Misreading Polling Data on Race

Caught up as we still are in the post-Floyd hysteria, the nation is seemingly transfixed on rooting out “systemic racism,” whatever and wherever that is. And with both the upcoming presidential election and a Black Lives Matter-like Proposition 16 to revive racial preferences on the November California ballot, accurate polling and survey data on racial attitudes is in demand and will be increasingly so.

Opinion surveys on racial issues, however, often obfuscate and confuse more than they clarify because their findings are notoriously susceptible to variations in what opinion is being sought and how terms are defined. Given the plethora of polls with often conflicting conclusions, it is thus necessary to compare, very carefully, the construction and conclusions of competing polls. That is precisely what Nate Silver’s well-known and highly regarded website, fivethirtyeight.com, purports to do. “Silver has established himself,” according to a biographical blurb at an MIT conference, “as today’s leading statistician through his innovative analyses of political polling.”

FiveThirtyEight, as it is known, has just turned its talents to analyzing a host of surveys of American racial attitudes. “We focused,” the authors write,

on areas of American life where there is documented racial inequality. We then searched for polling on those issues. Our aim was to find the most recent polling available, in part to see whether views on major issues had changed in the wake of Floyd’s death, but for many issues, we had to rely on older polling, conducted before Floyd was killed. We found results in four major areas: income inequality, education, housing and the workplace.

Unfortunately, FiveThirtyEight’s analysis of this racial data falls flat. It attempts to clarify apparently conflicting findings in at least two crucial areas of the Proposition 16 debate: affirmative action in college admissions and in the workplace.

College Admissions

Citing a 2019 Pew Survey finding that “most Americans say colleges should not consider race or ethnicity in admissions,” FiveThirtyEight noted—under a heading, “How white and Black Democrats view policies aimed at integrating schools”—that considering race or ethnicity was supported by 34% of white Democrats and 39% of black Democrats.






Considering race and ethnicity in college admissions decisions

Pew Research




Those results, however, do not match Pew’s findings at the link FiveThirtyEight cited, which reports that only 26% of respondents thought race or ethnicity should be considered (7% as a major factor and 19% as a minor factor). Moreover, Pew breaks down those results by race and party but not, so far as I could determine, distinguishing white from black Democrats.

Here is what Pew found: by race—22% of whites think race or ethnicity should be major or minor factor and 78% oppose; 38% of blacks think they should be a major or minor factor and 62% oppose. By party—16% of Republicans and 36% of Democrats think race or ethnicity should be a major or minor factor in admissions; 85% of Republicans and 63% of Democrats oppose their being considered.

One would expect that the surprisingly low levels of support, among blacks as well as whites, of the race preference affirmative action that is ubiquitous would require some comment, but all FiveThirtyEight says is “Notably, both Black and white Democrats are wary of race and ethnicity being factors in college admissions decisions.” Wary hardly seems an apt description of Pew’s finding that 73% of all Americans oppose the use of race or ethnicity, including 78% of whites and 62% of blacks.

The Workplace

Under a heading of “How white and Black Democrats view policies aimed at increasing workplace diversity,” FiveThirtyEight lists three survey results—two from Pew and one from Gallup.








Race should be considered in hiring and promotions

Pew Research





Affirmative action






Companies should promote racial diversity

Pew Research





“What’s going on here?” FiveThirtyEight asks, flummoxed by these seemingly contradictory results.

According to polls, both white and Black Democrats overwhelmingly want racially diverse workplaces [citing Pew] and, when generally defined, support affirmative action [citing Gallup]. But while both white and Black Democrats value workplace diversity and recognize unfairness in employment situations, neither group thinks race and ethnicity should be taken into account when making decisions about promotions or hiring, even though the objective is to increase workplace diversity. That view in some ways contradicts Black and white Democrats’ support of affirmative action and a racially diverse workplace.

FiveThirtyEight’s attempt to explain what it sees as contradictory results is feeble, half-hearted, and almost humorous:

Considering the long history of racial discrimination in employment, it’s likely that Black Democrats are worried that factoring race into the job application or promotion process would hurt them, even if that racial factor is supposed to benefit them. Alternatively, Black Democrats and their white counterparts might be hoping that diversity can be achieved without the direct consideration of race at the individual level, therefore explaining their wariness about considering race in hiring and promotions, as well as in college admissions.

Again, “wariness” is an odd way to describe the overwhelming opposition to taking race and ethnicity into account.

There is, of course, no contradiction whatsoever in these findings that needs to be explained. Respondents are simply answering different questions. When Pew asked how important respondents think it is “for companies and organizations to promote racial and ethnic diversity in their workplace,” 75% of all adults said very important or somewhat important (whites 73%, blacks 81%; Republicans 61%, Democrats 86%). But when Pew asked whether companies and organizations “should only take qualifications in account, even if it means less diversity” or should “also take race and ethnicity into account in order to increase diversity,” 74% of all adults (whites 78%, blacks 54%) said qualifications only, compared to only 24% (whites 21%, blacks 37%) who chose diversity.

Once again, FiveThirtyEight’s numbers listed in its chart above do not match the numbers that Pew published, nor are Pew’s numbers broken down by white Democrats versus black Democrats.

The story at Gallup is the same. Its question (shown on the topline linked at the bottom of the article FiveThirtyEight linked) was, “Do you generally favor or oppose affirmative action programs for racial minorities?” It found that 61% favor (whites 57%, blacks 72%) and 30% oppose (whites 34%, blacks 21%). As with Pew, these numbers are different from those listed by FiveThirtyEight in its chart.

The reason these results are not contradictory is that “diversity” and “affirmative action” are blank slate, feel good terms meaning in essence that one wants to be nice to minorities. As used by Gallup here, “affirmative action” was not “generally defined,” as indicated by FiveThirtyEight. It wasn’t defined at all. Consistently, Americans approve of “affirmative action” when it is not defined, and disapprove of it when it is.

Gallup knows this perfectly well. In late October 2018 (less than a month before the Gallup survey FiveThirtyEight included in its analysis Frank Newport, then Gallup’s editor in chief, described a survey on “The Harvard Affirmative Action Case and Public Opinion” that Gallup had just completed. “When specifics are outlined for respondents,” Newport wrote, summarizing numerous Gallup surveys, “support [for affirmative action] drops.”

Proving his point conclusively, Newport then quoted a question Gallup asked four times between 2003 and 2016:

Which comes closer to your view about evaluating students for admission into a college or university: applicants should be admitted solely on the basis of merit, even if that results in few minority students being admitted (or) an applicant’s racial and ethnic background should be considered to help promote diversity on college campuses, even if that means admitting some minority students who otherwise would not be admitted?

“Each of the four times Gallup has asked this question over a 13-year time period,” Newport emphasized, “between 67% and 70% of Americans chose the ‘solely on merit’ option.”

Perhaps those heretofore consistent opinions have been cancelled under the recent onslaught from Black Lives Matter and its allies among Democratic office holders, in academia, the press, and the corporations, but there is no evidence for that here. We shall see.

FiveThirtyEight has developed an enviable reputation as the pre-eminent parser of polls, relied on as the authority on what polls mean by pundits and journalists everywhere. But unfortunately its recent extensive survey of racial attitudes is disappointing—at least in its analysis of “diversity” and “affirmative action” in college admissions and the workplace.

Photo by Elliott Stallion on Unsplash

John S. Rosenberg

John S. Rosenberg

John Rosenberg blogs at Discriminations.

One thought on “Misreading Polling Data on Race”

  1. The other problem in turbulent times is what I refer to as “safety variance” — people will give the safe answer rather the honest one. For example, a lot of people who intend to vote for Trump are not going to tell anyone about it — they don’t want the hassle.

    They don’t want to be yelled at, they don’t want to be physically assaulted, they don’t want their property destroyed, they don’t want to lose their jobs — so they give the politically correct answers.

    At which point the data becomes bogus.

Leave a Reply

Your email address will not be published. Required fields are marked *