Potential moderators of effects in the actor–partner interdependence model (APIM) include variables that vary within dyads, between dyads, or both between and within dyads (i.e., mixed moderators). Another factor in the moderation of the APIM is whether dyads are indistinguishable (e.g., same‐sex friendship pairs) or distinguishable (e.g., heterosexual couples). For each possibility, what are the potential moderator effects (up to 8), how they might be estimated and tested, and how they can be interpreted are discussed. Submodels are also presented, based on patterns of moderation of the actor and partner effects, which are statistically simpler, more conceptually meaningful, and more powerful in testing moderator effects. Example analyses illustrate the recommended steps involved in an APIM moderation analysis.

We extend the actor-partner interdependence model (APIM), a model originally proposed for the analysis of dyadic data, to the study of groups. We call this extended model the group actor-partner interdependence model or GAPIM. For individual outcomes (e.g., satisfaction with the group), we propose a group composition model with four effects; for group-level outcomes (e.g., group productivity), we propose a model with two effects; and for dyad-level outcomes (e.g., liking of each of the other members of the group), a model with seven effects. For instance, for an individual outcome with gender as the group composition variable the effects are gender of the actor, gender of the other group members, actor similarity in gender to the others in the group, and the others' similarity in gender. For each of these models, we discuss the ways in which different submodels map onto social-psychological processes. We illustrate the GAPIM with two data sets.

A new method for assessing group synchrony is introduced as being potentially useful for objectively determining degree of group cohesiveness or entitativity. The cluster-phase method of Frank and Richardson (2010) was used to analyze movement data from the rocking chair movements of six-member groups who rocked their chairs while seated in a circle facing the center. In some trials group members had no information about others' movements (their eyes were shut) or they had their eyes open and gazed at a marker in the center of the group. As predicted, the group level synchrony measure was able to distinguish between situations where synchrony would have been possible and situations where it would be impossible. Moreover, other aspects of the analysis illustrated how the cluster phase measures can be used to determine the type of patterning of group synchrony, and, when integrated with multi-level modeling, can be used to examine individual-level differences in synchrony and dyadic level synchrony as well.

Research on in-group identification typically focuses on differences in individuals' identification at the individual level of analysis. We take a multilevel approach, examining the emergence of group influence on identification in newly formed groups. In three studies, multilevel confirmatory factor analysis confirmed two dimensions of identification-self-definition and self-investment (Leach et al., 2008)-at both the individual and the group level. As expected, the group had greater influence on individuals' identification the more group members interacted with each other. This was shown in experiments with varying amounts of real interaction (Study 1), in a longitudinal study of student project groups (Study 2), and in a longitudinal study that experimentally mimicked the development of online communities (Study 3). Together, these studies support a developmental model of identification at the group level that has implications for the understanding of social identity and small-group dynamics.

Members enter groups with different characteristics, for example, gender and ethnicity, and the Group Actor-Partner Interdependence Model (GAPIM) systematically tests several different effects of group composition for a given characteristic. By finding submodels of these effects, the GAPIM allows for empirically testing many theoretically meaningful models of differences within groups. Among the models that can be tested are models of diversity, relational demography, group norms, and contrast. This paper describes the four different steps of a GAPIM analysis and illustrates its application with two datasets. The first is an experimental dataset where gender composition is manipulated by presenting individuals with pictures of group members with whom they presumably would interact. The second dataset is a national sample of churchgoers who are members of different congregations, in which the effects of both a categorical and a continuous composition variable on a member-level outcome are assessed. SPSS and R syntax used for running the GAPIM is provided for each of these examples.

The present research investigated whether having out-group friends serves as a buffer for feeling misunderstood in interracial interactions. Across three experience sampling studies, we found that among ethnic minorities who have few White friends or are not interacting with White friends, daily interracial interactions are associated with feeling less understood. By contrast, we found that among ethnic minorities who have more White friends or are interacting with White friends, the relationship between daily interracial interactions and feeling understood is not significant. We did not find similar results for Whites; that is, having ethnic minority friends did not play a role in the relationship between daily interracial interactions and feeling understood. Together, these studies demonstrate the beneficial effects of intergroup friendships for ethnic minorities.

Little work has examined relationship dissolution or divorce in adoptive parents or same-sex parent couples. The current study examined predictors of relationship dissolution across the first 5 years of parenthood among a sample of heterosexual, lesbian, and gay male adoptive couples. Of the 190 couples in the study, 15 (7.9%) dissolved their relationships during the first 5 years of adoptive parenthood. Specifically, 7 of 57 lesbian couples (12.3%), 1 of 49 gay male couples (2.0%), and 7 of 84 heterosexual couples (8.3%) dissolved their unions. Results of our logistic regression analysis revealed that the odds of relationship dissolution were significantly higher for (a) couples who adopted a non-infant (i.e., older) child); (b) participants who reported feeling less prepared for the adoption, three months post-adoptive placement; and (c) couples in which both partners reported very low, or very high, pre-adoption levels of relationship maintenance behaviors. Findings have implications for adoption professionals seeking to support same-sex and heterosexual prospective adopters, as well as societal debates and policy regarding same-sex relationships and parenting.

The current longitudinal study examined patterns and predictors of parent-reported gender-typed play behavior in adopted boys and girls in lesbian, gay, and heterosexual two-parent families, across early childhood (Mage = 2.82 to 6.06 years). Specifically, using a sample of 181 couples (56 lesbian couples, 48 gay male couples, and 77 heterosexual couples), we examined parent reports of children’s gender-typed play behavior on the Pre-School Activities Inventory (PSAI; Golombok & Rust, 1993) at three time points (mean age = 2.82 years at T1, 3.93 years at T2, and 6.06 years at T3). Family structure variables (i.e., parents’ gender and sexual orientation; children’s gender and sibling status) were included as predictors. At T1, according to parent reports, children in lesbian-parent families had less gender-differentiated behavior (boys were less masculine, girls were less feminine) than children in heterosexual- and gay-parent families, whereas the degree of gender differentiation did not differ between heterosexual- versus gay-parent families. Findings from a Common Fate Growth Model (Ledermann & Macho, 2014) revealed that, regardless of family type, the parent-reported gender-typed behavior of boys, but not girls, significantly changed over time (i.e., boys’ behavior became more masculine). Our findings have implications for researchers who study gender development in children and adolescents, particularly those who are being raised by two mothers or two fathers.