Yes and no

 This post follows from my previous one, which was about opinions on abortion.  I mentioned that the GSS has questions about "whether it should be possible for a pregnant woman to obtain a legal abortion" in six different circumstances, and also one about whether it should be possible to obtain a legal abortion if "the woman wants it for any reason."  I said I would leave the last question out because "I suspect that some respondents interpret it as asking whether abortion should ever be allowed." This post reports on some further analysis of that question. I computed a variable as the sum of the number of "yes" answers for the six different conditions (risk to health, pregnancy due to rape, chance of birth defect, single and doesn't want to marry the father, married and doesn't want more children, too poor to care for a child).  A cross-tabulation of answers to the "any reason" question and the number of conditions for which they said abortion should be legal:

                 0          1          2         3          4         5         6

Yes           0%  0.1%   0.5%   1.7%  1.9%  3.1%    36.8%

No         8.3%  6.2%   9.4%  20.1%   5.9%  3.5%   2.4%

The base for the percentages is the whole sample, which is cumulative from 1977-2018--e. g., 36.8% of the respondents said that a woman should be able to get an abortion for any reason and in all six specific circumstances.  That's a consistent position in favor of legal abortion.  No on "any reason" is consistent with any number of yes answers to specific circumstances--it means that you think there should be some limits on legal abortion (even someone who says yes on all six conditions that the GSS asked about might think there are some cases in which abortion should not be legal).  But the ones highlighted in red are logically inconsistent--a woman should be able to get a legal abortion for any reason, but not for some of the reasons offered in the other questions.  Those answers add up to 7.3%.   There's some pattern in the "inconsistent" answers:  they are more common among less educated people and people who score lower on a test of vocabulary, suggesting that they're given  by people who don't recognize the contradiction or misunderstand the "any reason" question.

But there's also a pattern over time:

The "inconsistent" combinations are becoming more common.   

Consistent support for a general right to abortion (ie, "yes" on the any reason question and all six specific questions) has increased somewhat.  

Belief that abortion should (almost) never be allowed--that is, saying no to all six specific questions and the general question--also has increased.

That means that what you could call "mixed" opinions--that abortion should be allowed for some reasons, but not any reason--has declined.  Of course, you could say that the "inconsistent" positions are also mixed in a sense.  I wondered if the increase might be limited to Democrats, who might want to say that they are "pro-choice" in principle even if they have doubts about some cases, but it was about equally strong among independents and Republicans.  

I'm not sure how to interpret this.  I've noted before that there's been some movement towards extreme positions on abortion, but I don't know why "inconsistent" positions should have become more common.  If they were simply based on misunderstanding the "any reason" question, as I initially thought, they should have been roughly constant, or maybe declining because of the increase in educational levels. 

Abortion rights

 In the New York Times, Jon Shields notes that overall levels of support for legal abortion have remained almost the same over the last fifty years, even as opinions on many "social issues" have become more liberal.  He suggests that this is because both sides can justify their position by an appeal to rights--the rights of women or the rights of  "unborn children."  In contrast, with many social issues, the appeal to rights only works on the liberal side.  This led me to think about a series of General Social Survey questions about whether "whether it should be possible for a pregnant woman to obtain a legal abortion" in different circumstances:

1.  "there is a strong chance of a serious defect in the baby"

2.  "she is married and does not want any more children"

3.  "the woman's own health is seriously endangered by the pregnancy"

4.  "the family has a very low income and cannot afford any more children"

5.  "she became pregnant as a result of rape"

6.  "she is not married and does not want to marry the man"*

 I was particularly interested in the "serious defect" question.  Traditionally, that was one of the most widely accepted justifications for abortion, but recently I have seen some opponents of abortion argue that it amounts to discrimination against the disabled--that is, abortion because of the chance of a birth defect is doubly bad.  I wondered if this idea has had any impact on the general public.  

Support for legal abortion is considerably higher for #1, #3, and #5 than for #2, #4, and #6, so I show them in two different graphs:

 These are all circumstances which people who think of themselves as "pro-life" might see as justifying abortion.  Although large majorities are in favor of legal abortion in these cases, support has been declining.  The largest decline is for "a serious defect," suggesting that the discrimination argument may have had some impact.  

 The other questions show a different pattern--support declined until about 2000, but has risen since then.  The increase has been largest for the "married and does not want any more children"--support for legal abortion in those conditions is now as high or higher than it was in the 1970s.  I can't think of a good reason why that one is different from the others.  

I should point out that all of the changes are small, but I think they suggest that Shields is correct--that opponents of abortion are increasingly thinking in terms of the rights of the unborn.  That way of thinking makes it harder to justify exceptions for rape or a serious defect.  

I also considered the associations among opinions on abortion in the different cases.  Generally, they have increased--there's an increasing tendency to support or oppose abortion across the board.  But there is one clear exception to that pattern--the association between opinions about whether abortion should be allowed if the woman's health is endangered and if there is a defect has declined since the 1970s.  I think this also suggests that there has been a shift in the way people think--at one time, they were both thought of as "medical reasons" but now they are increasingly differentiated. 

*There is also one about if "the woman wants it for any reason."  I leave this out partly because it didn't start until 1978 (the others started in 1972) and partly because I suspect that some respondents interpret it as asking whether abortion should ever be allowed.

 

A matter of degree, part 2

 In my last post, I said that score on an index of "intellectual" tastes was associated with greater support for the left candidates (Gore or Nader) in the 2000 election, even after controlling for education.  The survey also had a lot of questions relating to politics.  I selected 20 of them, a combination of concrete issues (like laws on abortion) and general views that are commonly seen in left/right terms (like whether the growing number of newcomers threatens traditional American customs and values).  I controlled for a number of standard demographic variables plus education and "intellectual" tastes.  The estimated effect of intellectual tastes was in a liberal direction for 19 of the 20, and the estimate for education was in a liberal direction for 16 of 20.  By and large, the education and intellectual estimates were parallel--if one was large, so was the other.  I'm not sure if they are completely parallel, but that's the tendency. Going back to the question that started me on this, this suggests that there is a continuum, not a contrast between "elite" colleges and the rest of society--people who are more interested in school or do better in school tend to be drawn to liberal political views.  That is, I'd expect that students at flagship state universities are more liberal than those at second-tier state universities, and that honors students at a given university are more liberal than average students.  

Turning to specific questions, the association is stronger for "social" than "economic" issues--in fact, it's near zero for the (few) economic issues.  That's no surprise, but there are also substantial differences among the social issues.  The effects are strong for civil liberties, what you could call "cosmopolitanism" (acceptance of immigrants and concern with other nations), and opposition to racial discrimination.  Those are also in line with previous results on the effects of education.  But there was another pattern which I hadn't seen discussed before:  more educated/intellectual people were substantially more likely to disagree with the statements that racial discrimination was rare, and that "poor people have become more dependent on government programs."  I think those questions have something in common--they involve blame, or suggesting that people are responsible for their misfortune.  That goes along with something I have argued before:  that educated people have become less confident of their moral superiority.  In that sense, contemporary critiques of "meritocracy" have it backwards. 

[Data from the Roper Center for Public Opinion Research]

A matter of degree

 Matthew Yglesias has a piece pointing out that overwhelming majorities of students at "elite" universities favor the left.  Is this due to something specific to elite universities, or are they just one end of a continuum running from elite colleges through colleges at different levels of prestige until we get to people who don't go to college?  David French seems to favor the first hypothesis--he points to class and geographic differences in who applies to elite universities:  "we’re left with two Americas living two very different educational realities. At Harvard, Joe Biden wins almost 90 percent, and Donald Trump finishes third. At the University of Tennessee? Two weeks ago I sat in the student section with my son as the students led an entire stadium in a sing-song chant, 'F**k Joe Biden.'"  Yglesias seems to favors the second:  "it does seem to be true that the people who are best at school are on the left politically."  But neither one presents any systematic data. 

Most surveys don't ask people where they went to college, or how well they did in college, so you have to take a less direct approach.  One way is to look at the maps of precinct-level voting compiled by the New York Times.  You find a lot of college towns in Republican states are islands of blue--places like Morgantown WV, Vermillion SD, Manhattan KS.  They aren't as heavily Democratic as places like Princeton or Ithaca, but are definitely different from the surrounding areas.  But many of the votes in those towns may be from faculty or staff, or just people who like living in college towns, so they don't necessarily tell use about college students or college-educated people more generally.  A few surveys ask about cultural tastes, but most of those don't ask about politics.  In a post based on a Gallup poll from 1990, I noted that people who had read more "serious" authors were more likely to be Democrats and liberals.  Now I'll discuss a Pew survey from 2003 that asked about a wider variety of political issues.  This survey also asked people how they felt about a numbers of musicians, entertainers, and leisure activities.  A factor analysis suggested that four of them--"scientific shows and magazines," "Thai or Vietnamese food," "reading books," and "Mozart" formed a group.  I made an index adding together ratings (on a four-point scale from dislike very much to like very much) of those four.  The mean is 11 with a standard deviation of 2.7.  

 If you regress vote in 2000 on that index, the estimated chance of supporting a candidate on the left (Gore or Nader) increases by 1.5%, with a t-ratio of 2.9.  If you add race, gender, city size, age, income, and education as controls, the estimate is 1.7% with a t-ratio of 3.3.  The magnitude of the estimated effect was comparable to the estimated effect of education itself (the gap between college graduates and others was about 6.5%, controlling for the same variables).  That is, people with what you could call more intellectual tastes were more likely to vote for a candidate of the left, even controlling for education.  So the evidence seems to favor the second interpretation--the politics of elite universities are an extreme case of a more general tendency.  Why?  In my next post, I'll look at the association of "intellectual" tastes with opinions on some issues.  

[Data from the Roper Center for Public Opinion Research]