This paper examines a wide variety of forms, and full histories, of family
structure to test: existing theories of family influences and identify need
s for new theories. The focus is on links between childhood family structur
e and both completed schooling and risk of a nonmarital birth. Using a 27-y
ear span of panel (PSID) data for U.S. children, we find that: (a) change i
s stressful, (b) timing during childhood is relevant, (c) adults other than
parents are important, and (d) two more recently studied family structures
(mother-with-grandparent(s) and mother-with-stepfather) do not fit the mol
ds of existing theories. The findings suggest that new theories should cons
ider allocation of resources and reasons people group into family structure
s.