In the target article, the authors present results of existing published research to support their broad theoretical claim that romantic relationships matter more to men than women. The rationale for including in their review only research with cisgender men and women in opposite sex romantic relationships was that (1) they were interested in the comparison between the outcomes of men and women, and (2) not enough research currently exists with participants of other sexual identities and relationship configurations. In this commentary, I suggest that a methodological innovation introduced by West, Popp, and Kenny (Reference West, Popp and Kenny2008) stresses the need for the inclusion of people in gay and lesbian romantic relationships (point 2 above) to properly test for sex differences in romantic relationships (point 1 above).
West et al. (Reference West, Popp and Kenny2008) demonstrated that when the goal of research is to test sex differences in the romantic relationship context, data should be obtained from both partners in opposite-sex as well as same-sex (gay and lesbian) relationships. Using one of their examples for illustration, research with romantic couples with partners of the opposite sex has shown that women tend to idealize their partners to a greater degree than their male partners idealize them (Murray, Holmes, & Griffin, Reference Murray, Holmes and Griffin1996). At first glance, there seems to be a sex difference with respect to partner idealization. However, it is not known if the difference reported by Murray et al. (Reference Murray, Holmes and Griffin1996) stems from women idealizing romantic partners more (i.e., something inherent to women) or that men tend to be the target of idealization to a greater degree (i.e., something inherent about being in relationships with men). With data from only opposite-sex couples, it is not possible to tease apart these possibilities, as there are no proper reference groups.
More specifically, when testing sex differences, West et al. (Reference West, Popp and Kenny2008) pointed out that there are actually three effects to consider: (1) the effect of the sex of respondent – do men report, for example, higher scores relative to women?, (2) the effect of the sex of the respondent’s partner – do people with women as partners report, for example, higher scores relative to people with men as partners?, and (3) an interaction effect – do men report, for example, higher scores only when they have a woman as a partner? To tease apart these possibilities, with respect to study design, researchers need to go beyond collecting dyadic data and collect dyadic data from partners in gay and lesbian relationships as well. With respect to statistical models to tease apart these three effects, researchers need to go beyond comparing mean differences across opposite-sex, gay, and lesbian couples (the three-group method according to West et al., Reference West, Popp and Kenny2008) and adopt what they termed a factorial method. The factorial method is an extension of the traditional Actor-Partner Interdependence Model (APIM; Campbell & Kashy, Reference Campbell and Kashy2002; Kenny, Kashy, & Cook, Reference Kenny, Kashy and Cook2006) and can be easily estimated via Hierarchical Linear Modeling (HLM; West et al., Reference West, Popp and Kenny2008). Data from each partner from all dyads is structured such that predictor variables are linked with one’s own (actor) as well as the partner’s (partner effect) outcome variables. Analyses of the data are organized such that, for example, the effects of the respondent’s sex, partner’s sex, and the interaction between the two can be estimated.
With respect to the notion that romantic relationships matter more to men than women, imagine a study focusing on a particular benefit from being in a relationship (e.g., generic well-being). The sample for this hypothetical study includes responses from both partners of straight/opposite-sex, gay, and lesbian couples. A main effect of respondent sex may show that men report higher well-being scores compared to women. A main effect of partner sex may show that people in relationships with women report higher well-being scores compared to people in relationships with men. The conclusions of the target article suggest that an interaction should also emerge, but there are a few possible outcomes that could be consistent with the theorizing of the target article. First, the interaction effect could focus more on the sex of the respondent, with men reporting higher well-being regardless of the sex of their partner (i.e., straight and gay men in relationships benefitting the most). Second, the interaction effect could focus more on the sex of the partner, with people in relationships with women reporting higher well-being scores regardless of the sex of the respondent (i.e., straight men and lesbian women in relationships with women benefit the most). Third, the interaction effect may suggest that only men in relationships with women benefit the most. Analytically, these three possibilities represent different patterns of statistical significance and direction of model coefficients. But theoretically, these three possibilities differ in important ways.
Given this line of reasoning, I therefore view the conclusion of the target article, supported by existing research that admittedly has not included gay and lesbian couples for the purpose of teasing apart the sex differences in how relationships matter more to men than women, as preliminary and suggestive. That said, the strength of the arguments in the target article warrants a program of future research that is tailored to evaluate the notion that romantic relationships differentially benefit men relative to women.
You have
Access
In the target article, the authors present results of existing published research to support their broad theoretical claim that romantic relationships matter more to men than women. The rationale for including in their review only research with cisgender men and women in opposite sex romantic relationships was that (1) they were interested in the comparison between the outcomes of men and women, and (2) not enough research currently exists with participants of other sexual identities and relationship configurations. In this commentary, I suggest that a methodological innovation introduced by West, Popp, and Kenny (Reference West, Popp and Kenny2008) stresses the need for the inclusion of people in gay and lesbian romantic relationships (point 2 above) to properly test for sex differences in romantic relationships (point 1 above).
West et al. (Reference West, Popp and Kenny2008) demonstrated that when the goal of research is to test sex differences in the romantic relationship context, data should be obtained from both partners in opposite-sex as well as same-sex (gay and lesbian) relationships. Using one of their examples for illustration, research with romantic couples with partners of the opposite sex has shown that women tend to idealize their partners to a greater degree than their male partners idealize them (Murray, Holmes, & Griffin, Reference Murray, Holmes and Griffin1996). At first glance, there seems to be a sex difference with respect to partner idealization. However, it is not known if the difference reported by Murray et al. (Reference Murray, Holmes and Griffin1996) stems from women idealizing romantic partners more (i.e., something inherent to women) or that men tend to be the target of idealization to a greater degree (i.e., something inherent about being in relationships with men). With data from only opposite-sex couples, it is not possible to tease apart these possibilities, as there are no proper reference groups.
More specifically, when testing sex differences, West et al. (Reference West, Popp and Kenny2008) pointed out that there are actually three effects to consider: (1) the effect of the sex of respondent – do men report, for example, higher scores relative to women?, (2) the effect of the sex of the respondent’s partner – do people with women as partners report, for example, higher scores relative to people with men as partners?, and (3) an interaction effect – do men report, for example, higher scores only when they have a woman as a partner? To tease apart these possibilities, with respect to study design, researchers need to go beyond collecting dyadic data and collect dyadic data from partners in gay and lesbian relationships as well. With respect to statistical models to tease apart these three effects, researchers need to go beyond comparing mean differences across opposite-sex, gay, and lesbian couples (the three-group method according to West et al., Reference West, Popp and Kenny2008) and adopt what they termed a factorial method. The factorial method is an extension of the traditional Actor-Partner Interdependence Model (APIM; Campbell & Kashy, Reference Campbell and Kashy2002; Kenny, Kashy, & Cook, Reference Kenny, Kashy and Cook2006) and can be easily estimated via Hierarchical Linear Modeling (HLM; West et al., Reference West, Popp and Kenny2008). Data from each partner from all dyads is structured such that predictor variables are linked with one’s own (actor) as well as the partner’s (partner effect) outcome variables. Analyses of the data are organized such that, for example, the effects of the respondent’s sex, partner’s sex, and the interaction between the two can be estimated.
With respect to the notion that romantic relationships matter more to men than women, imagine a study focusing on a particular benefit from being in a relationship (e.g., generic well-being). The sample for this hypothetical study includes responses from both partners of straight/opposite-sex, gay, and lesbian couples. A main effect of respondent sex may show that men report higher well-being scores compared to women. A main effect of partner sex may show that people in relationships with women report higher well-being scores compared to people in relationships with men. The conclusions of the target article suggest that an interaction should also emerge, but there are a few possible outcomes that could be consistent with the theorizing of the target article. First, the interaction effect could focus more on the sex of the respondent, with men reporting higher well-being regardless of the sex of their partner (i.e., straight and gay men in relationships benefitting the most). Second, the interaction effect could focus more on the sex of the partner, with people in relationships with women reporting higher well-being scores regardless of the sex of the respondent (i.e., straight men and lesbian women in relationships with women benefit the most). Third, the interaction effect may suggest that only men in relationships with women benefit the most. Analytically, these three possibilities represent different patterns of statistical significance and direction of model coefficients. But theoretically, these three possibilities differ in important ways.
Given this line of reasoning, I therefore view the conclusion of the target article, supported by existing research that admittedly has not included gay and lesbian couples for the purpose of teasing apart the sex differences in how relationships matter more to men than women, as preliminary and suggestive. That said, the strength of the arguments in the target article warrants a program of future research that is tailored to evaluate the notion that romantic relationships differentially benefit men relative to women.
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Competing interests
The second author of the target article was my graduate school supervisor at Texas A&M University from 1996 to 2001.