Benjamin R. Karney’s research while affiliated with University of California, Los Angeles and other places

In relationship science, researchers have generated a wide array of constructs and corresponding self-report measures to characterize, explain, and predict relationship quality – the foremost studied outcome in the field. Collectively, however, the boundaries among these variables remain unclear. In the current research, we examined the extent to which measures of relationship quality and other important relationship constructs are empirically separable from one another. Across two studies of US census-matched participants (total N = 3,439), we applied latent variable techniques (e.g., exploratory bifactor analysis) on broad pools of items representing various prominent relationship-specific constructs. Results revealed robust evidence that a single general factor Q (representing global relationship sentiment) accounts for a vast majority of common variance across distinct relationship measures. Thus, respondents appear to draw primarily on their overall global relationship evaluations when reporting on an array of presumably-distinct relationship facets. This is consistent with a ‘sentiment override’ perspective. Our findings provide novel empirical evidence for a relationship-specific response bias that challenges prevailing assumptions and practice in the field, including the widespread use of self-report methods to capture meaningful aspects of relationship functioning.

Discrepant theoretical perspectives assert that challenges brought on by COVID-19 could either increase or decrease intimate partner aggression (IPA) between spouses. As reliance on retrospective or post-COVID-19 data cannot resolve these competing views, we turn to a sample of established mixed-sex married couples (N = 223) who provided six waves of pre-COVID-19 data and three waves of post-COVID-19 data. We examined whether the onset of COVID-19 affected IPA and whether any such changes were more pronounced for some couples than for others. Replicating prior research, dyadic piecewise regression showed that IPA declined significantly over time prior to COVID-19; these declines continued following the onset of the COVID-19 pandemic. Post-COVID-19 IPA intercepts were reliably lower than pre-COVID-19 intercepts, and the proportion of couples reporting any IPA dropped by 20% after the pandemic onset. Finally, post-COVID-19 IPA intercepts tended to be higher when, prior to the onset of COVID-19, spouses reported more IPA, less relationship satisfaction, and more stress, suggesting continuity rather than disruption in predictors of IPA. Thus, on average, among established mixed-sex married couples, (a) IPA declined over time, including from before to after COVID-19 onset, and (b) post-COVID-19 levels of IPA were predictable from pre-COVID-19 couple functioning.

The development of generative AI that can sustain complex conversations has created a growing market for companion chatbots promising social and emotional connection with their users. In this review and analysis, we apply theoretical tools established by over 50 years of research on relationships to evaluate the extent to which relationships between humans and chatbots fulfill this promise. Indeed, interactions between humans and chatbots do possess some of the characteristic features of close relationships: humans and chatbots can influence each other, and can engage in frequent and diverse conversations over extended periods of time. Chatbots can be responsive to their users in ways that humans perceive to be appropriate and supportive, and such responses can generate feelings of authentic connection, as well as opportunities for self-expansion and growth. Yet AI-powered chatbots cannot yet fully replicate the experience of being in a relationship with another human being. Their responses can lack depth and their ability to recall and build upon prior interactions remains limited, rendering them predictable and boring to some users. Because chatbots do not make more than the most superficial requests of their users, relationships with chatbots do not provide the many benefits of having to negotiate, compromise, and sacrifice for a partner. Although the relationships humans form with chatbots are clearly meaningful to some users, and will be preferable to social isolation for some, they have not reached the level where they can replace human relationships.

In relationship science, researchers have generated a wide array of constructs and corresponding self-report measures to characterize, explain, and predict relationship quality – the foremost studied outcome in the field. Collectively, however, the boundaries among these variables remain unclear. In the current research, we examined the extent to which measures of relationship quality and other important relationship constructs are empirically separable from one another. Across two studies of US census-matched participants (total N = 3,439), we applied latent variable techniques (e.g., exploratory bifactor analysis) on broad pools of items representing various prominent relationship-specific constructs. Results revealed robust evidence that a single general factor Q (representing global relationship sentiment) accounts for a vast majority of common variance across distinct relationship measures. Thus, respondents appear to draw primarily on their overall global relationship evaluations when reporting on an array of presumably-distinct relationship facets. This is consistent with a ‘sentiment override’ perspective. Our findings provide novel empirical evidence for a relationship-specific response bias that challenges prevailing assumptions and practice in the field, including the widespread use of self-report methods to capture meaningful aspects of relationship functioning.

As adults embark upon marriage, forming their own new nuclear families, they often remain connected to their families of origin. Nevertheless, spouses’ relationships with their parents and in-laws are likely to change across the early years of marriage. Given that empirical studies to date are typically limited to only a couple of measurement occasions, however, the stability and impact of such changes are unclear. Using Repeated Measures Latent Class Analyses, the present study addresses this gap by identifying classes of spouses (N=375 dyads) experiencing different types of relationships with their parents and in-laws across the first 9 years of marriage. Further, correlates and consequences of spouses’ longitudinal sentiment profiles were also examined. Three sentiment profiles (i.e., Mostly Helpful, Mostly Ambivalent, Mostly Indifferent, Mostly Difficult) emerged when looking at spouse’s sentiments toward their own parents and for husbands’ sentiments towards their in-laws, whereas 4 profiles (i.e., Mostly Helpful, Mostly Ambivalent, Mostly Indifferent, Mostly Difficult) were found for wives’ sentiments towards their in-laws. Mostly Ambivalent sentiment profiles showed the least stability, and spouses showed significantly higher odds of being in the Mostly Ambivalent class if they were living in a multigenerational household and if the wife was a 2nd generation immigrant. Spouses whose marriages dissolved showed significantly higher odds of being in the Mostly Indifferent, Mostly Ambivalent, and Mostly Difficult classes, compared to the Mostly Helpful class. These results provide an important starting point for future research aiming to understand how intergenerational bonds may promote or hinder successful marriages.

The social networks surrounding intimate couples provide them with bonding and bridging social capital and have been theorized to be associated with their well-being and relationship quality. These networks are multidimensional, featuring compositional (e.g., the proportion of family members vs. friends) and structural characteristics (e.g., density, degree of overlap between spouses’ networks). Most previous studies of couple networks are based on partners’ global ratings of their network characteristics or network data collected from one member of the dyad. This study presents the analysis of “duocentric networks" or the combined personal networks of both members of a couple, collected from 207 mixed-sex newlywed couples living in low-income neighborhoods of Harris County, TX. We conducted a pattern-centric analysis of compositional and structural features to identify distinct types of couple networks. We identified five qualitatively distinct network types (wife family-focused, husband family-focused, shared friends, wife friend-focused, and extremely disconnected). Couples’ network types were associated with the quality of the relationships between couples and their network contacts (e.g., emotional support) but not with the quality of the couples’ relationship with each other. We argue that duocentric networks provide appropriate data for measuring bonding and bridging capital in couple networks.

Although prominent theories of intimate relationships, and couples themselves, often conceive of relationships as fluctuating widely in their degree of closeness, longitudinal studies generally describe partners’ satisfaction as stable and continuous or as steadily declining over time. The increasing use of group-based trajectory models (GBTMs) to identify distinct classes of change has reinforced this characterization, but these models fail to account for individual differences within classes and within-person variability across classes and may thus misrepresent how couples’ satisfaction changes. The goal of the current analyses was to determine whether accounting for these additional sources of variance through growth mixture models (GMMs) alters characterizations of satisfaction changes over time. Applied to longitudinal data from 12 independent studies of first-married couples (combined N = 1,249 couples), GMMs that allowed for class-specific individual differences and within-person variability fit the data better than the GBTMs that constrained these to be equal across classes. Most notably, considerable within-person variability was evident within each class, consistent with the idea that spouses do indeed fluctuate in their satisfaction. Spouses who dissolved their marriages were 3.8–5.7 times more likely to be in classes characterized by greater volatility in satisfaction. Because the early years of marriage appear to be characterized by within-person fluctuations in satisfaction, time-varying correlates of these fluctuations are likely to be at least as important as time-invariant correlates in explaining why some marriages thrive where others falter.

Since the onset of COVID-19, a rise in loneliness has raised concerns about the social impact of lockdowns and distancing mandates. Yet, to date, the effects of the pandemic on social networks have been studied only indirectly. To evaluate how the pandemic affected social networks, the current analyses analyzed five waves of detailed social network interviews conducted before and during the first 18 months of the pandemic in a sample especially vulnerable to contracting the virus: mostly non-White couples (243 husbands and 250 wives) recruited from lower income neighborhoods. Pre-COVID interviews asked spouses to name 24 individuals with whom they interact regularly. Post-COVID interviews indicated a nearly 50% decline in face-to-face interactions and a nearly 40% decline in virtual interactions, with little recovery over the first 18 months of the pandemic. Compared with less affluent couples, those with higher incomes maintained more of their network relationships, especially when virtual interactions were taken into account.