Landon maintains lead

As part of an effort to avoid thinking too much about tomorrow's election, I'll return to a 1936 "straw poll" of college students that I wrote about in June.  I had found results for a number of colleges, but I knew that there were more, so I kept looking.  It turns out that the New York Times reported some additional results, mostly in a story of October 30, 1936.  It may seem strange that the Times ran a story on this, but before polls had developed, there was a lot of interest in anything that could provide hints about the outcome of the election.

Here are the additional colleges and universities, ranked from biggest Roosevelt lead on down:

                        Dem    Rep    Soc    Comm
CCNY                     62      4     12      23
New Rochelle             69 29     1       0
Manhattan                66 30      1       0
Rider                    53 43      1       1

Carnegie                 46 49      2       2
Delaware                 47 51      2 0
Drexel                   45 50      3 0
Franklin & Marshall      39 47     13 0
Upsala                   45 53      1 0
Swarthmore               37 48     13 1
Stevens                  38 52      7 2
Wash & Jefferson         25 47     26 1
Bucknell                 35 58      3 2
Penn State               37 60      2 0
Rhode Island             37 61      2 0
Rutgers                  37 62      1 0
Tufts                    32 59      4 4
MIT                      34 60      3 3
Sweetbriar               36 62      2 0
Conn Coll                30 67      2 1
Susquehanna              29 67      3 0
Gettysburg               30 69      2 0
Skidmore                 24 72      4 0
Temple                   20 67      4 9
Mass.                    22 72      5 0
Elmira                   24 73      2 0
Lehigh                   17 77      2 2

Mean                     38     55      4       2

The Vassar list was limited to elite universities, while the New York Times covered a wider range.  Despite this difference, the means were almost the same--today, I think that Republican support would be lower in elite universities than in others.  I wouldn't necessarily put much faith in the results for any particular one--it looked like turnout was generally high at the small colleges, but pretty low at the larger ones, but given the consistency it seems safe to say that college students leaned strongly Republican in the 1930s.  

Not that complicated

A few days ago, an article in the New York Times by Amanda Taub said that working-class support for Donald Trump reflected a "crisis of white identity."  Today, Ross Douthat said that it reflected the "thinning out of families."  The basic idea in both was that "working class" (ie less educated people's) opposition to immigration is a symptom of anxiety about something else.

In September 1957, the days of the baby boom and the "affluent society," when unions were strong and no one was talking about a crisis of white identity or masculinity, the Gallup Poll asked "UNDER THE PRESENT IMMIGRATION LAWS, THE HUNGARIAN REFUGEES WHO CAME TO THIS COUNTRY AFTER THE REVOLTS LAST YEAR HAVE NO PERMANENT RESIDENCE AND CAN BE DEPORTED AT ANY TIME. DO YOU THINK THE LAW SHOULD OR SHOULD NOT BE CHANGED SO THAT THESE REFUGEES CAN STAY HERE PERMANENTLY?"
42% said yes, and 43% said no.

In July 1958, another Gallup Poll asked "IN EUROPE THERE ARE STILL ONE HUNDRED AND SIXTY THOUSAND REFUGEES WHO LEFT HUNGARY TO ESCAPE THE COMMUNISTS. IT HAS BEEN SUGGESTED THAT THE U.S. PERMIT SIXTY-FIVE THOUSAND OF THESE PEOPLE TO COME TO THIS COUNTRY. WOULD YOU APPROVE OR DISAPPROVE OF THIS PLAN?"
33% approved and 55% disapproved

With both questions, education made a difference for opinions.  For example, in 1958, 55% of the people with a college degree favored letting the refugees come to the United States, compared to 31% of those without college degrees.  The only other demographic factor that made a clear difference was that Jews were more likely to favor letting the refugees stay.

The 1957 survey also had a question about the Brown vs. Board of Education decision against school segregation--people who approved were more likely to favor letting the refugees stay.  The 1958 survey had a series of questions about whether you would vote for various religious or racial minorities for president--people who were more tolerant were more likely to favor letting the refugees come to the United States.

The Hungarian refugees were white, Christian, and could be seen as part of a clear story of oppression vs. resistance.  Despite this, most people, especially less educated people, were not in favor of letting them stay in the United States.  So the contemporary opposition to immigration, and the tendency for it to be stronger among less educated people, are not a reflection of something specific to today, but continue a long-standing pattern.  Of course, an increase in the number of immigrants today presumably makes the issue more important.  But the basic pattern is not new.

[Data from the Roper Center for Public Opinion Research]

College professors and capitalism in the 1950s

In 1955, the Opinion Research Corporation ran a survey they called "Free Market versus Socialist Thinking," which was about basic economic issues.  I've written about it before, but until recently I hadn't noticed that it included special samples of a number of occupations:  Catholic clergy, Protestant clergy, college professors, high school teachers, business executives, other teachers, and manufacturing foremen.  They didn't give any details on how those samples were selected, but for what it's worth, here is the distribution of party identification for each group:

                              Rep    Dem   Other/None
Catholic clergy*     13%    52%     35%
Protestant clergy*  36%   37%     27%
Professors              31%    47%    22%
HS Teachers          40%   49%    11%
Executives*           62%   22%    16%
Other teachers       38%   50%     12%
Foremen                31%   47%     21%
Public                    37%   45%     18%              

The differences from the general pubic are statistically significant for the three groups marked with asterisks.

Today, college professors are far to the left of the general public.  I'm interested in figuring out how long this tendency has existed, so here are comparisons between the professor sample and the general sample on some other questions.

DO YOU THINK IT WOULD BE A GOOD THING FOR THE COUNTRY IF THE GOVERNMENT PUT A TOP LIMIT ON THE SALARY ANY MAN COULD MAKE?
   
                       Yes       No
Professors        9%     86%
Public             18%     78%

DO YOU THINK IT IS THE RESPONSIBILITY OF THE FEDERAL
GOVERNMENT TO SEE THAT EVERYONE WHO IS WILLING AND ABLE TO
WORK HAS A JOB?

                      Yes       No
Professors       43%     48%
Public             45%     48%

DO YOU THINK THE FEDERAL INCOME TAX SHOULD BE CHANGED TO
MAKE THE RATE STIFFER ON PEOPLE WITH BIG INCOMES, OR NOT
SO STIFF?

                           Stiffer   Less     As Is    
Professors           23%    22%       48%
Public                 43%    15%       36%

Professors were to the right of the general public on economic issues, although they may have been more likely to identify as Democrats.  I suspect they were to left on "social issues" like opposition to McCarthyism and racial segregation, although the survey didn't contain any questions on them.

[Data from the Roper Center for Public Opinion Research]

Predecessing the Unprecedented

In many ways, Donald Trump is unique in American politics, but as I've mentioned before, I think he has some parallels with Ross Perot.  Below the presidential level, I wondered about Jesse Ventura, who was elected governor of Minnestota as an independent in 1998.  Ventura had more political experience than Trump--he'd served one term as mayor of a city of about 85,000.  He also apparently had libertarian leanings, while Trump has mostly been a standard conservative since being nominated.  Still, it seems that for both of them, a basic part of their image was that they were outsiders who would apply toughness and common sense.

Minnesota has a state poll, sponsored by the Minneapolis Star Tribune, that goes back many years.  I combined results from two asked in the second half of October 1998 to get breakdowns by education and gender:

                                            Ventura
High school or less                 30%
Some College                         34%
College Graduate                    20%

Men                                         31%
Women                                    16%

The differences by education and gender are similar to those found with Trump today.

 Some contemporary analyses hold that the big difference by education in support for Trump and Clinton reflects long-term economic change--less educated people, especially men, are losing out from globalization.   I don't know about Minnesota in particular, but the economic situation in the late 1990s was generally regarded as good, so the example of Ventura counts against this analysis.  An alternative is that less educated people (maybe especially less educated men) are more attracted to tough-talking outsiders and less likely to be bothered by "gaffes" or outrageous statements.  That would suggest that large educational and gender gaps are specific to this election rather than a sign of things to come.

[Data from the Roper Center for Public Opinion Research]

Now that's what I call a gender gap

The gender breakdowns in responses to two recent survey questions:
         
Question 1:
        Yes      No
Men      37%     62%
Women    20%     80%

Question 2:

         Yes     No
Men      51%     46%
Women    36%     63%

The gender gaps are about equal to, or maybe somewhat bigger than, those found in the latest surveys of voting intentions in the Trump-Clinton race.  So what were these questions?

1.  Do you think the penny should be eliminated, or not?
2.  Do you think the penny is a useful coin, or should the government do away with it? ["do away with"="yes" in the table above.]

I'm not surprised that men gave more support to abolishing the penny--if I'd been asked before, looking at the results, I think I would have given odds of 3:1 or 4:1 that the difference would be in that direction--but why would the gap be so large?

[Data from the Roper Center for Public Opinion Research]

What (probably) never was

I wasn't planning to post again this soon, but this morning I read a review of The Populist Explosion (John P. Judis) in the NY Times Book Review.  The reviewer (Jonathan Alter) said "In 1968, Wallace, running as a third-party candidate, led the Democratic nominee, Hubert Humphrey, in the polls until October...."  I was surprised to read that, so I checked the Roper Center's iPOLL database for all surveys asking about choice among Nixon, Humphrey, and Wallace in 1968.  The results:

 Wallace was always behind Humphrey.  The smallest gap was in a poll taken at the end of September and beginning of October, when Humphrey had 26% and Wallace 19%, with Nixon at 41%.  The iPOLL database is not completely comprehensive, so maybe there were surveys in which Wallace led.  But I doubt it, given the substantial margin and general stability of the figures (which were from two different polling firms, Gallup and Harris).  Maybe Alter was thinking of 1992, when I think Perot did actually lead in some polls?

     I hadn't realized that Humphrey was generally leading in the polls until August.  The Republican convention took place at the beginning of that month and the Democratic convention took place at the end.  That connects to another issue that I was going to write about, so I might as well do it now.  Back in the primaries, I was thinking about the question of whether nominating an "extreme" candidate would make a difference.  People who say that an extreme candidate will lose votes always include George McGovern as a leading example.  McGovern definitely had a liberal voting record in the senate, but so did some more successful Democratic nominees (John Kerry and Barack Obama).

That led me to wonder if the relevant issue might be a divided convention, rather than "extremism."  The two things tend to go together, because the reason for a divided convention is often an ideological challenge where one or both sides feel strongly enough to fight on until the end.  By divided, I mean one in which there was open conflict that was visible to the TV audience.    I think that the following conventions are generally agreed to meet that standard:  Republicans in 1964 and 1976 (and now 2016), and the Democrats in 1968, 1972, and 1980.  You could argue that the whole idea of a divided vs. united convention doesn't apply before the TV era--the action took place behind the scenes, and multiple ballots were common.  However, the Democratic convention of 1924, which took over 100 ballots and involved a giant rally by the KKK after a motion to condemn them had been narrowly defeated, might deserve to be included.

I added a variable for a divided convention to Fair's model and got the following results:

        Model                               estimate    std err

1.    Fair plus divided (1968-80 only)        -2.36     (1.45)

2.    Fair plus divided (including 1924)      -2.51     (1.37)

3.    Modified plus divided                   -3.41     (1.59)

The third model modifies Fair's variables in two ways:  Gerald Ford is counted as an incumbent president, and the Democrats are counted as getting 34.8% of the two-party vote in 1924, which is what they actually got.  Fair does not count Ford as an incumbent on the grounds that he was not elected to either the presidency or the vice-presidency, which doesn't make sense to me.  There was an important third-party candidate (Robert LaFollette) in 1924, and Fair estimates that most of his voters would have gone to the Democrats if he had not been in the race.  That could be true, but it's not possible to be sure, and if you were adjusting totals it seems that a better case could be made for 1968, when surveys indicated that most of Wallace's voters would have gone to Nixon as a second choice.  

The figures mean that a divided Democratic convention is estimated to cost the Democrats about 2.4% and a divided Republican convention to help them by 2.4% (in model 1).  There's a good deal of uncertainty about the size of the effect, and it's not statistically significant at the 5% level in the first two models.  Still, I'd say that there's evidence that a divided convention hurts a party.

What might have been

Suppose that the major-party nominees for president had been nondescript, people like Mike Pence or Tim Kaine.  In that case, it seems that the outcome would depend on "how things are going"--if they're going well, keep a Democrat in office; if not, turn to the Republican. There have been a number of efforts to figure out exactly which "things" matter and how much difference they make.  One of the best-known ones was proposed by Ray Fair in 1978 and has been modified and updated since then.  In Fair's formula, support for the incumbent party depends (positively) on the rate of growth of per-capita GDP in the first three quarters of the election year and the number of quarters of rapid growth over the president's term and (negatively) on the rate of change in prices over president's term.  Of course, we don't know what those will be yet (currently the GDP data only goes through the first quarter of 2016), but if you put in his predictions of what those will be, the equation predicts the Republican would get 56% of the two-party vote.  Even if you assume very strong (4%) growth in the second and third quarters, that only brings the Republican share down to 52.7%.

So according to his formula, the objective conditions put the Democrats at a big disadvantage.  Why?  Part of the reason is that there have been few quarters of rapid GDP growth over the last term--as many people have pointed out, the economic recovery has been slow.  But that was also true in 2012, when the formula predicted that Barack Obama would win the popular vote.  The main reason for the difference is that the formula also includes variables for incumbency and for the duration of time the party has been in power.  Incumbency is estimated to increase vote share by 3.0%, and being in power eight years rather than four is estimated to reduce it by 3.8%.  Putting those together, the Democratic candidate in 2016 starts off 6.8% worse than Barack Obama did in 2012, and 7.6% worse than Obama did in 2008 (when duration was counting against the Republicans), before accounting for economic conditions.  These are big differences, both in absolute and relative terms (bigger than the difference that would result from a -5% rate of growth compared to a +5%).

When I checked earlier today, the Republican share of the two-party vote in the latest polls was about 47.5.  If that holds up, it will be the largest prediction error in the period covered by the estimates--the current record is 1992, when George HW Bush was predicted to get 53.6% of the two-party vote, but actually received only 48.3%.