Spill the tea,honey: Gossiping predicts well-being in same- and different-gender couples

Abstract

Despite its negative reputation, gossip is a common conversational behavior that can benefit relationships. However, its potential advantages have not been explored in romantic partnerships. Using a naturalistic observation method (the Electronically Activated Recorder; EAR) we assessed how frequently 76 same- and different-gender couples gossiped with each other, and whether this frequency was associated with self-reported happiness and relationship quality. Actor-Partner Interdependence Models revealed preliminary evidence that romantic partner gossip was positively associated with both happiness and relationship quality. This association held for both same- and different-gender couples. Given that romantic partners are primary sources of conversation and emotional exchanges for many adults, gossip may play a particularly important role in strengthening these bonds. Future research should replicate this finding and explore potential causal mechanisms and specific gossip characteristics that drive these associations, as gossip may be a fruitful avenue to test as a communication strategy for improving well-being in romantic partnerships.

Keywords

Electronically activated recorder (EAR)happiness naturalistic observation romantic relationships relationship satisfaction social interaction

People are inherently motivated to share social information. Over time, this tendency has been ascribed a label with a poor reputation: gossip – the exchange of information about a person who is not physically present or actively involved in a conversation (Baumeister et al., 2004). Despite its disrepute, gossip is common in everyday conversations and is generally not malicious (Robbins & Karan, 2020). Studies examining friend, coworker, and stranger interactions provide evidence that gossiping can facilitate positive relationship qualities between the sharers and receivers of the social information (Bosson et al., 2006). These qualities include greater cooperation, bonding, commonality, and liking (Peters & Kashima, 2007). Even negative gossip can promote trust and greater perceptions of emotional disclosure between gossip sharers (Dores Cruz et al., 2020). These findings suggest that gossiping between romantic partners might confer similar relationship benefits. However, to date, no research has examined romantic partners’ gossip and its associations with individual and relationship well-being. This study addresses this gap, using a naturalistic observation method to sample couples’ everyday gossip with each other, and explores its associations with self-reported happiness and relationship quality.

Couple communication

Romantic partners serve as the primary source of social support for many adults and, for better or worse, the quality of these relationships has consistently been linked to well-being outcomes (Braithwaite & Holt-Lunstad, 2017). Communication behaviors between partners, such as discussion of events and experiences, emotional exchanges, and interaction frequency are important aspects of relationship functioning. These exchanges can function as indicators of support, understanding, and commitment to the partner’s well-being (Johnson et al., 2021). Therefore, gossip may indicate romantic partners’ motivation to and comfort with disclosing their social experiences and thoughts with each other and provide opportunities to seek and receive support.

Relational gender and gossip

When studying gossip between romantic partners, it is important to consider how each partner’s gender might interact to influence gossip frequency. Approaching this topic using relational gender – a term based on the “gender-as-relational” approach (Thomeer et al., 2020) – allows researchers to account for the possibility that women and men communicate differently with their partner depending on that partner’s gender. For instance, men may gossip with a woman partner differently than a man partner. Research on gossip in non-romantic relationships suggests that gender mis/match may be an important factor when predicting how often and with whom people gossip (Robbins & Karan, 2020). These gender-based differences have been attributed to socialized gender norms and values. For instance, women are more likely to gossip about relationships, sexuality, and physical appearance, whereas men are more likely to gossip about achievement-related topics, social status, and wealth (Eckhaus & Ben-Hador, 2019). Thus, same-gender romantic partners may gossip with one another more frequently than different-gender partners due to similar gossip topic interests. Additionally, stereotypes classifying gossip as a behavior characteristic of women (McAndrew, 2014) might motivate men to avoid gossiping with women partners.

There is also evidence suggesting gender differences regarding with whom romantic partners choose to gossip. Self-report research found that men in different-gender (heterosexual) relationships claimed to gossip more with their romantic partner than with their same-gender friends, whereas women gossiped equally with their partner and same-gender friends (McAndrew et al., 2007). This finding mirrors a broader literature showing that men in different-gender marriages often engage in more emotional exchanges with their partner compared to their friends, while women generally engage more evenly with both their partner and same-gender friends (Stronge et al., 2019). Gossip may be a distinct form of communication that varies across relational genders, potentially reflecting differences in how partners interact conversationally. However, none of these past studies have examined gossip behavior dyadically, nor included people in same-gender romantic relationships, leaving these patterns largely unexplored.

Prior research demonstrates that same-gender romantic partnerships have unique strengths that enhance relationship quality, such as more concordant views on autonomy, sexual behaviors, intimacy, health behaviors (Joyner et al., 2019), more appreciation for individual differences, effective communication skills, positive emotions during interactions, commitment, and shared egalitarian ideals (Rostosky & Riggle, 2017). However, little is known about how these strengths play out in day-to-day conversations. Perhaps a tendency to gossip with one’s same-gender partner more frequently contributes to the relationship strengths often observed in same-gender couples.

As an important note, although we use inclusive language (e.g., ‘same-gender’ and ‘different-gender’ couples), we occasionally reference ‘heterosexual’ relationships when summarizing past research that used this terminology, for consistency. We acknowledge that such terms can conflate gender and sexual orientation and may not fully reflect participants’ identities. Further, there is no specific mention in past literature of gossip behavior among people who are gender diverse, to our knowledge, so our reference to ‘different-gender’ refers to couples where one partner identifies as a man and the other as a woman. We acknowledge this language may, and should, evolve as research increasingly includes gender-diverse individuals.

Studying romantic partner gossip using naturalistic observation

Gossip studies overwhelmingly rely on questionnaires, observations of experimentally constructed social interactions, or asking participants to predict their behavior in hypothetical scenarios (Watson, 2012). Consequently, our current knowledge of gossip is limited to what people are willing and able to report, or how they behave in artificial settings. While essential in advancing our understanding of how people perceive their gossiping behavior and how they engage in gossiping in experimentally-controlled social situations, these methods introduce biases. Considering that gossip is perceived as an immoral and socially undesirable behavior (Dores Cruz et al., 2019) and is often subtle or habitual, gossip may be underreported when using retrospective questionnaires. Indeed, studies have found discrepancies between observed and self-reported gossip engagement (Hartung & Renner, 2013), and associations between trait social desirability and reduced reports of gossip engagement (Nevo et al., 1993). The few gossip studies that have used observational methods only captured gossip in specific social settings, such as students in a university lounge (Levin & Arluke, 1985), or did not explicitly study gossip between romantic partners (Robbins & Karan, 2020). Lastly, people generally cannot accurately report their past behaviors or predict future behaviors (Nisbett & Wilson, 1977). Therefore, retrospective gossip measures and methods asking people to anticipate future or hypothetical gossiping could lead to inaccurate estimates. Addressing this topic using naturalistic behavioral observation is needed to obtain a comprehensive understanding of how romantic partners gossip in their everyday lives and whether it is reliably linked to relationship quality and well-being indicators.

The Electronically Activated Recorder (EAR) – a portable electronic device that records ambient sound from the participant’s environment throughout the wearer’s day – is a naturalistic observational measure that provides an ecologically valid account of social interactions. Data collected using this method are less affected by biases associated with social desirability, impression management, retrospection, and setting selection (Mehl et al., 2012); thus, several social psychologists have called for greater use of naturalistic observation methods over the last few decades (Robbins et al., 2021). Participants in EAR studies go about their daily routines and engage in conversations in both public and private settings, allowing researchers to obtain a real-world representation of gossip engagement. Therefore, in addition to being the first study to examine everyday gossip engagement between romantic partners, we also add a rare, naturalistic observation perspective to gossip research.

Present study

The current study uses a naturalistic observation method (the EAR), with a dyadic approach, as a first step examining how frequently same- and different-gender romantic partners gossip with each other. Additionally, we explore links between romantic partner gossip and self-reported happiness and relationship quality. We also account for whether this association differs depending on one’s own gender and the gender of their partner (i.e., relational gender). Data were collected from both partners and analyzed at the individual and couple levels. The following hypotheses and research question were tested:

More frequent, observed romantic partner gossip will be positively associated with participants’ self-reported happiness.

More frequent, observed romantic partner gossip will be positively associated with participants’ self-reported relationship quality.

RQ1

Do the associations between romantic partner gossip and well-being outcomes differ based on relational gender?

Method

Participants

The current study obtained data from both members of self-identified same-gender (i.e., woman in a relationship with a woman, man with a man) and different-gender (i.e., man with a woman) romantic couples. Couples were recruited throughout Southern California via flyers, which were distributed in public places, and through newsletters from the LGBT Community Center of the Desert in Palm Springs, from 2014–2018. Eligible couples were: (1) married or in a self-determined marriage-like relationship, (2) living together for at least 1 year, and (3) had no symptoms of physical or mental health conditions that impeded their daily functioning. These criteria ensured that we studied people in committed romantic relationships who were not coping with health problems that would change their normal activities. Participants were compensated $25 for each weekend they completed the study procedures, which totaled $50 per individual, and $100 per couple over the two-weekend period.

Of the 170 couples that expressed interest in the study, 51 couples (30%) were no longer interested after initial communication or were no longer able to participate, 17 couples (10%) were not living together, 17 couples (10%) did not meet the criteria for being in a marriage-like relationship, and 7 couples (4%) had medical issues that impeded their daily functioning (Robbins, Karan, & Randall, 2021; Robbins, Spahr, & Karan, 2024). Therefore, 78 couples were eligible and participated in the study. The analyses used in this paper include 76 of those couples (N = 152), due to technical difficulties with 2 couples’ audio files. Aside from these exclusions, time and financial constraints associated with recruiting eligible, willing couples were the sole factors influencing the sample size. Table 1 provides demographic characteristics of all participants.

Table 1.

Demographic characteristics of participants.

Sample characteristics	M	SD	Range	N	%
Age (years)	33.04	14.18	18–80	139
Relationship length (years)	7.43	8.26	1–40	139
Gender
Women				81	53.29
Men^a				71	46.71
Relational gender
Woman with a woman				50	32.89
Man with a man				40	26.32
Woman with a man				31	20.39
Man with a woman				31	20.39
Race/Ethnicity^b
Asian				12	7.89
Black				4	2.63
Hawaiian/Pacific Islander				1	0.66
Hispanic/Latinx				49	32.24
Middle eastern				6	3.95
White				77	49.34
Other				8	5.26
Education
Some high school				3	2.16
Graduated high school				8	5.76
Some college				61	43.88
Graduated college				25	17.99
Graduate education				42	30.22
Employment
Unemployed				16	11.51
Unemployed, seeking				10	7.19
Part time				35	25.18
Full time				65	46.76
Retired				12	8.63
Prefer not to answer				1	0.72

^aThree of the men identified as male transgender.

^bFor Race/Ethnicity categories, more than one option could be selected.

Two previous studies used this dataset. The first used data collected using ecological momentary assessments to examine associations between social network size, social interaction quality, and affect (Robbins & Karan, 2020). The second used the EAR data to examine the association between social network size and indicators of well-being, and whether these differed based on the relational gender of the couple (Robbins, Spahr, & Karan, 2024; Robbins, Jonnalagadda, & Spahr, 2024). None of the analyses performed in these studies overlap with the analyses conducted in the present study.

Procedure

Participants were first given detailed verbal information about the study and then indicated their willingness to participate in the study by signing a written consent form. Participants met with the experimenter on two separate Fridays, separated by one month, and were briefed on the naturalistic observation procedures that they would complete over the weekend. Participants were instructed to attach the EAR to their waistline and wear it as often as they were able during their waking hours. In accordance with ethics and California state law, participants were also asked to wear a bystander button with a picture of a microphone and the words “This conversation may be recorded” (Manson & Robbins, 2017). At the end of the specified recording period, participants were given the opportunity to privately review the recorded sound files and delete as many as they wished. They then returned their devices, and the sound files were uploaded to a secure storage platform where they could be accessed by trained research coders. Participants also completed several questionnaires during both the first and second weekend, which included demographics and measures of happiness and relationship quality.

Measures

The EAR

EAR device

The EAR application was installed on an iPod and worn in a protective case around the participants’ waistline (Mehl et al., 2012). It recorded ambient sound from participants’ immediate environments throughout their waking day at preprogrammed time intervals. In the current study, the EAR was programmed to record 50 s of sound every 6 min. The purpose of this far more intensive sampling procedure, compared to typical EAR studies, was for a separate methodological study of how sampling rate affects coded estimates of behavior and social settings (Karan & Robbins, 2017). This yields far more coding work than is necessary for robust estimates of the behaviors of interest (Mehl et al., 2012). Thus, we omitted every third sound file from coding to make the coding process maximally efficient for the project. This resulted in a variable sampling rate, where half the data were sampled with an interval of 6 min, and the other half were sampled with an interval of 12 min and 50 s (see Robbins, Spahr, & Karan, 2024; Robbins, Jonnalagadda, & Spahr, 2024 for more sampling information). The EAR recorded Friday afternoon or evening through Sunday night (i.e., recordings lasted about 2.5 days each, for five days total). The second weekend of monitoring, one month after the first, followed the same procedures. Participants were monitored over the weekend, as opposed to a weekday to maximize the amount of leisure time participants could choose to spend with their partner and others in their social network, and to avoid potential privacy concerns associated with being recorded at work. On average, the EAR recordings yielded a 14% sample of a participant’s 16-h waking day. Less frequent sampling rates (30 s every 12.5 min) have yielded reliable estimates of behavior in previous EAR research (Mehl et al., 2012). In prior studies, the EAR has demonstrated good acceptance and compliance in a variety of populations and is minimally bothersome to participants (Manson & Robbins, 2017). For more information about the EAR method, see the EAR Repository (https://osf.io/n2ufd/; Robbins et al., 2021).

Talking coding

Talking frequency was operationalized as the proportion of total recorded files in which the participant spoke.

Gossip coding

A team of research assistants were recruited and trained in the EAR coding process and given a coding scheme based on the gossip-related variables of interest. Coders were instructed to listen to all sound files in which the participant spoke (talking files), based on previous coding (Robbins, Spahr, & Karan, 2024; Robbins, Jonnalagadda, & Spahr, 2024). The presence or absence of gossip within a sound file was dichotomously coded (1 = present, 0 = absent). If the participant gossiped with any person within a given sound file, research assistants coded this category as present (total gossip). Gossip was defined as any discussion of a person who was not present in the participant’s conversation. To count as gossip, the participant needed to be contributing, and not merely listening, to the gossip. The conversation was not coded as gossip if the participant was discussing an event that happened to themselves, discussing a brand or company (does not qualify as a person), or describing an event that involved a non-specific person (e.g., “somebody robbed that house”).

Next, coders indicated who the participant was gossiping with (1) romantic partner, (2) acquaintance/close other (i.e., acquaintance, friend, family member, or coworker) and (3) stranger. These categories were not mutually exclusive. For instance, participants could be gossiping with their romantic partner and a friend simultaneously. The acquaintance/close other category was inclusive of multiple roles due to the difficulty of distinguishing between them. For example, in the absence of familial titles (e.g., Mom) or referencing a common family member, it is nearly impossible to distinguish a family member from a friend. Participant’s EAR-measured data were aggregated across both weekends, yielding an estimate of the amount of time spent gossiping.

Romantic partner gossip

Romantic partner gossip frequency was operationalized as the proportion of total recorded files over the two-weekend recording period. This included gossiping exclusively with one’s partner and gossiping with one’s partner in addition to another person. When discussing this variable in the dyadic analyses, to avoid confusion between the term “romantic partner” and discussion of “partner effects,” the terms “actor’s gossip with partner” and “partner’s gossip with actor” will be used to refer to actor and partner effects of romantic partner gossip frequency, respectively.

Self-report measures

Subjective happiness

Happiness was assessed using the 4-item Subjective Happiness Scale (SHS; Lyubomirsky & Lepper, 1999). This questionnaire asked participants to report how statements regarding happiness or unhappiness applied to them. For example, “Some people are generally very happy. They enjoy life regardless of what is going on, getting the most out of everything. To what extent does this characterization describe you?” with response options on a 7-point scale (1 = not at all, 7 = a great deal). Composite scores were calculated by averaging item responses.

Participants completed the SHS at two time points – immediately before the first (time (1) and second (time (2) weekends of EAR. This measure demonstrated good internal consistency at time 1 (α = .86) and time 2 (α = .87). Additionally, participants’ SHS composite scores across time points were highly correlated (r = .71, p < .001), therefore, these scores were averaged to create a single index of subjective happiness.

Relationship quality

Relationship quality was measured using the 32-item Dyadic Adjustment Scale (DAS; Spanier, 1976). Participants were asked to report the extent of agreement or disagreement between themselves and their partner and report their perceptions of relationship distress, satisfaction, and partner interactions. The response options varied across questions (e.g., 1 = always agree, 6 = always disagree; 1 = every day, 5 = never). Following the DAS scoring guidelines outlined by Spanier (1976), summary scores were generated by summing all item responses. Summary scores can range from 0 to 151, with higher scores indicating more positive dyadic adjustment.

Participants completed the DAS at time 1 and time 2. This measure demonstrated good internal consistency at time 1 (α = .89) and time 2 (α = .89). Additionally, participants’ DAS summary scores across time points were highly correlated (r = .87, p < .001), therefore, these scores were averaged to create a single index of dyadic adjustment. Person-mean imputation (i.e., using the average of the available item data in place of the missing scores) was used for participants that had missing data. Previous research has found that person-mean imputation methods provide good estimates of measure reliability in cases where the items missing within scales were 20% or less (Bono et al., 2007). Therefore, person-mean imputation was used for participants that responded to at least 80% of the questions included in the DAS. The only DAS participant data that was excluded based on this cutoff was one couple who did not respond to any DAS questions at time 3. Their time 1 DAS data were used in analyses.

Relational gender

Based on participants’ self-reported gender, couples were categorized based on their relational gender. Stemming from the “gender-as-relational” approach which proposes “gender within relationships is shaped by three key factors: own gender, partner’s gender, and the gendered relational context” (p. 220, Thomeer et al., 2020), relational gender accounts for how the gender composition of romantic couples might influence their gossiping behaviors. In this study, participants were categorized into one of four relational, rather than solely individual, genders: a woman in a relationship with a woman (WW), a man in a relationship with a man (MM), a woman in a relationship with a man (WM), or a man in a relationship with a woman (MW).

Extraversion

Extraversion was assessed using the 2-item extraversion subscale of the Ten-Item Personality Inventory (TIPI; Gosling et al., 2003). The TIPI asks participants the extent to which they agree a pair of traits applies to them on a 7-point scale (1 = disagree strongly, 7 = agree strongly). The statements in the extraversion subscale included “extraverted, enthusiastic” and “reserved, quiet” (reverse scored). Composite scores were calculated by averaging responses on the two items, with higher scores representing greater extraversion.

Social network

Two subscales of the Social Network Index (SNI; Cohen et al., 1997) were used to assess the total number of regularly contacted social network members (“total contacts”) and the number of roles (e.g., spouse, children, friends) in which these regularly contacted social network members belonged (“network diversity”). Regular contact was defined as contact that occurs at least once every two weeks. The 12 possible roles included: spouse, parent, child, child-in-law, close relative, close friend, church/temple member, student, employee, neighbor, volunteer, and group member. Total contact was calculated by summing the number of people the participant indicated they regularly contacted across all 12 categories of relationship roles. Participants were given the answer options of 1–7 or more for each category – but could leave the question blank if there were no regular contacts in a particular role – making the possible total contact range 0–84. Network diversity was calculated by summing the number of roles participants indicated having regular contact with, which could range from 0–12. Although these subscale scores were strongly correlated (r = .79, p < .001, CI [.75, .85]), we opted to control for both variables in our models as they could separately influence the association between variables of interest, and calculated the variance inflation factor (VIF) to test for issues with multicollinearity.

Data analyses

All data were analyzed using IBM SPSS Statistics Version 27–29. Before conducting dyadic analyses, we ran correlations to examine the raw associations between the study variables. To account for the dyadic nature of the data (i.e., individuals nested within couples), we used the Actor-Partner Interdependence Model (APIM), which allowed us to test for differences based on gender and relational gender (West et al., 2008). All continuous variables included in APIM analyses were z-scored to yield standardized estimates. We provide descriptives of predictor and outcome variables, that are statistically compared in the APIMs, while accounting for the non-independence in the data. Any differences in gossip engagement between relational genders, as well as their associations with happiness and relationship quality, were accounted for within these APIMs. To ensure that multicollinearity between the primary predictor variables and covariates (i.e., extraversion and social network index variables) would not pose issues during the APIM analyses, we calculated the VIF for all variables (Marcoulides & Raykov, 2019).

Three APIMs were run per dependent variable: happiness and relationship quality (6 models total). Due to the indistinguishable nature of the couples, an extension of APIM, the factorial method, was used (West et al., 2008). The total number of files in which the participant spoke was included as a covariate in our analyses to account for the possibility that talking more frequently overall might drive the link between gossip, happiness, and relationship quality. Participants’ reported number of social contacts and extraversion scores were included as covariates in four APIM analyses, as these variables have been linked to both happiness and relationship quality (Sayehmiri et al., 2020). The first model type used actor gender, partner gender, actor’s gossip with partner, partner’s gossip with actor as the predictors, with actor talking files, partner talking files, and actor age included as covariates. The second model type included additional covariates: extraversion, network diversity, and total social contacts. The third model type included two-way interactions between actor gender and partner gender, actor gender and actor’s gossip with partner, partner gender and actor’s gossip with partner, as well as a three-way interaction between actor gender, partner gender, and actor’s gossip with partner.

We followed the guidelines outlined by Hurlbert and colleagues (2019) to interpret our results, reviewing patterns, effect size magnitudes, and widths of confidence intervals, in addition to considering exact p-values, to judge the quality of evidence. Therefore, our conclusions were largely based on which effects were most likely to be meaningful and robust, rather than the arbitrary threshold of p ≤ .05. Our criterion for deconstructing interactions was p ≤ .20. This cutoff has been recommended in previous literature because it allows for the opportunity to detect differences rather than assuming there are none (Kenny & Ledermann, 2010). When interactions met the threshold, they were deconstructed using Case 1 of the online HLM 2-way and the HLM 3-way interaction utilities at quantpsy.org (Preacher et al., 2006).

Results

Descriptives

Romantic partner gossip

Out of the total number of sound files (98,873) recorded for the 152 participants, 31,440 files (31.80%) included instances of the participant talking. Of those 31,440 talking files, 3,790 (3.83% of all recordings, 12.05% of talking files) included gossip with any person and 2,942 (2.98% of all recordings, 9.36% of talking files) included gossip with one’s romantic partner.

Of the 76 couples, there was only one couple in which neither participant gossiped with their romantic partner. Additionally, there were two couples in which one partner engaged in zero instances of romantic partner gossip, but the other partner did engage in this form of gossip – thus indicating that one partner in these relationships gossiped but the other partner did not reciprocate within the sampled sound files. Descriptively, WW, on average, had the greatest number of talking files and instances of romantic partner gossip (see Table 2).

Table 2.

Descriptives for full sample and relational gender means (SD) for EAR and self-report data.

Measure	Full sample (N = 152)	WW (N = 50)	MM (N = 40)	WM (N = 31)	MW (N = 31)
EAR-based measures
Talking files	.32 (.11)	.37 (.12)	.27 (.10)	.32 (.10)	.32 (.07)
Total gossip	.04 (.03)	.05 (.04)	.03 (.02)	.04 (.02)	.04 (.02)
Romantic partner gossip	.03 (.02)	.04 (.03)	.03 (.02)	.03 (.02)	.02 (.02)
Self-report measures
Happiness	5.53 (1.16)	5.64 (.97)	5.64 (1.29)	5.24 (1.23)	5.53 (1.21)
Relationship quality	118.85 (13.01)	123.13 (11.45)	117.59 (12.32)	115.60 (16.01)	117.13 (11.68)
Extraversion	4.51 (1.69)	4.66 (1.78)	4.11 (1.74)	4.27 (1.59)	5.01 (1.47)
SNI - total contacts	19.01 (8.42)	19.48 (8.12)	16.6 (8.17)	17.19 (8.31)	23.23 (7.96)
SNI - network diversity	6.38 (1.74)	6.71 (1.37)	5.68 (1.80)	5.87 (1.71)	7.29 (1.74)

Note. WW = women with women, MM = men with men, WM = women with men, MW = men with women. Talking, total gossip, and romantic partner gossip variables are the proportion of files out of the total number of EAR files recorded.

Outcome variables: Subjective happiness and relationship quality

As shown in Table 2, participants’ mean happiness score was above the scale’s midpoint and is considered moderately high (range: 2.25–7.00; Lyubomirsky & Lepper, 1999). Based on averages across relational genders, WW and MM self-reported similar levels of happiness, and WM reported the lowest levels of happiness. WW reported the highest relationship quality on average, relative to other relational genders.

Covariates: Extraversion and social network index

As shown in Table 2, participants’ mean extraversion score (range: 1.00–7.00) was above the scale’s midpoint. Extraversion was similar across relational genders, although MW scored slightly higher. Participants’ mean number of total frequent contacts (range: 2–44) and diversity of frequent contacts (range: 2–12) are reported in Table 2. On average, MW had higher scores on the SNI subscales.

Correlations

Correlations between variables are displayed in Table 3. In support of our hypothesis, we observed a pattern of more frequent romantic partner gossip positively associated with happiness (r = .13, p = .114, 95% CI [−.03, .28]) and relationship quality (r = .23, p = .005, 95% CI [.07, .38]).

Table 3.

Correlation coefficients, p-values, and CIs for associations between gossip variables, relationship quality, happiness, extraversion, and social network domains.

	1	2	3	4	5	6	7
1. Talking files
2. Total gossip	.68p < .001[.58, .76]
3. Romantic partner gossip	.65p < .001[.54, .73]	.87p < .001[.83, .91]
4. Happiness	−.003p = .969[−.16, .16]	.14p = .081[−.02, .30]	.13p = .114[−.03, .28]
5. Relationship quality	.21p = .012[.05, .35]	.16p = .050[.00, .31]	.23p = .005[.07, .38]	.38p < .001[.23, .51]
6. Extraversion	.13p = .130[−.04, .29]	.18p = .035[.01, .34]	.15p = .082[−.02, .31]	.07p = .388[−.09, .24]	−.01p = .947[−.17, .16]
7. SNI - total contacts	.06p = .491[−.11, .22]	.08p = .337[−.08, .24]	.04p = .601[−.12, .20]	.08p = .333[−.08, .24]	.12p = .135[−.04, .28]	.27p = .001[.11, .42]
8. SNI - network diversity	.10p = .240[−.06, .25]	.08p = .330[−.08, .24]	.04p = .642[−.12, .20]	.05p = .549[−.11, .21]	.11p = .202[−.06, .26]	.29p < .001[.13, .44]	.79p < .001[.73, .85]

To determine whether controlling for extraversion and SNI variables in our models would lead to multicollinearity issues, we calculated the VIF, and values ranged from 1.12–2.93. These low values suggest that the moderate correlations between predictor variables in our model are not severe enough to indicate that multicollinearity would lead to problems when running the APIMs (Frost, 2017).

APIM hypothesis testing

Gossip and happiness

Table 4 presents the estimates from multilevel APIMs predicting self-reported happiness based on actor and partner gender, covariates, and the interaction between partners’ gender (relational gender), and relational gender interacting with actor gossip, to test how gossip frequency and relational gender predict happiness. The main effects of actors’ gossip with their partner consistently supported our hypothesis that more frequent gossip between romantic partners is positively associated with their own self-reported happiness. This positive association remained when controlling for actor and partner gender, and the number of talking files present for a given participant, (β = 0.33, t = 2.39, p = .019). Moreover, the relationship persisted even after additional controls for extraversion, network diversity, and total social contacts were included (β = 0.36, t = 2.49, p = .014). Finally, the associations remained directionally positive, but with a larger p-value, when two- and three-way interactions with actor and partner gender were accounted for (β = 0.97, t = 0.95, p = .345). None of the interactions met the p ≤ .20 threshold for deconstruction, meaning there was no substantial evidence for relational gender differences for the association between romantic partner gossip and happiness.

Table 4.

APIM hypothesis testing predicting happiness from romantic partner gossip frequency.

DV: Happiness	β	df	t	p	95% CI
Actor gender	0.10	130.71	0.52	.604	−0.27, 0.46
Partner gender	−0.25	132.20	−1.32	.189	−0.61, 0.12
Actor’s gossip with partner	0.33	109.96	2.39	.019	0.06, 0.60
Partner’s gossip with actor	−0.15	108.06	−1.06	.290	−0.43, 0.13
Actor talking files	−0.12	112.52	−0.89	.377	−0.39, 0.15
Partner talking files	−0.01	111.67	−0.10	.923	−0.29, 0.26
Actor age	0.01	82.52	1.62	.109	−0.002, 0.02
Actor gender	0.09	119.35	0.47	.642	−0.30, 0.48
Partner gender	−0.18	120.60	−0.86	.393	−0.58, 0.23
Actor’s gossip with partner	0.36	96.52	2.49	.014	0.07, 0.64
Partner’s gossip with actor	−0.17	94.31	−1.17	.246	−0.46, 0.12
Actor talking files	−0.10	93.12	−0.65	.514	−0.38, 0.19
Partner talking files	−0.03	96.64	−0.18	.856	−0.32, 0.27
Actor age	0.01	71.76	1.70	.093	−0.002, 0.03
Extraversion	0.03	115.31	0.46	.649	−0.09, 0.14
Network diversity	0.04	104.49	0.38	.706	−0.15, 0.22
Total social contacts	0.004	109.89	0.19	.852	−0.03, 0.04
Actor gender	−0.21	59.84	−0.31	.760	−1.59, 1.17
Partner gender	−0.42	61.31	−0.62	.536	−1.77, 0.93
Actor’s gossip with partner	0.97	83.65	0.95	.345	−1.07, 3.02
Partner’s gossip with actor	−0.18	93.50	−1.23	.223	−0.48, 0.11
Actor talking files	−0.10	91.88	−0.64	.522	−0.39, 0.20
Partner talking files	−0.04	94.11	−0.25	.807	−0.34, 0.26
Actor age	0.01	71.06	1.27	.210	−0.01, 0.03
Extraversion	0.02	110.48	0.34	.732	−0.10, 0.14
Network diversity	0.03	101.05	0.28	.782	−0.17, 0.22
Total social contacts	0.01	107.99	0.31	.761	−0.03, 0.05
Actor gender × partner gender	0.19	58.68	0.42	.678	−0.72, 1.10
Actor gender × actor’s gossip with partner	−0.52	78.29	−0.73	.469	−1.94, 0.90
Partner gender × actor’s gossip with partner	−0.19	83.69	−0.26	.796	−1.64, 1.26
Actor gender × partner gender × actor’s gossip with partner	0.18	83.21	0.38	.706	−0.78, 1.14

Gossip and relationship quality

Table 5 presents the estimates from multilevel APIMs predicting self-reported relationship quality based on relational gender, covariates, and their interactions. The main effects of actors’ gossip with their romantic partner supported our hypothesis that more frequent gossip between romantic partners is positively associated with self-reported relationship quality. However, as indicated by lower regression coefficients and t-values, and higher p-values, this association was not as robust as the association between actor gossip and happiness. The positive association was present when controlling for actor and partner gender and the number of talking files for a given participant, (β = 0.18, t = 1.53, p = .129) and persisted after additional controls for extraversion, network diversity, and total social contacts were included (β = 0.18, t = 1.54, p = .126). However, when two- and three-way interactions with actor and partner gender were included in the model, actor gossip was no longer positively associated with relationship quality (β = −0.58, t = −0.59, p = .558). This finding suggests that the positive association between gossip and relationship quality is not as robust as the association with happiness.

Table 5.

APIM hypothesis testing predicting relationship quality from romantic partner gossip frequency.

DV: Relationship quality	β	df	t	p	95% CI
Actor gender	−0.09	117.22	−0.51	.611	−0.43, 0.25
Partner gender	−0.28	122.63	−1.63	.106	−0.62, 0.06
Actor’s gossip with partner	0.18	128.30	1.53	.129	−0.05, 0.42
Partner’s gossip with actor	−0.04	113.75	−0.31	.759	−0.27, 0.20
Actor talking files	0.06	128.20	0.55	.587	−0.17, 0.30
Partner talking files	0.05	113.88	0.44	.658	−0.18, 0.28
Actor age	0.01	98.16	0.73	.469	−0.01, 0.02
Actor gender	−0.06	104.84	−0.35	.729	−0.41, 0.29
Partner gender	−0.22	110.37	−1.23	.223	−0.59, 0.14
Actor’s gossip with partner	0.18	118.16	1.54	.126	−0.05, 0.42
Partner’s gossip with actor	−0.06	100.51	−0.49	.627	−0.29, 0.18
Actor talking files	0.06	115.96	0.52	.604	−0.18, 0.30
Partner talking files	0.07	105.57	0.58	.561	−0.17, 0.31
Actor age	0.004	89.07	0.62	.536	−0.01, 0.02
Extraversion	−0.04	91.20	−0.87	.386	−0.14, 0.05
Network diversity	0.04	116.51	0.41	.680	−0.13, 0.20
Total social contacts	0.01	113.75	0.41	.684	−0.03, 0.04
Actor gender	−1.01	59.41	−1.53	.133	−2.33, 0.31
Partner gender	−1.20	66.37	−1.77	.081	−2.54, 0.15
Actor’s gossip with partner	−0.56	102.47	−0.59	.558	−2.45, 1.33
Partner’s gossip with actor	−0.08	99.37	−0.66	.508	−0.32, 0.16
Actor talking files	0.09	109.25	0.76	.450	−0.15, 0.34
Partner talking files	0.10	98.80	0.79	.430	−0.15, 0.34
Actor age	−0.001	83.95	−0.13	.896	−0.02, 0.01
Extraversion	−0.04	90.32	−0.86	.392	−0.14, 0.05
Network diversity	0.03	112.02	0.30	.767	−0.15, 0.20
Total social contacts	0.01	111.53	0.48	.635	−0.03, 0.04
Actor gender × partner gender	0.66	62.48	1.48	.150	−0.23, 1.55
Actor gender × actor’s gossip with partner	0.47	91.92	0.73	.467	−0.80, 1.74
Partner gender × actor’s gossip with partner	0.80	102.87	1.17	.244	−0.56, 2.16
Actor gender × partner gender x actor’s gossip with partner	−0.52	101.06	−1.17	.247	−1.41, 0.37

Note. These are three separate APIMs, each separated by a solid line. For gender, woman = 1, man = 2. DV = dependent variable.

Among the four interactions examined, only the interaction between actor gender and partner gender predicting relationship quality met the p ≤ .20 threshold for deconstruction (β = 0.66, t = 1.48, p = .145). Simple slopes revealed evidence that WM reported lower relationship quality compared to WW (β = −0.54, t = −1.98, p = .053), and weaker evidence for all the other comparisons (βs < .35, ps > .18).

Discussion

This naturalistic observation study revealed preliminary evidence that greater engagement in gossip with one’s romantic partner is associated with higher self-reported individual and relationship well-being among same- and different-gender couples. Initial raw correlations revealed a stronger positive association between romantic partner gossip and relationship quality; however dyadic analyses accounting for covariates showed that romantic partner gossip was more strongly and reliably related to happiness than relationship quality. Furthermore, these associations did not differ based on relational gender, meaning that they were consistent across men and women in same- and different-gender couples.

Gossiping between romantic partners, particularly about shared social networks or mutual acquaintances, may serve as a form of emotional bonding and facilitate more communication among partners (Bosson et al., 2006). Because gossip is a shared activity, it may reinforce the perception that partners are “on the same team,” enhancing feelings of connectedness, trust, and other positive relationship qualities (Dores Cruz et al., 2020), as well as contributing to overall well-being. For example, negatively gossiping with one’s romantic partner on the way home from a party could signal that the couple’s bond is stronger than with their friends at the party, while positively gossiping could prolong the fun experiences. Future research should test these causal pathways and mechanisms.

Another potential explanation for these findings is that gossip can function as a social regulation or norm-learning tool within relationships (Jolly & Chang, 2021). Discussing the behaviors and actions of others could help partners align their own expectations and behaviors, contributing to a harmonious relationship. This alignment may lead to greater happiness, as both partners feel understood and supported.

Descriptively, it appears that couples with more women incrementally increase in average total gossiping, despite the lack of statistical differences in romantic partner gossip in the APIMs. Based on a 16-h waking day, participants gossiped an average of 38.40 minutes per day. WW gossiped an average of 6.72 more min than WM, who gossiped 4.80 min more than MW, who gossiped 6.72 min more than MM. Examining relational, instead of individual gender differences, sheds more precise light on gender differences, revealing that the combination of partners’ genders may matter just as much as one’s own gender to predict differences in social behavior.

The only statistical difference found between relational genders in this study was unrelated to gossip engagement. The interaction between actor and partner gender was included as a control variable when predicting relationship quality, and the findings are noteworthy. The data suggest that WM reported lower relationship quality compared to WW. Though this finding was unrelated to the purpose of this study, it is consistent with a wealth of evidence that same-gender, compared to different-gender, romantic relationships tend to be higher quality (e.g., Joyner et al., 2019; Rostosky & Riggle, 2017).

Limitations and future research

Despite its strengths, this study was limited as the first step in a novel line of naturalistic observation research on couples’ gossip. As a first step, we did not examine the subject of gossip, setting, or other descriptive factors; but paved the way for future questions, such as whether different types of gossip differently relate to well-being, whether gossip facilitates relationship maintenance behaviors such as support, and whether gossiping within a couple promotes well-being. Although our sample was ethnically diverse, future research should explore whether these findings generalize to populations outside Southern California. This region tends to be more accepting of same-gender couples, which may facilitate greater well-being. On average, participants reported relatively high relationship quality and happiness, indicating potential selection effects—partners willing to participate in this research together may have higher quality relationships. Finally, and importantly, future research should move beyond binary gender categories and examine the social dynamics of all individuals in romantic relationships, regardless of their sex, gender, or sexual orientation (Meuwly & Randall, 2019).

Conclusion

This study is the first to investigate the frequency of naturally-occurring romantic partner gossip, explore relational gender differences in gossip engagement, and examine its associations with happiness and relationship quality. By using the EAR, a naturalistic observation method, we yield insights beyond the laboratory and what people can recall or report. These data revealed another potential positive side of a behavior with a bad reputation: gossiping with a romantic partner of any gender is positively associated with well-being.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Committee on Research of the Riverside Division of the Academic Senate Regents Faculty Development Award.

Open research statement

As part of IARR’s encouragement of open research practices, the authors have provided the following information: The research was pre-registered. The aspects of the research that were pre-registered were project materials, methods, and hypotheses. A portion of the preregistered materials and hypotheses are presented in the current study, with additional hypotheses to be investigated in subsequent research. There were minor deviations from the preregistration. The registration was submitted to https://osf.io/xqty2. The data used in the research is publicly posted. The data can be obtained at: https://osf.io/rc8a6/?view_only=96985350aaac4c11ab08394d3703653c. The materials used in the research are publicly posted. The self-report materials used in this study are searchable and publicly available. Information about the EAR method can be found in the EAR repository: .

Ethics considerations

This study received Institutional Review Board approval from the University of California, Riverside (HS-14-072). All participants were verbally given information about the study and then received a written informed consent form, where they indicated their willingness to participate.

ORCID iDs

Chandler M. Spahr

Megan L. Robbins

Data Availability Statement

All primary data and analysis scripts are publicly available ().

References

Baumeister

R. F.

Zhang

Vohs

K. D.

(2004). Gossip as cultural learning. Review of General Psychology, 8(2), 111–121. https://doi.org/10.1037/1089-2680.8.2.111

Bono

Ried

L. D.

Kimberlin

Vogel

(2007). Missing data on the center for epidemiologic studies depression scale: A comparison of 4 imputation techniques. Research in Social and Administrative Pharmacy: RSAP, 3(1), 1–27. https://doi.org/10.1016/j.sapharm.2006.04.001

Bosson

J. K.

Johnson

A. B.

Niederhoffer

Swann

W. B.

(2006). Interpersonal chemistry through negativity: Bonding by sharing negative attitudes about others. Personal Relationships, 13(2), 135–150. https://doi.org/10.1111/j.1475-6811.2006.00109.x

Braithwaite

Holt-Lunstad

(2017). Romantic relationships and mental health. Current Opinion in Psychology, 13, 120–125. https://doi.org/10.1016/j.copsyc.2016.04.001

Cohen

Doyle

W. J.

Skoner

D. P.

Rabin

B. S.

Gwaltney

J. M.

Jr. (1997). Social ties and susceptibility to the common cold. Journal of the American Medical Association, 277(24), 1940–1944. https://doi.org/10.1001/jama.1997.03540480040036

Dores Cruz

T. D.

Beersma

Dijkstra

M. T. M.

Bechtoldt

M. N.

(2019). The bright and dark side of gossip for cooperation in groups. Frontiers in Psychology, 10, 1374. https://doi.org/10.3389/fpsyg.2019.01374

Dores Cruz

T. D.

Nieper

Testori

Martinescu

Beersma

(2020). An integrative definition and framework to study gossip. Group & Organization Management, 46(2), 252–285. https://doi.org/10.31234/osf.io/b8x57

Eckhaus

Ben-Hador

(2019). Gossip and gender differences: A content analysis approach. Journal of Gender Studies, 28(1), 97–108. https://doi.org/10.1080/09589236.2017.1411789

Frost

(2017). Multicollinearity in regression analysis: Problems, detection, and solutions. Statistics by Jim. https://statisticsbyjim.com/regression/multicollinearity-in-regression-analysis/

10.

Gosling

S. D.

Rentfrow

P. J.

Swann

W. B.

Jr. (2003). A very brief measure of the big five personality domains. Journal of Research in Personality, 37(6), 504–528. https://doi.org/10.1016/S0092-6566(03)00046-1

11.

Hartung

F. M.

Renner

(2013). Social curiosity and gossip: Related but different drives of social functioning. PLoS One, 8(7), Article e69996. https://doi.org/10.1371/journal.pone.0069996

12.

Hurlbert

S. H.

Levine

R. A.

Utts

(2019). Coup de grâce for a tough old bull: “Statistically significant” expires. The American Statistician, 73(sup1), 352–357. https://doi.org/10.1080/00031305.2018.1543616

13.

Johnson

M. D.

Lavner

J. A.

Mund

Zemp

Stanley

S. M.

Neyer

F. J.

Impett

E. A.

Rhoades

G. K.

Bodenmann

Weidmann

Bühler

J. L.

Burriss

R. P.

Wünsche

Grob

(2021). Within-couple associations between communication and relationship satisfaction over time. Personality and Social Psychology Bulletin, 48(4), 534–549. https://doi.org/10.1177/01461672211016920

14.

Jolly

Chang

L. J.

(2021). Gossip drives vicarious learning and facilitates social connection. Current Biology: CB, 31(12), 2539–2549. https://doi.org/10.1016/j.cub.2021.03.090

15.

Joyner

Manning

Prince

(2019). The qualities of same-sex and different-sex couples in young adulthood. Journal of Marriage and Family, 81(2), 487–505. https://doi.org/10.1111/jomf.12535

16.

Karan

Robbins

M. L.

(2017). Does quantity equate quality? An exploration of differing sampling rates using the electronically activated recorder (EAR). Society for Ambulatory Assessment.

17.

Kenny

D. A.

Ledermann

(2010). Detecting, measuring, and testing dyadic patterns in the actor–partner interdependence model. Journal of Family Psychology, 24(3), 359–366. https://doi.org/10.1037/a0019651

18.

Levin

Arluke

(1985). An exploratory analysis of sex differences in gossip. Sex Roles, 12(3-4), 281–286. https://doi.org/10.1007/BF00287594

19.

Lyubomirsky

Lepper

(1999). A measure of subjective happiness: Preliminary reliability and construct validation. Social Indicators Research, 46(2), 137–155. https://doi.org/10.1023/a:1006824100041

20.

Manson

J. H.

Robbins

M. L.

(2017). New evaluation of the Electronically Activated Recorder (EAR): Obtrusiveness, compliance, and participant self-selection effects. Frontiers in Psychology, 8, 1–9. https://doi.org/10.3389/fpsyg.2017.00658

21.

Marcoulides

K. M.

Raykov

(2019). Evaluation of variance inflation factors in regression models using latent variable modeling methods. Educational and Psychological Measurement, 79(5), 874–882. https://doi.org/10.1177/0013164418817803

22.

McAndrew

F. T.

(2014). The “sword of a woman”: Gossip and female aggression. Aggression and Violent Behavior, 19(3), 196–199. https://doi.org/10.1016/j.avb.2014.04.006

23.

McAndrew

F. T.

Bell

E. K.

Garcia

C. M.

(2007). Who do we tell and whom do we tell on? Gossip as a strategy for status enhancement. Journal of Applied Social Psychology, 37(7), 1562–1577. https://doi.org/10.1111/j.1559-1816.2007.00227.x

24.

Mehl

M. R.

Robbins

M. L.

Deters

F. G.

(2012). Naturalistic observation of health-relevant social processes: The electronically activated recorder methodology in psychosomatics. Psychosomatic Medicine, 74(4), 410–417. https://doi.org/10.1097/PSY.0b013e318254547

25.

Meuwly

Randall

A. K.

(2019). Relationships, gender, and sexual orientation: Getting beyond the binary. In Schoebi

Campos

(Eds.), New directions in the psychology of close relationships (pp. 100–117). Routledge/Taylor & Francis Group. https://doi.org/10.4324/9781351136266-7

26.

Nevo

Derech-Zehavi

(1993). The development of the tendency to gossip questionnaire: Construct and concurrent validation for a sample of Israeli college students. Educational and Psychological Measurement, 53(4), 973–981. https://doi.org/10.1177/0013164493053004010

27.

Nisbett

R. E.

Wilson

T. D.

(1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84(3), 231–259. https://doi.org/10.1037/0033-295X.84.3.231

28.

Peters

Kashima

(2007). From social talk to social action: Shaping the social triad with emotion sharing. Journal of Personality and Social Psychology, 93(5), 780–797. https://doi.org/10.1037/0022-3514.93.5.780

29.

Preacher

K. J.

Curran

P. J.

Bauer

D. J.

(2006). Computational tools for probing interactions in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31(4), 437–448. https://doi.org/10.3102/10769986031004437

30.

Robbins

M. L.

Spahr

C. M.

Karan

(2024). Re-evaluating the honing framework: Naturalistic observation of same- and different-sex couples’ conversations. Personal Relationships, 31(1), 156–174. https://doi.org/10.1111/pere.12533

31.

Robbins

M. L.

Jonnalagadda

Spahr

C. M.

(2024). Rebalancing social and personality psychology methods: The case for naturalistic observation. Social and Personality Psychology Compass, 18(1), Article e12917. https://doi.org/10.1111/spc3.12917

32.

Robbins

M. L.

Karan

(2020). Who gossips and how in everyday life? Social Psychological and Personality Science, 11(2), 185–195. https://doi.org/10.1177/1948550619837000

33.

Robbins

M. L.

Karan

Randall

A. K.

(2021). Initial examination of social compensation and honing frameworks for sexual minority social life and well-being. Journal of Social and Personal Relationships, 38(1), 301–321. https://doi.org/10.1177/0265407520961840

34.

Rostosky

S. S.

Riggle

E. D. B.

(2017). Same-sex couple relationship strengths: A review and synthesis of the empirical literature (2000–2016). Psychology of Sexual Orientation and Gender Diversity, 4(1), 1–13. https://doi.org/10.1037/sgd0000216

35.

Sayehmiri

Kareem

K. I.

Abdi

Dalvand

Gheshlagh

R. G.

(2020). The relationship between personality traits and marital satisfaction: A systematic review and meta-analysis. BMC Psychology, 8(1), 15. https://doi.org/10.1186/s40359-020-0383-z

36.

Spanier

G. B.

(1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of Marriage and Family, 38(1), 15–28. https://doi.org/10.2307/350547

37.

Stronge

Overall

N. C.

Sibley

C. G.

(2019). Gender differences in the associations between relationship status, social support, and wellbeing. Journal of Family Psychology: JFP: Journal of the Division of Family Psychology of the American Psychological Association (Division 43), 33(7), 819–829. https://doi.org/10.1037/fam0000540

38.

Thomeer

M. B.

Umberson

Reczek

(2020). The Gender-as-Relational approach for theorizing about romantic relationships of sexual and gender minority mid- to later-life adults. Journal of Family Theory & Review, 12(2), 220–237. https://doi.org/10.1111/jftr.12368

39.

Watson

D. C.

(2012). Gender differences in gossip and friendship. Sex Roles, 67(9-10), 494–502. https://doi.org/10.1007/s11199-012-0160-4

40.

West

T. V.

Popp

Kenny

D. A.

(2008). A guide for the estimation of gender and sexual orientation effects in dyadic data: An actor-partner interdependence model approach. Personality & Social Psychology Bulletin, 34(3), 321–336. https://doi.org/10.1177/0146167207311199