Comparing One-Partner and Couple Data on Sensitive Marital Behaviors: The Case of Marital Violence.

摘要: Data on sensitive marital relationship issues such as violence, conflicts, or sexuality are often subject to considerable reporting bias. This bias can affect validity and prevalence estimates, well correlations between the issue under consideration other variables. Couple data render it possible assess some (though certainly not all) effects results. The following analyses focus violence but applicable research topics. Many surveys dealing with behaviors, including 1975 1985 National Family Violence Surveys (NFVS), lack empirical evidence concurrent (Egley, 1991; Straus, 1992a). Studies comparing spouses' responses about have relied small nonrepresentative samples (Jouriles & O'Leary, 1985; Malone, 1992; Szinovacz, 1983) clinical populations (Browning Dutton, 1986; Edleson Brygger, Langhinrichsen-Rohling Vivian, 1994). Results from these studies indicate disagreement spouses whether what types of occurred. Larger community premarital recently married couples revealed moderate (McLaughlin, Leonard, Senchak, O'Leary et al., 1989). Relations one-partner couple predictors appeared be similar. However, this latter finding may due used for studies. Given NFVS' exclusive reliance data, further clarification seems essential. article reassesses basis a nationally representative set, Survey Families Households (NSFH). ONE-PARTNER AND COUPLE DATA One-partner based only one partner. Spouses give valid information partner objective characteristics socioeconomic background history. they usually cannot provide their partner's attitudes perceptions relations interactions, because attributions differ would given (Thompson Walker, 1982). Similar discrepancies occur in socially disapproved behaviors. In contrast, relying obtain same variables both partners interactions partners. thus allow answers questions enable researcher differences ("his" "her" marriage). addition, permit dyadic approach analysis: Relevant attitudinal available analyses. Moreover, necessary measure and/or construct agreement. Finally, theoretical relevance, also serve methodological purposes. estimate underreporting undesirable which associated underreporting, show associations explanatory dependent affected by selective reporting. is uses applied measurement violence. DISCREPANCIES BETWEEN SPOUSES' RESPONSES Generally, derive three sources: random error, systematic relationship. self-reports, error typically caused ambiguities scale items. For example, if item refers several behaviors (e.g., hit throw things at other), spouse answer regard first behavior (hitting), whereas refer (hitting throwing). …