A new essay in the journal The American Sociologist raises important questions about the intersections between research and politics — highlighting the tensions between scholars’ professional responsibilities to understand social phenomena as comprehensively and accurately as possible, and their personal desires to promote particular socio-political outcomes.
Full Reference: al-Gharbi, Musa (2018). “Race and the race for the White House: On social research in the age of Trump.” The American Sociologist. DOI: 10.1007/s12108-018-9373-5
Abstract: As it became clear that Donald Trump had a real base of political support, even as analysts consistently underestimated his electoral prospects, they grew increasingly fascinated with the question of who was supporting him (and why). However, researchers have also tended to hold strong negative opinions about Trump, and have approached research with uncharitable priors about the kind of person who would support him and what they would be motivated by. This essay presents a series of case studies showing how analyses of the roles of race and racism in the 2016 U.S. Presidential Election seem to have been systematically distorted as a result. However, motivated reasoning, confirmation bias, prejudicial study design, and failure to address confounds are not limited to questions about race or Trump. Presented evidence suggests that research with strong adversarial or advocacy orientations may be most susceptible to systemic distortion. Activist scholars and their causes may also be among those most adversely impacted by the resultant erosion of research reliability and credibility. Ultimately, however, these are problems which all social scientists must remain vigilant against, and which we all have a stake in working to address.
Musa al-Gharbi, a sociologist at Columbia University and a research associate with Heterodox Academy, argues:
Some degree of bias is inevitable in social research. We are all shaped by our limited experience, our personal commitments and aspirations, etc. – and these inform social inquiry in obvious and subtle ways. However, in a context where people with diverse and competing commitments, backgrounds and experiences are simultaneously exploring a particular issue, these errors should approach random distribution — they would largely cancel each-other out. “Zones of agreement” should emerge in the process, forming a basis for reliable knowledge. However, in a world where everyone shares the same basic commitments, errors are not randomly distributed. They are far less likely to be recognized as errors and addressed at all – indeed, they are likely to cascade as an integral part of the ‘consensus’ position, even giving rise to entire ‘null fields’ of research.
Among social researchers, one commitment which virtually everyone seems to share is a passionate distaste for Donald Trump, his agenda, and what he seems to represent. As it became clear that Trump was a real contender for the Republican nomination, even as analysts consistently underestimated his prospects — both in the primary and the general election – researchers became transfixed by questions of just who was supporting him and why. These questions are perfectly reasonable and legitimate. The problem is that, as a result of their disgust with Donald Trump, many social researchers also seem to have developed strong assumptions about the type of person who would support him, and what they would be motivated by. These presuppositions have consistently undermined research about the President and his base. The most typical problems include:
- Failure to deal with obvious confounds
- Prejudicial study design
- Strange interpretations of data (apparently to make them fit with researchers’ preferred theses)
- Uncharitable interpretations of the words and actions of Trump and (especially) his supporters – or even outright misrepresentation through selection and editing. (p. 1-2)
Al-Gharbi then drills into a handful of prominent works exploring the role of race and racism in the 2016 U.S. general election — prioritizing those that have had a substantial impact both within the academy and beyond, and which are representative of broader lines of research.
Case Study #1: White Supremacy/White Rage Narratives
The first case-study from this literature responds to the “White Supremacy”/“White Rage” narratives about Trump’s victory — advocated most prominently by Ta-Nehisi Coates (here) and Emory sociologist Carol Anderson (see e.g., here). Al-Gharbi demonstrates that the theory seems to be robustly confounded by the very data it seeks to explain:
- The election did not represent a rejection of Barack Obama as a politician or a symbol: Obama’s popularity remained high throughout the 2016 cycle. In fact, his popularity rose over the course of the campaign, even as the ratings for Trump and Clinton plummeted. A year into Trump’s presidency, Barack Obama remains highly-popular.
- The most decisive votes for turning the election came from districts that went for Barack Obama in 2008 and 2012 but flipped to the Republicans in 2016. It is unclear why these voters, if horrified at the prospect of a black president, would have voted Obama into office to begin with — let alone given him another four years to advance his agenda when he stood for reelection.
- Trump did not spearhead a white uprising: participation rates among whites were roughly equivalent to 2012 and lower than 2008. In fact, whites made up a smaller share of the electorate than they had in previous cycles, while Hispanics and Asians were better represented.
- Overall, Trump won only about 37% of eligible non-Hispanic white voters. 36% abstained, 24% voted for Clinton, and 3% voted for other candidates. In short, it would be consistent with the data that an overwhelming majority of white voters did find Trump’s rhetoric disqualifying.
- Among those who did head to the polls, Trump actually won a lower share of the white vote than Mitt Romney.
- He was nonetheless able to win because he won a larger share of Hispanics and Asians than his predecessor, and won a larger share of the black vote than any Republican since 2004.
Case Study #2: Racism Motivated Trump Voters, More than Authoritarianism
The second case study critiques a widely trafficked and prominently cited essay by Ohio State political scientist Thomas Wood. Comparing endorsement of “symbolic racism” across Democratic and Republican voters from 1988 through 2016, Wood observed that the gap between partisans was larger in 2016 than in any previous cycle — suggesting race likely played an especially important role in this election’s outcome.
Setting aside well-known criticisms of symbolic racism (see e.g., here, here, and here) and authoritarianism (see, e.g., here and here), al-Gharbi demonstrates that the apparent significance of race in the 2016 cycle was actually due to a radical shift among Democratic voters, who were less likely to endorse “racist” sentiments in the 2016 cycle than they were in the entire 28 years of data presented. Indeed, the data show Democrats grew more “racist” when Obama was on the ballot, and then dropped to unprecedented lows in this cycle. This is an interesting phenomenon which the study author failed to meaningfully address at all. Instead, he took an effect driven entirely by Democrats in order to impute negative motives to Trump supporters — despite the fact that his own data showed a decrease in endorsement of “racism” among Republican voters in 2016. That is, according to Wood’s data, Trump voters were actually less racist than Romney voters. Yet the data were contorted in such a way as to suggest that Trump supporters were particularly “racist.”
Beyond these inferential gaps, the aforementioned confound problem also persists: correlations between “racist” attitudes and support for Trump cannot be translated into a plausible causal story unless one can overcome the objections raised in the first case study.
But more broadly, this section criticizes a tendency towards prejudicial study design about the motives of Trump supporters. Most studies in the genre assume that Trump voters were motivated primarily by something negative or even pathological: poverty, fake news, ignorance, anti-minority/ white supremacist sentiments, sexism/misogyny, etc. The studies merely set out to determine which of these bad traits was more explanatory. Meanwhile, he argues, there is little research exploring “the extent to which Clinton voters were motivated by negative impulses (the same ones or others). It is implied that, by contrast, they must have been guided primarily by benign traits such as love for others, patriotism, knowledge of the issues, rational self-interest, etc. (Stanovich, 2017)” (p. 13).
Case Study #3: Ethnographic Safaris into “Trump Country”
Finally, al-Gharbi addresses Berkeley sociologist Arlie Hochschild’s bestselling books Strangers in their Own Land: Anger and Mourning on the American Right. This book was actually published two months prior to the 2016 Presidential Election, but “the perceived insight and prescience of the work helped elevate it into being a New York Times bestseller and national book award finalist” (p. 13). Since the election, others have “gone on safari” in “Trump Country” to better understand those who voted for him (see e.g., here, here, and here), with the hope that achieving such an understanding can help Democrats in the upcoming 2018 and 2020 elections.
Al-Gharbi however, demonstrates that despite being widely touted as having strong explanatory power for understanding the electoral outcome, there is actually little reason to believe that her findings generalize to most Trump voters — and many reasons to doubt this, in fact:
Hochschild conducted forty “core interviews” (Hochschild 2016a, 247-50) — all with white Tea Party supporters, concentrated in one narrow region (Lake Charles) of a single state (Louisiana). This is an extremely small and homogenous sample for explaining why some 63 million people, from all across the country and all walks of life, cast their ballots for Donald Trump. Indeed, as we have previously explored the race was likely decided by people who voted for Obama in the previous election(s) but flipped for Trump in 2016. Few of these voters would have been staunch Tea Party supporters (they may even have very negative views of the Tea Party). And again many of the critical defections were not from whites at all: were it not for Republican gains (and Democratic losses) among blacks, Hispanics, and Asians – Trump likely would have lost, given that he actually won a smaller share of the white vote than Mitt Romney. Yet, there does not seem to be one single county in the state Hochschild studied that flipped from Obama to Trump.
Indeed, by her own reckoning, only about 11% of white Louisiana voters supported Obama in 2012. In 2011, up to 50% of Louisianans supported the Tea Party. The region is described by Hochschild herself as the “geopolitical heart of the right” (p. 11-12). Simply put: these votes were not decisive in the 2016 election – there was no real contest for Louisiana at all. And so, empirically speaking, it is not clear how Hochschild’s project — which basically attempts to understand why whites in a solidly-red state vote Republican — would explain much about why so many former Obama supporters, many of them minorities, voted for Donald Trump over Hillary Clinton in 2016. This is perhaps the most substantive question for understanding why Trump won, and it is not addressed in her book — whatever its other merits.
Additionally, al-Gharbi flags aspects of the work in which Hochschild’s own political agenda seems to distort the study framing, and Hochschild’s interpretations of her data. He concludes:
In a break with Hochschild’s groundbreaking previous work, the goal of Strangers in their Own Land is not so much to understand these people on their own terms, for their own sake – or to shine a light on the challenges they face, and their needs as they understand them. Instead, Hochschild’s Tea Party supporting subjects and their stories are used as a means to political ends — ones which they would never, themselves, endorse: Hochschild wants to stall the ‘rise of the right’; she wants to help progressives understand how to advance their agenda and candidates in red states; she now wants to neutralize the appeal of Donald Trump. Hochschild seems to paternalistically believe that securing these political objectives would benefit her subjects in the end – irrespective of their own preferences. However, this goal seems secondary at best in reading the book. Although she typically refers to subjects as her ‘Tea Party friends’ — in the end they do not seem to be on the same page, and they are not on the same team.
The essay closes with a discussion of how counterproductive these research tendencies can be, particularly for those who, likely, are the most opposed to Trump.
For one, Trump’s opposition (among which al-Gharbi counts himself, and I do as well) is undermined when they pursue edifying narratives that reinforce poor research feeds into bad political strategies:
Throughout the 2016 cycle, al-Gharbi argued that Trump was likely to win, and that Democrats’ strategy was poorly calibrated — both because they did not seem to take the threat Trump posed seriously enough, and also because their rhetoric seemed likely to alienate key voting constituencies (through a phenomenon he dubbed “negative intersectionality”). Since the election, he has argued that Trump stands a better chance of winning reelection in 2020 than most analysts and pundits seem willing to recognize — and that distressingly little seemed to have been learned from Trump’s 2016 “black swan” victory.
However, the stakes extend far beyond the next presidential election: al-Gharbi argues that when research disparages Trump and his supporters on weak evidentiary grounds, the credibility and viability of the broader social research enterprise is called into question as well. Many on the right already view the humanities and social sciences as essentially “partisan propaganda,” he reminds, and have called for defunding social research on these grounds. It is therefore imperative that research about these already-skeptical constituents be as fair-minded and rigorous as possible.
Why is this important?
Whatever one’s opinion of Donald Trump is, it is hard to argue that his election to the Presidency in 2016 was an insignificant event. It is important to have a clear-eyed understanding of how he won for many reasons.
However, social researchers have another major stake in addressing biased research on this topic, related to the continued viability and impact of their fields:
People tend to exert loyalty and commitment to institutions they feel represented in, and to undermine or disinvest in those which they feel alienated from. Because conservatives have no meaningful representation in social research fields, they have been moving to devalue and defund it.
The Trump Administration has suggested defunding the National Endowment for Humanities and the National Endowment for the Arts. Congressional Republicans have tried to strip away National Science Foundation support for social science as well.
However, in addition to currently controlling every branch of the federal government, Republicans control 32 state legislatures nationwide (they hold governorships in 25 of these states too). In these states, moves to cut state support for institutions of higher learning are growing increasingly pronounced. Often lawmakers explicitly refer to the exclusion of conservative perspectives from campus in order to justify the legislation their constituents (who are typically, themselves, Republicans). In many instances where cuts happen, humanities and social science budgets are the first on the chopping block – as STEM fields are perceived as both more immediately useful, but also more “objective” disciplines.
Minorities, low-income and other disadvantaged Americans are disproportionately likely to attend or to teach at state-funded schools (as opposed to elite private schools), thus, as previously noted, such cuts are likely to have a disproportionate impact on the very groups activist scholars ostensibly want to assist.
Social research is unlikely to be respected or funded, let alone adopted or utilized, by Republicans insofar as it is seen as ideologically biased against conservatism. So given that Republicans control all branches of the federal government and most state governments, the impact of social research is severely diminished to the extent that social researchers only engage with the left.
Read the entire article (ungated): al-Gharbi, Musa (2018). “Race and the Race for the White House: On Social Research in the Age of Trump.” The American Sociologist. DOI: 10.1007/s12108-018-9373-5