Sampling, Experimental Control, and Generalizability in the Study of Marital Process Models

Abstract

The Oregon Youth Study has yielded invaluable insights into the adult antisocial lifestyle (Shortt, Capaldi, Dishion, Bank, & Owen, 2003), deviant interpersonal processes (Dishion, Nelson, Winter, & Bullock, 2004), psychological and physical aggression within romantic couples (Capaldi & Crosby, 1997; Capaldi, Shortt, & Crosby, 2003), and even the influence of male peer groups on aggressive behaviors directed at romantic partners (Capaldi, Dishion, Stoolmiller, & Yoerger, 2001). Kirn, Capaldi, and Crosby (this volume) have framed their latest article describing Oregon Youth Study couples as a test of the generalizability of marital process models presented in Gottman, Coan, Carrere, and Swanson (1998, referred to by Kirn et al. as the Newlywed Study). Citing Lykken (1968), Kim et al. remind readers that the gold standard of repeatability for any study is constructive replication, not literal replication, and especially not statistical significance. They emphasize that constructive replication does not refer to the literal duplication of some earlier study's sampling procedures, research methodology, specific measurement approaches, or even specific results but rather refer to the replication of conclusions about a set of theoretical propositions. We wholeheartedly agree with this perspective but also note that the theoretical propositions that formed the basis of the interest of Kim et al. in constructively replicating the process models described in Gottman et al. (1998) in their sample of lower income, at-risk couples were somewhat unclear. Although they describe their report's purpose as being to examine whether the process models described in Gottman et al. would also be predictive in a sample with differing characteristics from that used in the Newlywed Study, they offer little else in the way of specific rationales for their hypotheses. In contrast to Kim et al., we would not expect many of the process models described in the Newlywed Study to generalize to the Oregon Youth Study. On the one hand, this is on the basis of our own experience in studying patterns of affect within lower income, at-risk couples (e.g., Babcock, Waltz, Jacobson, & Gottman, 1993; Coan, Gottman, Babcock, & Jacobson, 1997; Jacobson, Gottman, Gortner, Berns, & Shortt, 1996; Jacobson et al., 1994) but also because of the important differences between married and cohabiting couples, differences that are relatively independent of factors such as income (e.g., Dush, Cohan, & Amato, 2003). On the other hand, we would expect to see higher levels of negative affect reciprocity in the Oregon Youth Study and that if any of the process models described in the Newlywed Study would be found to be of consequence in Oregon Youth Study interactions, it would be the male escalation of negative affect. As it happens, and as we describe in greater detail below, these predictions are consistent with what Kim et al. actually report. In this commentary, we highlight that (a) in contrast to the Newlywed Study, the Oregon Youth Study sought out and utilized a sample intended to describe a unique population, one that is of lower income and at higher risk for antisocial behavior; (b) some of the experimental procedures used in the Oregon Youth Study may limit the generalizability of conclusions that can be drawn from it; and (c) despite these potential difficulties, the findings they report are consistent with other research on at-risk couples, and even to a limited extent the Newlywed Study itself. SAMPLING At-Risk Couples Among the goals of the Newlywed Study was to build an empirically tested process model of marriage from detailed observations of marital interactions in the early stages of normative marital relationships. To be eligible for the Newlywed Study, couples had to have married for the first time within 6 months of participating in the study and they had to be childless. The demographic breakdown of the sample was representative of the greater Seattle area, according to the Seattle City Metropolitan Planning Commission Report (Gottman et al. …