opinion Research]