Key problems in social stratification hinge on how the duration of exposure to social contacts affects individuals' outcomes, i.e., on overlap effects. In the first chapter, I conceptualize the overlap effect as the cumulative effect of an individual’s observed and unobserved characteristics over the duration of overlap. I develop a formal model for grandparent overlap effects on grandchild test scores, which encapsulates this conceptualization, and in which the overlap effect is (a) confounded by, and (b) varies with individuals’ observed and unobserved characteristics. Conventional methods would fail to identify overlap effects because the effect of unobserved confounders interacts with overlap and constitutes the overlap effect. I show that the overlap effect is identifiable and develop a new cumulative fixed effects (CFE) estimator to recover various overlap estimands. I demonstrate the unbiasedness of this estimator by simulation. In the second chapter, I study the grandparent overlap effect on children’s language test scores using the Danish register data. I find positive grandparent overlap effects on children’s language tests for all grandparent lineages with a new method, CFE. While conventional fixed effects would underestimate the grandparent overlap effects, OLS regression would overestimate such effects. The effects of grandparent overlap effect are also heterogeneous across grandparent lineages, class, and health. Given the rising inequality in mortality and the effect heterogeneity of overlap for different social groups, grandparent overlap may amplify the multigenerational transmission of social inequality. The third chapter explores the intergenerational exchanges of support in stepfamilies; how parents and adult children respond to each other’s past signals of support with monetary transfer, contact and interpersonal help depending on the children’s step/biological status. I propose and compare four hypotheses regarding how different motivations of help and norms play out in biological kin vs. step kin interactions. Using the HRS data, I find that in spite of step kin’s lower levels of support when no signals of past help are present, step kin responds more sensitively to each other’s signals of help with a larger increase in parents’ monetary transfer, contact, and children’s senior care support, converging to the levels of exchange between biological kin.