October was a rough month for progressives, like sociologist Philip Cohen, who seek to minimize or deny the importance of family structure. New studies highlighting the connection between strong families and child well-being, as well as economic growth, were spotlighted in the New York Times, the Washington Post, and the Los Angeles Times. And then, on Halloween, an anonymous blogger took apart Cohen’s critique of Harvard economist Raj Chetty’s important findings on family structure and economic mobility.
Last year, Chetty and his colleagues sought to answer this question: What “factors are the strongest predictors of upward mobility in multiple variable regressions”?
As I wrote in Slate, the answer they found was this:
Family structure. Of all the factors most predictive of economic mobility in America, one factor clearly stands out in their study: family structure. By their reckoning, when it comes to mobility, “the strongest and most robust predictor is the fraction of children with single parents.”
Cohen did not like this finding one bit and sought to discredit it. He ran a regression focusing on just the 100 largest U.S. communities (or commuting zones [CZ]) in Chetty’s data — not all of Chetty’s data — and claimed that by incorporating the share of black residents in a community, he was able to minimize the effect of family structure on economic mobility in the data. After adding race to the models, according to Cohen, the association between family structure and economic mobility became statistically insignificant. In his words, “Percent Black statistically explains the relationship between single motherhood and intergenerational immobility across U.S. labor markets.” He even took to Twitter to say: “The single mother effect in Chetty is all in the % Black effect he left out.”
Seems like a potentially damning indictment of Chetty’s family structure finding, right?
Wrong. Along comes “Random Critical Analysis” (RCA), an anonymous data blogger who started poking around at Cohen’s analysis. RCA found four particularly noteworthy problems with Cohen’s work:
1. RCA reran the data Cohen had used and found that single motherhood remained a statistically significant, negative predictor of intergenerational mobility in the 100 largest CZs, even after including race in his model.
2. RCA then analyzed all of Chetty’s data and found, miracle of miracles, “percent single-mothers is truly a substantially stronger predictor than percent black (or any of the other covariates I found).” The figures below illustrate the bivariate links, first, between the share of single mothers in a community and mobility and then, second, between percent black and mobility, according to RCA’s analysis of Chetty’s data.
3. RCA also found that race cannot account for all of the link between single motherhood and economic immobility because the single mother–immobility link is actually stronger in communities with fewer African Americans: “Quite contrary to Cohen’s assertion that single-motherhood predicts because it is associated with blackness, I actually find that the association is notably stronger in less-black (usually whiter) counties and commuting zones (CZ).”
4. Finally, RCA found some evidence that the negative association between race and economic mobility may be explained in part by single parenthood, rather than vice versa. “Once we crudely control for single-motherhood, the association between percent black and mobility is vastly reduced,” in RCA’s words.
In the wake of RCA’s devastating takedown of his critique of Chetty, Cohen weakly replied on his website, “I now must admit that I overspoke myself on Twitter.” Apparently, when you’re in denial about the connection between family structure and the health of the American Dream, it can be hard to admit, straight up, that you made a mistake in your online campaign to minimize the empirical links between single motherhood and economic mobility in America.