When surveys were young

I've continued to see mentions of the claim that the Republican vote among "minorities" in 2020 was higher than at any time since 1960.  While doing a search to get a sense of just how common this is, I ran across an article from Forbes by Avik Roy disputing that claim.  It included a figure on the Republican vote among blacks, which I reproduce here:

What caught my eye was the first year, 1932--it says that Republicans got 77% of the black vote, far higher than they got in 1936 or any later year.  The first modern survey (the Gallup Poll) didn't begin until 1935, so where did these data come from?  The note says "Joint Center for Political and Economic Affairs," but I couldn't find the relevant report on their web site.  Admittedly I didn't do a thorough search, since I knew they couldn't be the original collector.  A little more searching on the internet gave this, published in the Washington Examiner:  "In 1932, Franklin D. Roosevelt lost the black vote overwhelmingly, receiving just 21% support from black Americans in Chicago (there were no national polling organizations). . . Yet just four years later, in 1936, the newly established Gallup Poll revealed that FDR had received 76% of the black vote." I found a couple of other references to that survey of Chicago, although none gave any further information, so I'll guess that's the source of the Forbes number too (figuring that third parties got a few percent).  However, even if those numbers for Chicago are reliable, my impression is that there was a lot of local variation in the black vote at that time, so it might have been very different in other cities.  I went back to the Gallup Poll, which regularly asked people how they voted in previous elections.  Unfortunately, they didn't do that in their first few years, so it appears that only one survey asked about 1932.  

The results:

                                   Hoover         FDR         Thomas     Didn't Vote        Too young

Unspecified                    36%           48%            2%              9%                    5%

White                              15%           51%             1%             12%                  21%

"Colored"                        24%          55%             1%             17%                    1%

So black support foe Republican candidate was higher than white support, but considerably lower than support among people whose race wasn't recorded (who were a majority of the sample).  Who were those people?  The early Gallup polls had a variable for economic level, which was coded by inspection:  "average plus," "average," "poor," or "on relief."  Race was recorded only for people who were "poor" or "on relief."  Either they didn't interview blacks who looked like they were middle class, or they classified them as "poor" regardless of their appearance.  I think it was the latter--I seem to remember seeing some early Gallup instructions that said that all black people go in the lower categories.  So we can assume that all the people whose race wasn't recorded are white.  Under that assumption, and limiting it to people who voted for one of the three major candidates in 1932:

                                 Hoover       FDR       Thomas

White                          36%         62%        2%

"Colored"                   30%         68%        2%

That is, in 1932, the black vote was already mostly for the Democrats, although if you control for economic level it was not much different from the white vote.  I should say that there is a good deal of uncertainty, because there were only 71 blacks in the sample, and only 57 of those said they voted in 1932.  The early Gallup polls had big samples (this was about 5500 people), but they didn't try to represent groups by their proportion in the population, but by what they thought was their proportion of the voters.  But even with these qualifications, it's safe to say that Roosevelt got far more than 21% of the black vote in 1932, and probably got more than half.  

By the way, the Washington Examiner op-ed piece I mentioned, which was published in 2019, argued that Trump was poised to make big gains among black voters in 2020.  That was the point of the comparison to the 1930s:  they said that just as Roosevelt had made big gains then, Trump could repeat that accomplishment now.  As I've mentioned before, many conservatives seem to want to believe that the Republicans are a multicultural party, or are on the verge of becoming one.  

[Data from the Roper Center for Public Opinion Research]