The logistic regression carried out earlier and reported in table Al shows that the coefficient for number of moves since grade 5 is 10 times its standard error, the variable with the strongest overall effect of any variable in the equation, including the measures of human and financial capital in the family (socioeconomic status) and the crude measures of family social capital introduced in the earlier analysis. Translating this into an effect on dropping out gives 11.8% as the dropout rate if the family has not moved, 16.7% if it has moved once, and 23.1% if it has

QUESTION: Why does moving so strongly predict dropout even controlling for socioeconomic status? Is it disruption of social capital, or an unmeasured confounder? Builds on: "Frequent family moves disrupt social capital and harm outcomes."

Coleman, SocialCapitalCreation-1988, loc. 179