Abstract
Although a large body of research has been devoted to investigating human mate preferences, much of this research focuses exclusively on monogamous relationships. While monogamy is a prevalent mating system, concurrent, non-monogamous mating has also been common both throughout human history and across the ethnographic record. Here, we seek to address this gap in the literature by exploring ideal preferences for two concurrent romantic partners among both monogamous and polyamorous participants. In Studies 1 and 2, using a budget allocation paradigm, we performed a k-Means cluster analysis to explore the types of ideal concurrent partners preferred by two largely monogamous samples of participants. In Study 3, we applied these same methods among a sample of polyamorous participants. Across all studies, the results revealed that a plurality of participants preferred two well-rounded partners, in which all traits were valued similarly. However, monogamous men were significantly more likely than monogamous women to prefer at least one partner for whom attractiveness and skill in bed were prioritized above all other traits. These findings only partially generalized to polyamorous participants, who also most often preferred two well-rounded partners, but did not exhibit gender differences in cluster assignment. In addition, we found that, across studies, partner preferences are likely influenced by intended investment between partners, suggesting that preferences are sensitive to the specific contexts and structures of these relationships. These findings emphasize the need for more research on the psychological systems involved in non-monogamous mating.
Keywords
Introduction
Concurrent mating, such as polygyny and polyandry, has been recurrent across human history and cultures (Gangestad & Grebe, 2015) and consensual non-monogamy is an increasingly common relationship structure across North America (Balzarini & Muise, 2020; Scoats & Campbell, 2022). Nonetheless, despite an extensive body of literature devoted to examining mating psychology in general and mate preferences in particular, much of this work asks about participants’ preferences in relation to a single partner and among exclusively monogamous samples of participants. This focus on monogamy ignores the cross-cultural prevalence of non-monogamous mating throughout human history and across the ethnographic record (Gangestad & Grebe, 2015). Concurrent, non-monogamous relationships offer their own unique suites of benefits and costs, and it is therefore possible that our mating psychology is designed to produce distinct sets of psychological and behavioral outputs in concurrent mating contexts. In the present research, we test this possibility across three studies. To do so, we examined ideal partner preferences in relation to two partners pursued simultaneously using both monogamous (Studies 1 and 2) and polyamorous (Study 3) samples of participants.
Here, we use consensual non-monogamy as an umbrella term for describing a relationship structure in which individuals are able to take on multiple (long or short-term) partners simultaneously with the consent and knowledge of all individuals involved (Conley et al., 2013). This relationship structure has become increasingly common; for example, in one large North American survey published in 2014, 5.3% of respondents indicated that they were currently involved in a consensually non-monogamous relationship (Rubin et al., 2014). In more recent research, nearly one in five respondents reported current or past involvement in a consensually non-monogamous relationship in their lifetime (Haupert et al., 2017). One subtype of consensual non-monogamy—and the subject of the present work—is polyamory. Polyamory refers to a form of consensual non-monogamy in which all relationship partners can take on multiple long-term partners simultaneously (Loue, 2006). In one study, 3.7% of respondents indicated they were currently involved in polyamorous relationships (Haupert et al., 2017).
While monogamy—an exclusive, dyadic, long-term relationship—has been the most common mating system observed across human cultures and history, non-monogamy is far from a novelty of the Western world or the modern age; in fact, concurrent mating (i.e., taking on multiple partners at one time, though not necessarily consensually) appears to have historically been present across a large majority of human societies in the ethnographic record (Gangestad & Grebe, 2015). For instance, polygyny (in which men take on multiple female partners) has been observed in over 80% of societies in the Standard Cross-Cultural Sample (Murdock & White, 1969). Polyandry (in which women take on multiple male partners) is less common than polygyny, but it has nonetheless been observed across a number of geographically distant societies (Starkweather & Hames, 2012). Indeed, although polyandry appears to be practiced most commonly in the Tibetan plateau and the Marquises islands, it has also been observed across at least 53 additional small-scale societies (Starkweather & Hames, 2012). The prevalence of concurrent mating suggests that pursuing and acquiring multiple mates may have been a recurrent feature of human mating over evolutionary time. For this reason, the possibility exists that humans may possess a mating psychology designed, in part, for responding to concurrent mating contexts.
Despite the evolutionary recurrence of concurrent mating systems and growing contemporary interest in consensual non-monogamy, little research to-date has explored the preferences people hold for concurrent long-term partners. Instead, the overwhelming majority of prior literature on mate preferences has focused nearly exclusively on the preferences reported or observed in monogamous contexts (e.g., Buss, 1989; Hill, 1945; Li et al., 2002; Walter et al., 2020). The large body of research on monogamous mate preferences has consistently revealed that both sexes value traits such as health, intelligence, and kindness above all others (Buss, 1989). These shared preferences likely reflect the shared adaptive problems faced by both sexes (Buss, 1989), including the need for a highly cooperative partner, the need for a heavily invested partner, and the need for a partner of high genetic “quality.” But in addition to these adaptive problems common across both sexes, other selection pressures were likely to have been more relevant to members of each sex. For women, more than men, reproductive success would have historically been limited by the caloric, energetic, and time costs associated with pregnancy and lactation; for men, more than women, reproductive success would have historically been limited by the availability of partners high in reproductive value. Sex differences in preferences reflect this and have been accordingly robust (e.g., Buss, 1989; Hill, 1945; Townsend & Wasserman, 1998; Walter et al., 2020). Men, more than women, consistently report preferring younger and more physically attractive partners; conversely, women, more than men, prefer older partners with greater access to resources (Buss, 1989; Walter et al., 2020).
Despite the consistent results observed in this body of literature across cultures and time, nearly all of this work has asked about preferences in relation to a single, long-term romantic partner. Consequently, it remains unclear whether we should expect the findings from this literature to apply in concurrent mating contexts. Importantly, the costs and benefits of any given trait (e.g., attractiveness, generosity, resource access, etc.) may differ between monogamous and non-monogamous contexts. For instance, in the context of polyamory, the costs and benefits associated with a trait such as generosity may depend, in large part, on the value of this trait in other partners. Because the total sum of resources at one’s disposal may derive from multiple partners or a single partner, a stingy partner may be of little consequence when this partner is retained in the presence of one or more additional partners high in generosity. Nonetheless, this approach may not hold across traits. Although a single wealthy partner can provide the resources of two average partners, a single female partner high in fertility cannot provide the offspring of two fertile partners. In this way, whereas preferences for traits associated with fertility may be expected to persist across partners, preferences for other traits, such as generosity, may decrease in importance as the number of other partners increase.
Given the ubiquity of concurrent mating over evolutionary time and the inequalities in the costs and benefits associated with various traits across partners, the possibility remains that humans possess a mate preference psychology designed to produce distinct sets of preference outputs across ecological contexts. Under this account, humans may possess a mate preference psychology designed to output either a single set of preferences (in monogamous contexts) or distinct sets of preferences in relation to each partner (in concurrent contexts). Exactly which preferences are likely to change across partners may be expected to vary as a function of the ease with which a given trait in one partner (e.g., resource access) can offset the relevance of this trait in subsequent partners.
Indeed, in the small body of literature which exists in this area, some evidence for these kinds of shifts has already been identified. For instance, men report distinct sets of preferences in relation to marriage partners and affair partners, prioritizing traits such as youth and physical attractiveness in affair partners and valuing a wider array of traits, such as kindness, in marriage partners (Choy et al., 2023; Li & Kenrick, 2006). A similar pattern appears to emerge among men in the Himba, a semi-nomadic pastoralist group in Namibia in which men and women regularly take on informal, secondary partners in addition to their primary marriage partner (Scelza & Prall, 2018). Among women, however, the same benefits are often obtained in relation to each partner, resources and protection (Starkweather & Hames, 2012; Stephens, 1988). In Scelza and Prall’s (2018) work with the Himba, women typically reported similar preferences between formal marriage partners and informal (secondary) partners (Scelza & Prall, 2018).
Nonetheless, the body of literature exploring these questions is exceptionally small, and all existing work compares preferences exclusively in relation to primary and secondary partners (marriage partners vs. affair partners: Li & Kenrick, 2006; formal marriage partners versus informal secondary partners: Scelza & Prall, 2018). Because primary and secondary partners may meaningfully differ from other forms of concurrent mating, it is unclear whether these preferences translate to other concurrent mating contexts. Here, we address this gap in the literature by exploring concurrent preferences across three samples. We sought to address the question: When given the opportunity to take on multiple long-term partners, are ideal preferences for these partners shared or distinct? In Studies 1 and 2, we asked a majority-monogamous sample to rate their ideal preferences across two concurrent partners. In Study 3, we examined these effects in an exclusively polyamorous sample. Additionally, Studies 2 and 3 asked participants to indicate their preferred degree of investment across their created ideal partners. Doing so allows us to assess partner preferences across multiple long-term partners without restricting participants to a particular relationship structure, such as primary versus secondary partners. Moreover, such a design allows us to test whether the traits preferred across multiple ideal partners may differ as a function of one’s investment in each.
To test the possibility that two preferred partners may be qualitatively distinct, we used k-means cluster analysis. Rather than comparing the traits of each preferred partner across a series of t-tests, cluster analysis permits a comparison in the overall pattern of preferences reported in relation to each partner. However, to prevent clusters from emerging which only capture differences in mate value (e.g., a high-desirability cluster and a low-desirability cluster), we used a budget allocation paradigm. Participants were given a “budget” of points which they could “spend” on traits across two ideal partners. Both the use of a budget allocation paradigm and the use of cluster analysis mirror prior research on monogamous mating (e.g., Li & Kenrick, 2006). All materials, data, and code are publicly available on the Open Science Framework (https://osf.io/p8xj4/overview).
Study 1
In Study 1, we sought to examine both whether, and how, participants preferred concurrent partners to differ. As polyamorous samples can be challenging to recruit, Study 1 included a general sample (i.e., not limited to self-identified polyamorous participants) and asked about their ideal preferences in relation to two concurrent long-term partners. We chose to focus on seven key traits: Ambition, physical attractiveness, intelligence, kindness, social status, financial prospects, and “good in bed.” The first six of these traits were chosen because they are commonly used in existing research on mate preferences (e.g., Buss, 1989; Walter et al., 2020). We chose to include one additional trait, “good in bed”, to mirror Scelza and Prall’s (2018) existing work on concurrency in the Himba—one of the few existing studies assessing concurrent partner preferences. As the current work is largely exploratory, we did not have specific hypotheses and instead considered two possibilities: (1) ideal preferences for concurrent partners are distinct, such that some traits are prioritized in one partner, while different traits are prioritized in the other; or (2) ideal preferences are shared, such that people may seek to maximize the same ideal trait preferences across both partners.
Method
Participants
An initial sample of 289 participants was recruited through Prolific. The survey was advertised as “Exploration of Romantic Partner Traits”, with no mention of polyamory or monogamy in the description. We set the balance criterion in Prolific to recruit an equal number of male and female participants. Only participants residing in the United States were recruited for the survey.
Participants’ responses were excluded if they did not fully complete the budget allocation items necessary for the analyses (n = 16). Furthermore, we excluded participants who reported identifying as a gender other than man or woman (n = 6). We excluded non-binary participants because we intended to explore sex/gender differences in our findings and lacked a large enough sample size of non-binary participants to make statistical comparisons.
Our final sample consisted of N = 267 participants, with n = 133 (49.81%) participants identifying as women and the remaining n = 134 (50.19%) as men. Participants ranged in age from 19 to 89 years old, with an average age of M = 36.79, SD = 11.85 (Mdn = 34.00). Most participants were White (n = 192, 71.91%); however, n = 23 participants (8.61%) identified as Black/African American, n = 20 participants (7.49%) identified as Asian/Pacific Islander, n = 13 participants (4.87%) identified as Hispanic/Latino, n = 3 participants (1.12%) identified as Native American/American Indian, and n = 14 identified as multi-racial (5.24%). Two participants (.75%) declined to answer this question. Most of our sample identified as heterosexual (n = 221, 82.77%), with n = 31 (11.61%) identifying as bisexual, n = 9 (3.37%) as homosexual, and the remaining participants identifying with a sexuality not listed. Additionally, 10 (3.75%) participants identified as polyamorous. Regarding relationship status, n = 68 (25.47%) were single, n = 75 were in a relationship (28.09%), n = 12 were engaged (4.49%), and n = 112 were married (41.95%). Of these participants in a relationship, n = 4 (1.50%) participants reported currently having more than one romantic partner.
Procedure and Materials
Participants provided informed consent and then completed a survey on Qualtrics. They first were asked to complete standard demographics items, such as those assessing participant age, sex assigned at birth, sexual orientation, gender, race, and relationship status. In addition to those standard items, they were also asked whether they would consider themselves polyamorous. Polyamory was defined for participants as “a relationship orientation in which participants can pursue more than one long-term romantic partner and where these additional relationships are consented to and known about by all people involved”. Participants also reported how many romantic partners they currently had. They completed other measures that are not discussed in the present paper.
Following these questions, participants completed a budget allocation task (Li et al., 2002). This budget allocation task instructed participants to “imagine [they had] two ideal long-term romantic partners: partner Orange and partner Blue”. Participants were instructed to imagine that “both partners are aware of the other and have consented to [the participant’s] relationship with the other person”. They were then asked to allocate 70 points across 7 traits (ambition, physical attractiveness, intelligence, good in bed, kindness, social status, and financial prospects) between partner Orange and partner Blue. Participants could allocate as few as zero (“Bottom 1 out of 100 people”) or as many as 10 (“Top 1 out of 100 people”) points for any given trait of a specific partner. A total of 70 points was chosen as it allowed participants, should they desire, to allocate 5 points (“Average. Top 50 out of 100 people”) across all traits for both partners. The order of ideal partners (Blue and Orange) was counterbalanced. Participants were also asked about the ideal gender for partner Orange and partner Blue after they had allocated these points across their ideal partners. Colors, rather than numbers, were chosen to avoid forcing participants to contextualize these partners as primary and secondary partners (e.g., in the event that participants preferred to view both partners equally). The order that participants received the measures about their own traits, their current partner(s) traits, the budget allocation task, and openness to and involvement in polyamory was randomized.
Data Processing and Analyses
We conducted a k-Means cluster analysis to determine whether participants wanted their concurrent ideal partners to be categorically different from one another and whether the types of partners desired by participants differed as a function of participant gender. Data from the budget allocation measure—preferences for seven traits (ambition, physical attractiveness, intelligence, good in bed, kindness, social status, and financial prospects) in each partner—were used in this analysis. We used a scree plot to determine how many clusters best captured the variation in the data. We then created vectors of the trait preference means for each cluster to determine the levels of each trait held by the average ideal partner in a given cluster.
Following this cluster analysis, we conducted a series of chi-squared tests of independence to determine (1) whether there were differences in the cluster combinations ideal partners fell into based on participant gender, (2) whether the cluster of one ideal partner was related to the cluster of the other, and (3) whether either gender was more likely than the other to want at least one partner in a specific cluster.
Results
Budget allocations Across Traits and Partners by Gender – Means and Standard Deviations
Note. The mean budget allocation (from a total budget of 70 points to allocate across 7 traits and 2 ideal partners) is displayed for each trait, separated by Partner Blue and Partner Orange, and by participant gender. Means are presented first, and the corresponding standard deviation is listed next in parentheses. Subscripts represent within-column comparisons. Means that share letters within a column are not significantly different from one another. For rows in which between-gender comparisons were significant, the highest mean is bolded. Significant comparisons had p < .05, and all comparisons were Bonferroni adjusted.
Each of our 267 participants completed a budget allocation task for two concurrent partners; thus, 534 ideal partners were sorted into clusters using a k-Means cluster analysis. Based on a scree plot, we determined that 2 clusters provided the best fit for the data. The mean trait preference values for these clusters are displayed in Figure 1. Two Clusters of Ideal Long-Term Partners. Note. Two clusters emerged from the k-Means cluster analysis. The mean trait ratings of each cluster are plotted, with each cluster labelled descriptively with the most valued traits of that cluster. The dashed line indicates a trait value of 5, which would be the value should the budget have been equally allocated across traits and partners
Each cluster was named descriptively for the traits within the cluster that had the highest values. The number of ideal partners within each cluster varied. Overall, the clusters were “Well-Rounded” (n = 330) and “Good in Bed and Attractive” (n = 204). The “Well-Rounded” cluster was named as such because most traits centered around a value of 5, with less variation between trait means than in the other cluster (Var = 0.50 in the Well-Rounded Cluster; Var = 5.78 in the Good in Bed and Attractive cluster). The “Good in Bed and Attractive” cluster was named as such because the traits valued most highly included both “attractiveness” and “good in bed.”
We conducted a chi-squared test to determine whether cluster preference for one ideal partner (partner Blue) was significantly dependent on the cluster preference for the other ideal partner (partner Orange). We found that the cluster of one ideal partner was significantly associated with the cluster of the other ideal partner, χ2 (1, 267) = 55.78, p < .001, W = .465, such that having a partner in one cluster significantly increased the likelihood that the other ideal partner would also fall within that same cluster. Furthermore, 74.53% of participants created ideal partners that fell into the same cluster; a permutation test revealed that this was significantly greater than would be expected by chance, 95% CI = [46.92%, 58.05%]. The most common cluster combination was two “Well-Rounded” partners (49.06%). The percentage of participants whose ideal partners fell within each combination of clusters is shown in Figure 2. Percentage of Participants with Ideal Partners in Each Combination of Clusters. Note. Matrix plots illustrating the percentage of ideal partner pairs in each combination of clusters. Higher percentages are darker in color. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed. The y-axis represents the cluster of partner Orange, while the x-axis represents the cluster of partner Blue. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed
Although most participants had both partners in the same cluster, over 25% of participants created ideal partners who fell into different clusters. To further explore this pattern, we examined the relationship between cluster choice and participant gender. We found that cluster combination was significantly dependent on participant gender, χ2 (2, 267) = 12.385, p = .002, W = .215. Figure 3 shows the breakdown of partner cluster combinations separated by participant gender. For both men and women, the most common combination included two Well-Rounded clusters: 38.81% of men and 59.40% of women created ideal partners that both fell into this cluster. However, gender differences were observed in preferences for partners in the “Good in Bed and Attractive” cluster. The second most common cluster combination for ideal partners created by male participants included two partners in the “Good in Bed and Attractive” cluster (32.84%). However, this was the least desired cluster combination among female participants (18.05%). Additionally, female participants were significantly more likely than male participants to create at least one ideal partner who fell within the “Well-Rounded” cluster, χ2 (1, 267) = 6.93, p = .008, W = .17, and male participants were significantly more likely than female participants to create at least one ideal partner who fell within the “Good in Bed and Attractive” cluster, χ2 (1, 267) = 10.52, p = .001, W = .21. Frequency of Each Cluster Combination Separated by Participant Gender. Note. Matrix plots illustrating the percentage of ideal partner pairs in each combination of clusters. There are separate plots for female participants (A) and male participants (B) Higher percentages are darker in color. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed. The y-axis represents the cluster of partner Orange, while the x-axis represents the cluster of partner Blue. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed
Discussion
In Study 1, we found that when evaluating two concurrent ideal romantic partners, a majority of participants reported similar preferences in relation to both partners (i.e., two “Well-Rounded” partners). However, over 25% of participants reported distinct sets of preferences in relation to each partner; traits commonly emphasized in short-term mate preferences—including physical attractiveness and skill in bed—seem to be prioritized in a second partner, and male participants, in particular, seem to be expressing these preferences most strongly. These results largely align with prior literature on preferences in the Himba, (e.g., Scelza & Prall, 2018), as well as work on preferences for affair partners (e.g., Li & Kenrick, 2006), and work on the benefits of polyamory in western cultures (Balzarini et al., 2017).
As Study 1 was exploratory, we conducted an additional study to replicate these findings, as well as to correct a minor error in the survey 1 . Additionally, Study 1 was limited in that it did not include any follow-up questions about these ideal partners, including how much participants envisioned investing in each partner. It is possible that participants imagined investing in one partner more heavily than another. Additionally, although participants were repeatedly informed that both of these ideal partners would be long-term, some participants may have nonetheless been envisioning one long-term partner and one short-term partner. To address this possibility, Study 2 sought to replicate the results observed in Study 1 while also assessing participants’ desired investment in each partner.
Study 2
Methods
Participants
An initial sample of N = 295 participants was recruited through Prolific. Exclusion criteria were the same as in Study 1, except for the inclusion of additional attention checks. Everyone received a minimum of two attention checks; however, if participants received additional blocks of questions for measures not discussed in this paper, they received an additional attention check question. We excluded participants who incorrectly answered at least one attention check question (n = 64). We also excluded participants who did not identify as either a man or a woman (n = 4). After applying these exclusion criteria, the final sample consisted of N = 227 participants.
Of this final sample, n = 111 (48.90%) participants identified as women and n = 116 (51.10%) participants identified as men. Participants ranged in age from 19 to 83 years old, with an average age of M = 40.77, SD = 14.11 (Mdn = 37.00). Most participants were White (n = 175, 77.09%); however, n = 19 participants (8.37%) identified as Black/African American, n = 11 participants (4.85%) identified as Hispanic/Latino, n = 9 participants (3.96%) identified as Asian/Pacific Islander, n = 1 participant (.44%) identified as Native American/American Indian, and n = 12 identified as multi-racial (5.29%). The majority of our sample was heterosexual (n = 192; 84.58%), while n = 17 (7.49%) were bisexual, n = 11 (4.85%) were homosexual, and the remaining n = 7 (3.08%) identified as another, unlisted sexuality. 10 (4.41%) participants identified as polyamorous. Additionally, most participants reported having a romantic partner: n = 117 were married (51.54%), n = 2 were engaged (0.88%), and n = 40 were in a relationship (17.62%). The remaining n = 68 participants (29.96%) reported being single. 5 participants (3.14%) reported being currently involved with more than 1 long-term romantic partner.
Procedures and Measures
Participants provided informed consent and then completed a survey on Qualtrics. All survey materials were identical to those used in Study 1 except for the addition of several investment questions, as well as some additional measures unrelated to the present project. Following the budget allocation task, we asked participants about how much they would invest in partner Orange compared to partner Blue across a series of three questions. They were reminded of the points they allocated across traits for each partner, and then asked, “If you were in a long-term, romantic relationship with both Partner Blue and Partner Orange, how would you divide your financial investment between the two partners?” with a 7-point scale ranging from “Entirely in Partner Blue” to “Entirely in Partner Orange”. Using the same scale, participants were also asked about how they would divide their time between each partner and their emotional closeness to each partner.
Data Processing and Analyses
Our processing and analyses were the same as those described in Study 1. We additionally examined whether people were more likely to invest differently in each partner if the cluster of each partner was different. To do so, we conducted a series of ordinal logistic regressions exploring the relationship between cluster combination for partners Orange and Blue and investment. We ran a series of ordinal logistic regressions with 2 predictor variables: whether the clusters for partners Orange and Blue are the same or different (0 = same, 1 = different) and participant gender (female = 0, male = 1), with deviation from equal investment as our outcome variable. Deviation from equal investment was computed as the absolute deviation between a participant’s investment rating and the midpoint of the scale (4, representing equal investment in both partners). We ran a separate regression for each of the three outcome variables: financial investment, time investment, and emotional closeness.
Initial ordinal regression models using time and emotional closeness as outcome variables did not meet the assumption of proportional odds required for an ordinal regression. Thus, for models involving these outcome variables, we transformed investment deviation into a dichotomous variable: equal investment (0) or unequal investment (1). We also ran a series of ANOVAs exploring investment as a function of partner Orange and partner Blue cluster, with separate ANOVAs for each investment type. In these models, the financial investment and time investment variables maintained their original 7-point scale, and the emotional closeness variable maintained its original 5-point scale. We initially fit each ANOVA with partner Blue cluster, partner Orange cluster, and their interaction as predictors of investment. If we did not find a significant interaction, we removed the interaction term and re-fit a model examining only the main effects of partner Blue and partner Orange cluster. Of the analyses conducted exploring investment between ideal partners, only time investment is reported in the main text for brevity; results for financial investment and emotional closeness can be found in the Supplementary Materials.
Results
Budget allocations Across Traits and Partners by Gender – Means and Standard Deviations
Note. The mean budget allocation (from a total budget of 70 points to allocate across 7 traits and 2 ideal partners) is displayed for each trait, separated by Partner Blue and Partner Orange, and by participant gender. Means are presented first, and the corresponding standard deviation is listed next in parentheses. Subscripts represent within-column comparisons. Means that share letters within a column are not significantly different from one another. For rows in which between-gender comparisons were significant, the highest mean is bolded. Significant comparisons had p < .05, and all comparisons were Bonferroni adjusted.
We again found that two clusters emerged (see Figure 4). We named these clusters valued most highly within each cluster: “Good in Bed and Attractive” (n = 186) and “Well-Rounded” (n = 268). As in Study 1, in the “Well-Rounded” cluster, no traits stood out as being substantially more valued than the others, and there was less variation across trait values in the “Well-Rounded” cluster (Var = .64) than what we observed in the “Good in Bed & Attractive” cluster (Var = 4.92). Two clusters of ideal long-term partners. Note. Two clusters emerged from the k-Means cluster analysis. The mean trait ratings of each cluster are plotted, with each cluster labelled descriptively with the most valued traits of that cluster. The dashed line indicates a trait value of 5, which would be the value should the budget have been equally allocated across traits and partners
We conducted a chi-squared test to determine whether there was a relationship between the cluster of one ideal partner and that of the other. We found that the cluster of one partner was significantly dependent on the cluster of the other, χ2 (1, 227) = 11.59, p < .001, W = .24. The majority of participants (63.00%) wanted two ideal partners of the same cluster. This was significantly different from chance, 95% CI = [45.37%, 57.71%].
The most common cluster combination preferred by participants included two “Well-Rounded” partners, with 40.53% of participants creating two ideal partners of this combination. Figure 5 shows the percentage of participants preferring each cluster combination. Percentage of participants with ideal partners in each combination of clusters. Note. Matrix plots illustrating the percentage of ideal partner pairs in each combination of clusters. Higher percentages are darker in color. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed. The y-axis represents the cluster of partner Orange, while the x-axis represents the cluster of partner Blue. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed
Although it was most common for participants to have two partners in the same cluster, 37.00% of participants created ideal partners in different clusters. To further explore this, we examined the relationship between cluster choice and gender using chi-squared tests. Replicating the results from Study 1, cluster combination was significantly dependent on participant gender, χ2 (2, 227) = 15.72, p < .001, W = .26.
In contrast with the results of Study 1, where men most often created two ideal partners who fell within the “Well-Rounded” cluster, the most common cluster combination created by men was one “Well-Rounded” partner and one “Good in Bed and Attractive” partner, with 37.93% of men reporting this ideal partner combination. The second-most common combination for men included two “Good in Bed and Attractive” partners (31.90%). Two “Well-Rounded” partners was the least desired cluster combination (30.17%) among men. On the other hand, female participants most commonly created two ideal partners that fell into the “Well-Rounded” cluster (51.35%), followed by one partner in the “Well-Rounded” cluster and one in the “Good in Bed and Attractive” cluster (36.04%). The rarest combination of partners created by women included two ideal partners who fell within the “Good in Bed and Attractive” cluster (12.61%).
Women were significantly more likely than men to have a least one ideal partner fall into the “Well-Rounded” cluster, χ2 (1, 227) = 11.03, p < .001, W = .23. Compared to 68.10% of men, 87.39% of women had at least one ideal partner in this cluster. Additionally, men were significantly more likely than women to have at least one partner fall into the “Good in Bed and Attractive” cluster, χ2 (1, 227) = 9.70, p = .002, W = .22. Compared to 48.65% of women, 69.83% of men had at least one ideal partner in this cluster. Figure 6 shows the full breakdown of cluster combination selection separated by participant gender. Percentage of participants with ideal partners in each combination of clusters. Note. Matrix plots illustrating the percentage of ideal partner pairs in each combination of clusters. There are separate plots for female participants (A) and male participants (B) Higher percentages are darker in color. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed. The y-axis represents the cluster of partner Orange, while the x-axis represents the cluster of partner Blue. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed
To explore what could be driving these cluster preferences, we examined how participants would divide their investment between their two ideal partners. Here, we report analyses where time investment is the outcome variable, though additional analyses exploring financial investment and emotional closeness can be found in the Supplementary Materials. First, we asked whether participants were more likely to report unequal investment between the two partners when those partners fell into different clusters. We conducted an ordinal logistic regression with time investment as the outcome variable. Here, there was no significant interaction; instead, we found a significant main effect of cluster (same or different) on equality of time investment, χ2 (1) = 18.52, b = −.87, SE = .20, z = −4.30, p < .001. Having ideal partners in the same cluster was associated with higher odds of equal investment compared to ideal partners in different clusters.
We also explored whether investment in one ideal partner over the other was driven by partner cluster by exploring which cluster received the most investment. Time investment was measured using a 7-point scale, where 1 represented investment entirely in partner Blue and 7 represented investment entirely in partner Orange. A 4 on this scale represents equal desired investment across both partners. When exploring desired time investment between partners, we found a significant main effect of partner Blue cluster, F (1, 222) = 5.76, p = .017, and a significant main effect of partner Orange cluster, F (1, 222) = 4.53, p = .034: Participants preferred to invest more time in partner Blue when partner Blue was Well-Rounded (M = 3.79, SD = 1.18; values lower than 4 indicate more investment in Blue over Orange) compared to when partner Blue was Good in Bed and Attractive (M = 4.17, SD = 1.21). When partner Orange was Well-Rounded, participants reported wanting relatively equal investment between the partners (M = 4.04, SD = 1.20), and when partner Orange was Good in Bed and Attractive, participants reporting wanting more time investment in their other ideal partner, partner Blue (M = 3.80, SD = 1.20). Across all investment items, including those reported in the Supplementary Materials, participants reported greater investment in partner Blue relative to partner Orange when partner Blue was Well-Rounded. Additionally, although we randomized the order in which participants saw the budget allocations for partner Blue and partner Orange, there also appeared to be a slight bias towards investing in partner Blue generally. This might indicate that although we chose these color-based names to avoid notions of primary or secondary partners, some participants may have found partner Blue more appealing based on the name alone.
Discussion
In Study 2, participants preferred concurrent ideal partners that separated into clusters resembling those observed in Study 1. As in Study 1, many participants preferred two well-rounded partners, but men, more often than women, preferred at least one “Good in Bed and Attractive” partner. These results largely replicate the findings and themes observed in Study 1. In previous work using cluster analysis to examine preferences in relation to a single partner, participants preferred a partner who fell into a “Well-Rounded” cluster (Li & Kenrick, 2006). Here, we found similar results in that people preferred at least one “Well-Rounded” partner. However, many participants reported a distinct set of ideal preferences when given the opportunity to consider a second partner.
Results from the investment analyses indicate that planned investment divisions between two partners may be driving some of the differences observed here: participants whose partners fell into different clusters reported desiring more unequal investment between them. Furthermore, partners who were well-rounded appeared to command more investment than partners who fell into the “Good in Bed and Attractive” cluster. These results suggest that people may be flexible in the mating strategy they employ when holding preferences for concurrent partners, and the strategy chosen—distinct or shared preferences—may be partially driven by intended investment between partners. In the study that follows, we explore these ideal preferences in a sample of participants engaged in concurrent mating.
Study 3
In Study 3 we sought to replicate Studies 1 and 2 in a sample of polyamorous participants. Doing so allowed us to examine whether the results of Studies 1 and 2 are specific to monogamous populations’ conceptualizations of polyamorous relationships or if these results generalize across both people who identify as monogamous and those who identify as polyamorous. We preregistered this study on AsPredicted (https://aspredicted.org/x2sf-fchf.pdf).
Methods
Participants
We sought to obtain a sample size of N = 263 polyamorous participants based on an a priori power analysis using the smallest effect sizes observed in our chi-squared analyses in Studies 1 and 2. This power analysis was based on an effect size of W = .178, observed when we had previously believed that 3 clusters best fit our data. In order to recruit a sample of polyamorous participants, we advertised a pre-screen on Prolific titled “Pre-Screen for Future Study”, advertised broadly to participants in the United States, which asked whether participants identify as polyamorous. As in prior studies, polyamory was defined for participants as a “relationship orientation in which participants can pursue more than one long-term romantic partner and where these additional relationships are consented to and known about by all people involved.” To avoid disclosing the nature of the main study, we embedded this question among a series of distractor questions. We recruited N = 5000 people to complete this prescreen, of whom n = 397 identified as polyamorous. We then advertised our main survey through Prolific to only this subset of 397 participants.
We initially recorded N = 416 responses to our main survey. This exceeds the number of people to whom the survey was advertised, indicating that some participants possibly failed to complete the survey the first time or completed the survey more than once. Based on the Prolific submission record, N = 359 unique participants completed the survey. We addressed these issues through the use of additional exclusion criteria.
First, at the beginning of the main survey, we asked participants again if they identified as polyamorous, using the same phrasing and definition as the pre-screen survey. Only N = 247 participants again indicated that they were polyamorous and were retained in the final sample.
We also asked participants three attention checks dispersed throughout the survey and eliminated n = 57 participants who incorrectly answered any of those questions, leaving a sample size of n = 191 polyamorous participants. Finally, we excluded participants who did not identify as either a man or woman, or who desired an ideal partner who was neither a man nor a woman, reducing our sample size to n = 169 participants. Although we initially sought a sample size of N = 263 in accordance with our a priori power analysis, it proved difficult to recruit a sizeable sample given the size of this population. The effect size that our initial power analysis was based on was also smaller than those observed in our updated results with 2 clusters. Thus, we conducted an additional power analysis to determine the power we would have to detect the smallest effect from Study 2 given our sample of N = 169. Using the pwr function in R, we were able to determine that, given our sample size, we had 80.03% power to detect the smallest chi-squared test effect size observed in Study 2, W = .22. Due to the sample size being smaller than intended, it is possible that we are at risk of a Type II error for smaller effects; consequently, null results should be interpreted with some caution.
Of this final sample, n = 64 participants identified as women (37.87%) and n = 105 as men (62.13%). Most participants were White (n = 104, 61.54%). Nonetheless, n = 24 participants (14.20%) identified as Black/African American, n = 11 participants (6.51%) identified as Hispanic/Latino, n = 10 participants (5.92%) identified as Asian/Pacific Islander, n = 2 participants (1.18%) identified as Native American/American Indian, n = 1 participant (.59%) identified as Middle Eastern/North African, and n = 16 identified as multi-racial (9.47%). One participant (.59%) declined to answer this question. 85 (50.30%) identified as heterosexual, n = 57 (33.73%) as bisexual, n = 20 (11.83%) as homosexual, and n = 7 (4.14%) as an unlisted sexual orientation. Participants ranged in age from 18 to 69 years old, with a mean age of M = 38.44, SD = 10.56 (Mdn = 36.00)—mirroring the age range of the previous two samples. 45 (26.63%) participants reported being single, while n = 122 (72.19%) reported being partnered (either married, engaged, or in a relationship), n = 2 (1.18%) participants declined to respond, and n = 71 (58.20%) participants reported having more than one romantic partner. Of participants with more than one partner, n = 61 participant had two partners, and n = 10 participants had three or more partners.
Procedures and Measures
Procedures and measures were the same as those described in Study 2. Additional items were included in the survey for other research projects.
Data Processing and Analyses
Our analyses are the same as those detailed in Study 2.
Results
Budget allocations Across Traits and Partners by Gender – Means and Standard Deviations
Note. The mean budget allocation (from a total budget of 70 points to allocate across 7 traits and 2 ideal partners) is displayed for each trait, separated by Partner Blue and Partner Orange, and by participant gender. Means are presented first, and the corresponding standard deviation is listed next in parentheses. For partner Blue, subscripts represent within-column comparisons, so the columns for Men and Women should be considered separately. Means that share letters within a column are not significantly different from one another. For partner Orange, there was no significant interaction between trait and gender, so the subscripts are shared across the columns for Men and Women. For rows in which between-gender comparisons were significant, the highest mean is bolded. Significant comparisons had p < .05, and all comparisons were Bonferroni adjusted.
To explore the partner “types” that participants created, we once again used a scree plot and determined that two clusters best fit our data. We graphed the mean trait values of these clusters (Figure 7) and named each cluster according to the most valued traits: “Kind and Good in Bed” (n = 203) and “Well-Rounded” (n = 135). Once again, the “Well-Rounded” cluster was named as such because of the low variance in mean trait ratings in that cluster relative to the other (Var = .11 vs. Var = 4.89). Two Clusters of Ideal Long-Term Partners. Note. Two clusters emerged from the k-Means cluster analysis. The mean trait ratings of each cluster are plotted, with each cluster labelled descriptively with the most valued traits of that cluster. The dashed line indicates a trait value of 5, which would be the value should the budget have been equally allocated across traits and partners
We conducted a chi-squared test to explore the relationship between the cluster of one partner and that of the other. We found that the cluster of one partner was significantly dependent on the cluster of the other, χ2 (1, N = 169) = 35.42, p < .001, W = .470, with most participants (74.56%) preferring two ideal partners of the same cluster. Based on the results of a permutation test, this was significantly greater than chance, 95% CI = [44.97%, 59.17%]. The most common cluster combination ideal partners fell into was two “Well-Rounded” partners (47.34%). Figure 8 shows the percentage of participants with each cluster combination of ideal partners. Percentage of participants with ideal partners in each combination of clusters. Note. Matrix plots illustrating the percentage of ideal partner pairs in each combination of clusters. Higher percentages are darker in color. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed. The y-axis represents the cluster of partner Orange, while the x-axis represents the cluster of partner Blue. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed
While the majority of participants created ideal partners who fell into the same cluster, 25.44% of participants did desire two different types of ideal partner: one “Well-Rounded” and one “Kind and Good in Bed”. To further investigate this and explore whether there was a difference in cluster preference for men, relative to women, we examined the relationship between cluster choice and gender using chi-squared tests. Unlike the results of Studies 1 and 2, cluster combination was not significantly dependent on participant gender, χ2 (2, 169) = .77, p = .680; both men and women most commonly designed two partners who fell within the “Well-Rounded” cluster (46.67% of men; 48.44% of women). Additionally, participant gender was not significantly related to whether participants reported desiring at least one partner in the “Well-Rounded” cluster, χ2 (1, 169) = .15, p = .700, nor was participant gender significantly related to whether participants preferred at least one partner in the “Kind and Good in Bed” cluster, X2 (1, 169) = .004, p = .948. Figure 9 shows the breakdown of cluster combinations separately as a function of participant gender. Although we preregistered additional analyses, these analyses were contingent on finding significant gender differences in cluster choices; as a result, these analyses were not performed. Percentage of participants with ideal partners in each combination of clusters. Note. Matrix plots illustrating the percentage of ideal partner pairs in each combination of clusters. There are separate plots for female participants (A) and male participants (B) Higher percentages are darker in color. Some cells are blank because cells of the same combination, regardless of order (partner Orange/partner Blue vs partner Blue/partner Orange), are collapsed. The y-axis represents the cluster of partner Orange, while the x-axis represents the cluster of partner Blue
To further understand this discrepancy from what we observed in Studies 1 and 2, we also explored how participants sought to divide their investment between their two ideal partners. As with results from Study 2, here, we report analyses where time investment is the outcome variable. However, analyses exploring financial investment and emotional closeness can be found in the Supplementary Materials. As in Study 2, we used an ordinal logistic regression to examine whether participants preferred more unequal investment between their two partners if each ideal partner fell within a different cluster. Ideal partner cluster (same or different), gender, and their interaction were included as predictor variables in this model, and equality of time investment (equal or unequal) was once again the outcome variable. We found only a significant main effect of clusters, χ2 (1) = 4.07, b = −.52, SE = .25, z = −2.07, p = .039. Like in Study 2, having ideal partners in the same cluster was associated with higher odds of equal investment compared to ideal partners in different clusters.
Furthermore, we again ran a series of ANOVAs to explore whether investment in one ideal partner over the other was driven by partner cluster to determine which cluster type, if any, received more investment. We found no significant main effects of cluster on ideal time investment between partners for partner Blue’s cluster F (1, 165) = 2.21, p = .139, or partner Orange’s cluster, F (1, 165) = .18, p = .674. Thus, unlike our findings in Study 2 with a monogamous sample, polyamorous participants’ desired investment division between the two ideal partners was not influenced by partner type.
Discussion
As in Studies 1 and 2, a plurality of polyamorous participants preferred two “Well-Rounded” partners. Additionally, the proportion of polyamorous participants preferring two partners of the same type was similar to the proportions observed in the prior studies (Study 3: 74.56%; Study 2: 63.00%; 74.53% in Study 1). However, unlike Studies 1 and 2, polyamorous men and women did not differ in their partner preferences from one another; instead, both men and women most commonly preferred two ideal partners of the same type. Furthermore, unlike in Study 2, polyamorous people did not express differences in their intended investment as a function of cluster type—though this null effect may be attributable, in part, to a power issue. Indeed, in keeping with this possibility and aligning with the results observed in Study 2, participants in this study did report significant differences in intended investment as a function of whether their two partners fell within the same or different clusters. In this way, while not fully replicating the results previously, these results nonetheless align with Study 2 in highlighting the importance of partner investment as a predictor of the ideal partners designed by participants.
General Discussion
When given the opportunity to take on multiple concurrent partners, what preferences do people hold? Because previous research has focused primarily on monogamous mating psychology, this question has largely been ignored. Here, we conducted three studies exploring ideal preferences for long-term concurrent partners in both monogamous and polyamorous participants in the United States. These findings suggest that most participants, whether monogamous and polyamorous, tend to hold similar preferences across both partners. Across all studies, the partner combination preferred most commonly by participants included two well-rounded partners. Additionally, gender differences were also observed among monogamous participants, with men, relative to women, more commonly preferring at least one “Good in bed and Attractive” partner (i.e., in which attractiveness and skill in bed were prioritized above all other traits). However, these gender differences were not found among polyamorous participants.
That most participants desired two partners of the same “type”, and these most often were two “Well-Rounded” partners across studies, suggests a preference psychology that outputs shared preferences for a long-term partner, regardless of concurrency. However, that many participants still reported distinct preferences for each partner also suggests that there may be specific contexts and considerations that lead to distinct sets of preferences across concurrent partners–particularly among men. Although men in previous research have been found to consistently value attractiveness more than women (Buss, 1989; Symons, 1979), both genders reporting valuing other traits, such as kindness and intelligence, most highly. However, the results of Studies 1 and 2 suggest that when monogamous men are given the opportunity to take on two concurrent partners, many report exaggerated gender-differentiated preferences in at least one of their two ideal partners—valuing traits such as attractiveness over and above traits more commonly prioritized traits such as intelligence and kindness. This finding mirrors work by Scelza and Prall (2018), which found that Himba men value traits such as grit and respect in formal marriage partners while valuing attractiveness above all other traits in informal (secondary) partners. This also mirrors previous work on male preferences for affair partners, with male respondents prioritizing traits otherwise valued in short-term partners, such as physical attractiveness (Li & Kenrick, 2006). Furthermore, these findings are in line with the primary functional benefit that men obtain from concurrent mating opportunities–increased reproductive success via increased mating opportunities (Mogilski et al., 2023).
Although a plurality of both monogamous and polyamorous participants preferred two well-rounded partners, the second cluster which emerged among polyamorous participants was slightly different from the cluster which emerged among monogamous participants: Whereas the second cluster emerging among monogamous participants captured preferences for an ideal partner who was attractive and good in bed, the second cluster emerging among polyamorous participants captured preferences for an ideal partner who was kind and good in bed. The prioritization of kindness observed here among polyamorous participants aligns more closely with prior findings on long-term partner preferences.
Moreover, among polyamorous participants, there were no gender differences in cluster preferences; the finding observed in Studies 1 and 2 —in which some men exhibited exaggerated gender-specific preferences in one partner—were unique to our primarily monogamous samples. This may be the result of participants’ expectations about investment in each partner: Both monogamous and polyamorous participants reported more unequal investment between partners (across all types of investment for monogamous participants and for time investment among polyamorous participants; see Supplemental Materials for more details) when those partners fell into different clusters. This seems to suggest that ideas about the structure of these relationships may influence preferences for specific types of partners: Envisioning a less equal relationship dynamic (e.g., one primary partner and one secondary partner in whom one is less invested) may influence the expression of specific preferences. Indeed, opportunities for concurrent mating take different forms, within and across cultures, including having short-term affair partners and a single long-term partner, hierarchical and non-hierarchical long-term polyamorous relationships, polygyny, and polyandry, and each of these relationship structures may possess distinct costs and benefits to participants. It is possible that concurrent partner preferences may reflect anticipated differences in investment between partners or overall relationship/household structure. We did not collect data from participants about their intended relationship structure (e.g., hierarchical, non-hierarchical, etc.) with their ideal partners, or polyamorous participants’ actual experienced relationship structures. We discuss this limitation later in the discussion.
Because investment expectations seemed to be a key driver of the kinds of preferences reported by participants in relation to each partner, the present results may not necessarily suggest a distinct “non-monogamous” mating psychology designed for concurrent mating specifically. Instead, these results may align with existing research suggesting distinct sets of preferences in relation to casual, short-term partners and long-term romantic partners; in the present work, some of the preferences commonly observed in relation to casual short-term partners also emerged in relation to secondary long-term partners in whom participants reported less interest in investing. Thus, these results are consistent with the possibility that, at least among men, short-term mate preferences may sometimes emerge even in relation to long-term partners when one or more long-term partners are invested in less heavily than others.
Limitations and Future Directions
This current research was a necessary first step in examining ideal preferences for concurrent partners. However, the present research has some limitations. Most notably, our sample of polyamorous subjects in Study 3 was smaller than what we originally intended to obtain on the basis of our pre-registration. Although a post-hoc power analysis suggests sufficient power to detect the expected effect sizes, the smaller sample size risks Type II error for smaller effects. Thus, the null effects observed in Study 3 should be interpreted with this caveat in mind. Future research should attempt to replicate these findings using a larger polyamorous sample, perhaps recruited via other platforms.
Another limitation of this work is that we did not ask polyamorous participants about the type of polyamory they practiced and thus do not know whether they practiced hierarchical (primary/secondary) polyamory or non-hierarchical polyamory. It is unclear how participants are interpreting the task of designing two ideal partners with regard to relationship structure, which limits the interpretation of results. It is possible that the monogamous participants implicitly assumed a hierarchical relationship, whereas polyamorous participants might have more varied experiences of different concurrent relationship structures, explaining the differences in their preferences. We used this budget-allocation two-partner paradigm to allow us to explore preferences without restricting participant interpretation to a single “type” of consensual nonmonogamy (i.e., relationship structure), however doing so resulted in ambiguity in how participants perceived the two “ideal partners” and the relationship structure they would have and prevents us from exploring the relationship between specific polyamorous relationship structures and concurrent partner preferences. Thus, future work may benefit from intentionally recruiting participants of different polyamorous relationship structures and should explore preferences among polyamorous participants of different relationship configurations. Additionally, all participant samples in the current work were recruited exclusively in the United States. Future research could examine concurrent ideal preferences in other countries and societies, including those that have varying norms surrounding concurrency. Furthermore, whereas we have focused on ultimate, functional explanations for these observed patterns as informed by evolutionary psychology, other theoretical perspectives could also offer useful explanations for these findings. We encourage additional research on this topic from a wide variety of perspectives, especially those that can explain the role of sociocultural forces in the development of these patterns (e.g., social norms, sexual scripts, etc.).
Another important note is that the partner “types” identified with cluster analysis are not necessarily real, discrete types of people that exist in the world or in the mind. The exact clusters that emerge – in this case often “Well-Rounded” and “Good in Bed and Attractive” – are influenced by the design and constraints of the budget allocation task and are best conceptualized as patterns of allocation within this task. That we observe these patterns of allocation suggests that variation does exist between people in relative prioritization of these traits. However, future work should explore concurrent partner preferences using other paradigms that can help triangulate the psychological processes that give rise to these apparent clusters.
Additionally, in the current work, we examined a limited subset of potential mate preferences; future work should examine additional preferences in the context of concurrency. We are also limited in the demographics information we collected, as we did not collect information about socioeconomic class, location within the United States, or disability. Furthermore, because of our analytic approach and the small number of non-binary participants in our sample in Studies 1 and 2, we excluded non-binary participants from our analyses. This resulted in an especially large number of exclusions among our polyamorous sample in Study 3. This presents a limitation because it is unclear whether these results would generalize to non-binary participants, who seem particularly represented in consensually non-monogamous communities. Future research exploring concurrent partner preferences among non-binary participants may be particularly fruitful.
Moreover, the current research touches on only one component of mating psychology—mate preferences. Many other aspects of mating psychology may differ between monogamous and polyamorous individuals, including jealousy, sociosexual orientation, interest in sexual variety, and so on.
Finally, future research may benefit from more closely examining actual and intended investment in partners—both among those who practice concurrency and those who practice monogamy. Because polyamorous participants reported intentions to invest more equally across both partners, it is possible that, compared to monogamous people, polyamorous people dedicate more investment and energy towards mating and less toward other endeavors. Alternatively, polyamorous people may dedicate the same amount of investment and energy toward mating and simply devote half as much investment in each partner. Future research is needed to better understand the role of investment in polyamorous relationships, as well as how it differs from, or is similar to, investment in monogamous relationships.
Conclusion
The present studies were the first to evaluate ideal mate preferences for concurrent long-term partners in a Western sample. Overall, most people preferred two well-rounded partners, in which all traits were valued similarly. However, of the monogamous participants preferring different types of partners, men, in particular, exhibited exaggerated sex-specific preferences for physical attractiveness and sexual skill in one of their two partners. This is in line with prior theoretical work on the benefits of polygyny for men, as well as findings from another, non-Western sample. In addition, we found that partner preferences are likely influenced by intended investment between partners, suggesting that preferences are sensitive to the specific contexts and structures of these relationships. This pattern of results suggests that there may not be a distinct “non-monogamous” mating psychology; instead, preferences may be influenced by intentions surrounding partner investment (e.g., short-term vs. long-term partners, primary and secondary partners, etc.).
Footnotes
Ethical Considerations
These studies were approved by the University of California, Santa Barbara Research Ethics Committee (approval: 28-23-0486) on August 18, 2023. Respondents gave written consent before beginning the surveys.
Consent to Participate
These studies were approved by the University of California, Santa Barbara Research Ethics Committee (approval: 28-23-0486) on August 18, 2023. Respondents gave written consent before beginning the surveys.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by a University of California Academic Senate Faculty Research Grant awarded to Tamsin German.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Open Research Statement
As part of IARR's encouragement of open research practices, the authors have provided the following information: Studies 1 and 2 of this research were not pre-registered. Study 3 was pre-registered. The registration was submitted to AsPredicted and can be found here: https://aspredicted.org/x2sf-fchf.pdf. The data and materials used in the research are publicly posted. This can be obtained at: 
.