Comment on Knodel’s “starting, stopping, and spacing during the early stages of fertility transition”

Abstract

the behavioral changes that accompanied historical fertility declines. One of the central questions in this discussion has been the relative importance of two alternative, and perhaps complimentary, fertility-limiting behaviors. The first of these behaviors, "stopping," refers to the attempted truncation or termination of childbearing after some desired family size has been achieved. Until recent times both theories and methods of historical demographic analysis have implicitly assumed that the only significant behavioral change resulting in early fertility declines was the widespread onset of stopping behavior. A second behavior that may be used to limit achieved fertility is the intentional lengthening of interbirth intervals or delay during marital childbearing, that is, "spacing." In at least some fertility declines (e.g., American), lengthened interbirth intervals appear to have been significant. The possible significance of historical birth spacing poses both theoretical and methodological problems. Theoretically, spacing challenges the model of discontinuous decision making in which a couple's fertility is unaffected by the surrounding circumstances and context until a desired limit is reached. Addressing the possible lengthening of interbirth intervals presents a more complicated, and perhaps more realistic, portrayal of decision making that is responsive to both the immediate circumstances and speculations concerning future circumstances. The challenge that spacing behavior poses to traditional methods of historical fertility analysis is of equal import. It is perhaps natural that spacing should receive greater attention in studies influenced by the more recent focus on life-course patterns, the increasing availability of historical life-course data, and contemporary methods of event-history analysis. The larger body of historical fertility research, however, has relied heavily on indices and measurements that implicitly assume that only stopping behavior contributes to fertility "limitation" [e.g., Coale and Trussell's M and m (1974, 1978)]. Knodel, presumably seeking to avoid the inherent limitation of such indices relies largely on age at last birth and McDonald's (1984) supposed decomposition of spacing and stopping effects. Although there are broad theoretical and methodological issues surrounding spacing behavior, this comment is confined to an illustration that neither age at last birth nor McDonald's decomposition adequately assesses the presence of birth spacing. Knodel employed age at last birth as an indicator of stopping behavior, but he also noted that increases in interconfinement intervals may be anticipated to increase open intervals and thereby lower the age at last birth. The effect is demonstrated in Figure 1. The parity-specific mean birth intervals for women with both 6 and 8 children ever born from the 1860 birth cohort of once-married women in Utah (Anderton and Bean, 1985) is used for illustration. The plot presents the observed mean birth intervals by parity (labeled CEB 6 and CEB 8) and predictions of the following birth intervals at the next higher parity should