Abstract Partner selection requires decisions about whom to approach, how to interpret interaction, and whether to continue searching. These decisions draw on information of different kinds and occur in markets where interest must be reciprocal. We propose a two-process framework that links candidate screening to the updating of dyadic evaluations. Screening allocates attention using observable attributes and anticipated fit. Updating revises evaluations using information generated through interaction. The two processes can overlap and recur, and neither is sufficient for relationship formation. We distinguish consensual desirability, unique romantic liking, perceived compatibility, and subsequent relationship quality, so that attraction is not treated as evidence of durable fit. We also distinguish the available pool from the options a person actually examines and the alternatives that remain psychologically salient. This distinction yields a conditional account of abundance: additional options may improve the chance of encountering a mutually acceptable partner while increasing comparison costs, depending on exposure, search rules, and access to diagnostic interaction. The synthesis generates predictions about the changing value of information, reciprocal interest, and the interaction between pool size and search architecture. Three proposed studies separate initial pursuit, repeated interaction, and experimental manipulation of search. The framework extends existing work by specifying how opportunities, information acquisition, and continued comparison jointly shape partner selection. Its causal mechanisms and its applicability across populations remain to be tested.
Keywords: partner selection; emerging adulthood; dyadic evaluation; romantic attraction; choice overload; online dating
Introduction Having more potential partners need not make selecting a partner easier. A larger pool can increase opportunities to encounter someone appealing, but searching that pool requires attention, and a preferred candidate may not reciprocate interest. Even mutual attraction supplies incomplete information about how a relationship would function. A theory of partner selection must therefore explain both the allocation of attention across candidates and the interpretation of information acquired within particular dyads.
The Internet has expanded the routes through which people meet partners outside existing social networks [1]. This change makes the organization of search theoretically consequential. However, neither an increase in accessible profiles nor the prevalence of online dating shows that individuals have many feasible, mutually interested partners. Geography, orientation, relationship goals, platform visibility, and willingness to reciprocate constrain the effective pool. Abundance should therefore be measured rather than inferred from the availability of a dating application.
The present article develops a framework for emerging adults seeking a romantic relationship. University settings are one relevant context, but university attendance is not a defining condition. The framework does not assume that everyone seeks exclusivity, that singlehood is undesirable, or that choosing no partner constitutes failure. Its primary outcome is the formation of a mutually desired relationship consistent with participants’ stated goals, followed by evaluation of its quality where a relationship forms.
A recent critical review identifies fragmentation across perspectives, difficulties measuring preferences, and challenges predicting partner choices [2]. The present framework addresses part of that integration problem by linking opportunity sets to dyadic information acquisition. It makes three contributions. First, it separates two information-processing functions, screening and updating, and specifies what distinguishes them. Second, it distinguishes pool size, encountered candidates, and considered alternatives, so that opportunity benefits and comparison costs can be modeled separately. Third, it derives predictions and a three-study program that can test these claims. The article is a theoretical synthesis, not a systematic review or an empirical test. The literature discussed here does not jointly validate the proposed mechanisms, and the synthesis is intended to make them testable.
Conceptual foundations Desirability and dyadic evaluation Consensual desirability is the extent to which a target elicits favorable evaluations across perceivers within a specified population and setting. In a social relations model, the partner effect captures this shared component. It is not an intrinsic or universally stable measure of a person’s worth. Changing the population of perceivers, the available alternatives, or the relationship goal can change the evaluation.
Unique romantic liking is the component of one person’s evaluation of another that remains after accounting for the perceiver’s general tendency to like others and the target’s tendency to be liked. The social relations model represents it as a relationship effect [3]. This effect is directional: one person’s distinctive interest in another need not be reciprocated. With only one observation per directed dyad, the relationship component also includes measurement error, so repeated measurement is required to distinguish stable dyadic evaluation from transient noise.
Perceived compatibility is a person’s judgment that interaction and anticipated coordination with a particular partner fit their needs and goals. It can concern communication, intimacy, relationship expectations, or arrangements for everyday life. Unique liking and perceived compatibility may correlate, but they are not interchangeable. A residual attraction score does not identify which feature of a pairing produced it, nor does it show that the pair would function well over time.
Romantic chemistry denotes a subjective experience of connection or attraction. In qualitative content analysis, Devenport and colleagues [4] found that participants described chemistry through multiple facets that varied between individuals and co-occurred. The study offers evidence about lay concepts, not a validated predictive measure or a causal account of compatibility. Chemistry is therefore treated here as a candidate indicator of dyadic experience, not a diagnostic test of fit. Relationship quality concerns satisfaction and other evaluations within an existing relationship. These distinctions prevent a circular argument in which compatibility is defined as liking and then said to explain liking.
Pool size and exposure The available pool comprises candidates who could in principle be encountered under the study’s eligibility criteria. The encountered set comprises candidates actually viewed or met. The consideration set comprises candidates retained for serious evaluation. Psychologically salient alternatives are candidates, including imagined future candidates, that remain relevant when evaluating a current option. These sets can differ substantially.
A larger available pool need not produce a larger consideration set, and a small encountered set can still evoke a strong sense that better alternatives are available elsewhere. This separation is central to the framework. It allows opportunity benefits and comparison costs to vary independently, although ordinary platform designs may cause them to covary. Search architecture refers to the rules governing exposure, revisiting candidates, shortlisting, and ending a search episode. Its effects must be distinguished from the effects of candidate quality and interaction opportunity.
Screening and updating as overlapping processes Screening allocates opportunities for interaction Screening determines which candidates receive further attention. Profile attributes, appearance, communicated intentions, and practical constraints can influence this allocation. Screening judgments are forecasts made with incomplete information. A preference for a quality does not guarantee that a person can identify that quality from a profile, or that selecting on it improves later outcomes.
Evidence on ideal partner preferences illustrates the distinction. Eastwick and Finkel [5] found that familiar sex differences in stated preferences did not translate into corresponding differences in associations with romantic interest during speed dating. A subsequent meta-analysis found that attractiveness predicted romantic evaluations in both sexes, while earning prospects had a smaller association, and that differences between the sexes in these associations were small and nonsignificant [6]. This evidence concerns prediction across particular measures and settings. It does not establish that preferences are irrelevant or that sex differences can never occur.
Cross-cultural research provides evidence relevant to competing explanations of stated preferences. Buss [7] reported sex-linked patterns across 37 samples in 33 countries, and Walter and colleagues [8] reported a large preregistered replication across 45 countries. These surveys establish broad patterns in reported ideals, but they do not alone show how those ideals govern choices in a particular encounter. Li and Meltzer [9] argue that sex-differentiated selection may emerge under theoretically specified conditions. In experiments, Li and colleagues [10] found context-dependent differentiation, including differences between long- and short-term goals and between pools with and without low-value candidates. In four longitudinal studies of married couples, Meltzer and colleagues [11] reported that attractiveness predicted husbands’ marital satisfaction more strongly than wives’. These results address distinct populations, choice stages, and outcomes. Together they qualify any blanket conclusion that sex differences either always or never shape partner selection.
More recent evidence strengthens the need for qualification. A registered report involving 10,358 participants in 43 countries found significant preference-matching effects, with estimates differing by analytic metric [12]. The corrected pattern effect was smaller after accounting for normative preference matching, and trait-specific interaction effects were very small on average. The framework therefore allows preferences to influence screening while treating their incremental predictive value as an empirical question. It does not rely on a universal ranking of attractiveness, resources, or personality.
Communicated kindness, social ease, and responsiveness might also influence initial judgments. Where they enter a screening process depends on when they become observable. A photograph cannot provide the same information as conversation, and a brief conversational impression may partly reflect the perceiver’s own attraction. Rather than assigning social competence to a fixed stage, research should record the source and timing of each judgment and examine whether it adds information beyond prior evaluation.
Interaction updates evaluations of a dyad Updating occurs when new evidence changes an evaluation of a candidate or a pairing. Conversation can reveal whether attention is reciprocated, whether communication feels responsive, and whether expectations are compatible. Updating can begin at the first encounter.
The distinction between screening and updating lies in their inputs and outputs, not in their timing. Screening uses information available before or outside a specific dyad to decide where attention goes. Updating uses information generated within a dyad to revise an evaluation of a specific candidate. The same judgment, such as perceived responsiveness, can serve either function depending on when and how it is formed. This is why the framework specifies two information-processing functions rather than a compulsory sequence in which desirability precedes compatibility.
Baxter and colleagues [3] found that initial evaluations contained both consensual and relationship-specific components, and that these components predicted later romantic interest and dating. Their findings support the relevance of unique early evaluations and challenge a rigid model that postpones dyadic information until after screening. The proposed distinction is instead between evaluating a candidate from currently available cues and revising that evaluation as additional information is acquired.
Reciprocity changes which options remain feasible. Eastwick and colleagues [13] distinguished selective desire for a particular person from a general tendency to desire others, and found different patterns of reciprocation. The implication is that candidate evaluation and candidate availability cannot be modeled independently. Continued interest can itself change the interaction, and an initially appealing candidate may cease to be an attainable option. A unilateral ranking of profiles is therefore an incomplete representation of partner selection.
Once relationships exist, their experienced characteristics become especially relevant. Across 43 longitudinal couples datasets, Joel and colleagues [14] identified relationship-specific self-reports as strong predictors of relationship quality. Individual differences and partner reports added little beyond participants’ own relationship-specific reports, and change in quality was difficult to predict. These findings justify studying experiences within a dyad, but they do not validate early compatibility measures or show that changing a measured predictor would improve a relationship. Established couples are also selected survivors of earlier processes, and their reports can share evaluative content with satisfaction measures.
Continuation and renewed search After an interaction, participants may continue with the same person, return to search, or stop seeking a relationship. None of these actions shows that the preceding judgment was accurate or inaccurate. Continuing can reflect attraction, limited alternatives, or inertia; stopping can reflect incompatible goals or a preference for singlehood. Outcomes should be interpreted relative to the participant’s aims and the information then available.
The distinctive prediction is that the value of additional search depends on what has already been learned. When current evidence is sparse, meeting another candidate can supply both a new opportunity and a comparison benchmark. When a mutually acceptable pairing has generated credible information, further profile comparison may offer less useful evidence while sustaining uncertainty. The account permits renewed search after adverse information and does not prescribe an irreversible stopping rule.
Similarity and coordination in established couples Similarity evidence helps delimit what the framework means by compatibility. In a sample of 291 newlywed couples, Luo and Klohnen [15] found substantial partner similarity in attitudes but little in personality, and associations between similarity and marital quality differed across domains. Later dyadic analyses using polynomial regression generally found that Big Five similarity contributed little to satisfaction beyond each partner’s trait levels, although limited longitudinal patterns appeared [16]. A review similarly concluded that personality effects on satisfaction were more consistent than incremental similarity effects [17]. These findings concern established couples and cannot establish which cues people should screen for when dating. They also caution against inferring a benefit from similarity merely because couples resemble each other.
For the present framework, compatibility refers to the perceived ability to coordinate on personally consequential expectations, not a generic match between trait profiles. Whether actual or perceived similarity helps forecast coordination is an open, domain-specific question. Studies should distinguish attitude or value similarity, perceived similarity, and personality similarity, and should model actor and partner traits before testing incremental dyadic fit.
A conditional account of option abundance Opportunity benefits An elementary benchmark clarifies the opportunity mechanism. If each independently encountered candidate has probability p of meeting a person’s minimum criteria and reciprocating interest, then the probability of at least one acceptable pairing among n encounters is 1 − (1 − p )^n . This expression is an illustrative probability identity, not an estimated model of dating. It assumes independent opportunities and a fixed p , assumptions likely violated by assortative exposure, competition, and correlated preferences. It also concerns encounters rather than the number of profiles nominally available.
The benchmark shows why additional exposure can increase opportunity under some conditions, with diminishing gains. It does not imply greater relationship quality, guaranteed formation, or improved decision satisfaction. If extra candidates are inaccessible, redundant, or unlikely to reciprocate, the effective gain can be small. The opportunity mechanism must therefore be tested using feasible candidates and mutual outcomes, not only a participant’s evaluation of the most attractive profile.
Hu and Markowitz [18] provide recent evidence against a uniformly adverse effect of large sets. In two preregistered experiments, participants selected from six or thirty-one profiles, and those in the larger-set condition reported greater pursuit of their selection. Perceived compatibility or similarity statistically mediated the effect. This supports the possibility of opportunity benefits but does not establish improved mutual matching or durable commitment. The manipulation changed exposure, and the mediation evidence does not by itself identify a causal mechanism.
Comparison costs Classic experiments show that larger assortments can reduce selection or satisfaction in some consumer tasks [19]. The broader evidence does not support a universal law. Scheibehenne and colleagues [20] found a near-zero mean effect and substantial variation across experiments. Chernev and colleagues [21] identified choice complexity, task difficulty, preference uncertainty, and decision goal as moderators. These findings make conditional predictions essential and limit straightforward transfer from consumer choices to reciprocal relationships.
Dating experiments give the comparison account direct but mixed support. Wu and Chiou [22] reported that a larger profile set increased search and was associated with poorer selection in their experimental sample. D’Angelo and Toma [23] found lower satisfaction one week after selecting from 24 rather than six profiles, with the lowest satisfaction when participants could reverse a large-set choice. These results concern particular interfaces, samples, and short-term judgments. They do not show that every larger real-world pool harms matching or relationship outcomes. They motivate separating search behavior, reversibility, pursuit, and satisfaction in new tests.
Within the proposed framework, comparison costs arise when evaluating or revisiting alternatives consumes attention without supplying enough useful information to justify that effort. Relevant outcomes include search time, subjective difficulty, decision satisfaction, and regret. These need not move together: more time can reflect productive deliberation, and high confidence can accompany a poor decision. Maximizing tendencies may moderate these processes, and their association with dissatisfaction and regret [24] suggests, but does not establish, that imposing a search limit would improve outcomes.
Opportunity benefits and comparison costs are thus separable hypotheses, not an established reconciliation of all findings about online dating. They can coexist, cancel, or dominate in different environments. A study might observe increased pursuit alongside greater regret, or observe neither effect because additional exposure is uninformative. These possibilities make the framework more discriminating than a claim that abundance is simply beneficial or harmful.
Search architecture and information quality Some dating decisions also involve sequential search. In an economic optimal-stopping task, van de Wouw and colleagues [25] found that biased expectations about the quality of future options helped explain participants’ sampling decisions. This is evidence about sequential economic decisions, not a test of dating choices. It motivates measuring expectations about future alternatives where candidates are encountered one at a time. A consideration limit may reduce comparison effort, but it may also remove candidates before their diagnostic value has been assessed. Which effect dominates is an empirical question addressed in Study 3.
Finkel and colleagues [26] distinguished access, communication, and matching in their analysis of online dating. That distinction explains why the medium alone is too coarse a predictor: increased access need not improve the information supplied by communication, and profile matching does not establish that a pairing will succeed. Their 2012 review is a foundation for this framework, not evidence about the performance of every contemporary algorithm. The present framework focuses on observable exposure and interaction rules that can be tested independently of proprietary matching claims.
Predictions and conditions that would challenge the framework P1: Changing predictive value of information
After initial evaluation is accounted for, information acquired during interaction should improve prediction of subsequent pursuit when it contains reliable dyad-specific information. The critical test compares models using preinteraction attributes alone with models that add separately measured postinteraction experiences. Prediction should be evaluated on held-out participants or events. A reliable interaction measure that repeatedly fails to improve prediction would challenge this claim. A noisy measure would primarily challenge the measurement strategy.
P2: Reciprocity constrains opportunity benefits
Increasing feasible encounters should improve the probability of a mutually acceptable pairing more than increasing displayed but unavailable candidates. The effect should weaken where reciprocal eligibility is low. Support requires behavioral or independently reported mutual interest. A gain confined to unilateral pursuit intentions would support an exposure effect but would leave the matching mechanism unconfirmed.
P3: Updating trajectories supply information beyond initial liking
Changes in perceived compatibility across repeated encounters should add predictive information beyond initial liking and elapsed contact. A positive trajectory is not assumed to outperform every stable high evaluation. The test must compare trajectories at equivalent initial levels and account for the selective continuation of promising pairs. If properly measured change contributes no replicable predictive information, the framework’s trajectory claim should be revised.
P4: Pool size interacts with search architecture
When all participants can inspect the assigned pool, a manageable shortlist with structured comparison is predicted to reduce regret more in a large pool than in a small pool, particularly when preferences are uncertain. Preservation of pursuit is a separate prediction that requires a prespecified noninferiority margin; a nonsignificant reduction is insufficient evidence of preservation. If reduced regret comes at a meaningful cost to mutual pairing or sustained pursuit, the proposed benefit of bounded comparison has not been demonstrated.
The stronger causal claim is that changes in comparison account for regret effects, while changes in access to acceptable candidates account for opportunity effects. Correlational mediator reports cannot establish this claim. It requires independent manipulation of the hypothesized mechanisms, or a clearly qualified mediation analysis with its assumptions and sensitivity checks reported.
Proposed empirical program Study 1: Initial screening and reciprocal interaction Recruit relationship-seeking emerging adults from multiple university and community settings. Construct interaction events using reciprocal eligibility, including gender, orientation, age preferences, and relationship goals. Collect preinteraction evaluations using a common brief profile format, then obtain private postinteraction evaluations after standardized conversations. Measure attraction, perceived compatibility, and desire for further contact separately. Record mutual opt-in contact exchange and follow-up communication at prespecified intervals.
Participants must rate multiple eligible partners and be rated by multiple partners so that perceiver, partner, and directional relationship components can be estimated. Fit a social relations model with repeated indicators where feasible, followed by a multilevel model for subsequent contact. The model should include event structure and dependence between reciprocal observations. The primary comparison tests incremental prediction from postinteraction experience beyond baseline evaluation. Validation must hold out whole events or participants to prevent information leakage across dyads that share a person.
Study 2: Repeated interaction and relationship formation Follow mutually interested pairs from Study 1, or a separate cohort, through repeated voluntary encounters. Measure compatibility and liking after each encounter using distinct items, with interaction opportunities and actual contact recorded separately. Assess relationship status from both participants at fixed follow-ups, and assess satisfaction only for relationships that have formed. Three or more observations per continuing pairing are a minimum design objective, not a guarantee of an estimable trajectory; reliability and power must be evaluated before recruitment.
Estimate within-dyad change separately from differences between dyads. A joint longitudinal and continuation model can address the dependence between evaluation and the chance of another encounter. Prespecify sensitivity analyses for informative dropout, since disappointed participants may be less likely to respond. Predictions about relationship quality among formed couples must be labeled conditional: excluding nonforming pairs changes the population and can induce selection bias. Report nonformation, loss to follow-up, and formation as distinct outcomes.
Study 3: Pool size and bounded comparison Randomize participants to a 2 × 2 design crossing small versus large encountered pools with free versus structured comparison. Six and thirty-one profiles provide a possible starting point for comparability with Hu and Markowitz [18], subject to piloting. Randomly draw profiles from a common calibrated population and counterbalance their order. All participants inspect their assigned pool. Structured comparison then uses a capped shortlist and explicit review criteria, while free comparison permits revisiting the full assigned pool. Equalize interface usability and instructions unrelated to the manipulation.
This design tests bounded comparison after exposure. It does not test whether truncating search before exposure preserves access to compatible candidates. Log viewing, revisiting, shortlist revisions, time, and deviations from instructions. Measure immediate regret and pursuit separately, with later willingness to initiate contact as an additional endpoint. Unless participants encounter real reciprocally eligible candidates, results concern simulated selection rather than relationship formation. A field replication would be required to evaluate mutual outcomes.
Prespecify the pool-size-by-comparison interaction for regret as the primary contrast, a separate preservation test for pursuit, and secondary process measures. Estimate effects by randomized assignment, retaining noncompliance in the primary analysis. Report effect sizes and uncertainty rather than treating statistical significance as evidence of practical benefit. A later dismantling experiment should distinguish shortlist size, reversibility, and exposure to alternatives; otherwise any observed effect belongs to the intervention package as a whole.
Shared methodological requirements Determine sample sizes with simulations that represent event clustering, dyadic dependence, repeated observations, attrition, the primary interaction, and any noninferiority analysis. The number of dyadic ratings cannot substitute for the number of independent participants or events. Prespecify the smallest effects that would matter theoretically, measurement reliability targets, exclusion rules, missing-data assumptions, and multiplicity control. Do not calculate power from selectively favorable published estimates alone.
Develop and validate compatibility measures before confirmatory hypothesis testing. Use cognitive interviewing and a separate pilot to distinguish compatibility judgments from liking and global satisfaction, and examine measurement invariance where group comparisons are intended. Do not assume an established scale exists merely because the construct has a familiar label. Preregister primary tests, and reserve subgroup and alternative-model analyses for clearly identified exploratory work unless they are adequately powered.
Ethical procedures should include informed consent, private ratings, mutual permission before contact exchange, and an option to decline without explanation. No intervention should pressure participants to continue an unwanted interaction or disclose another person’s rejection. Share analysis code and appropriately deidentified materials; access to sensitive relational data may require controlled rather than unrestricted sharing. Gender and orientation should be measured in ways appropriate to eligibility and theory, rather than reduced to a universal binary moderator.
Scope and limitations The framework is an account of information use during voluntary partner search. It does not explain every influence on relationship formation, including discrimination, coercion, network constraints, or structural inequality. Early evaluations may depend on these influences, so a statistical dyadic effect should not be interpreted as an unconstrained expression of personal fit. Cultural expectations and relationship goals may alter both the cues available and the meaning of continuation.
Emerging adulthood is a proposed application, not an empirically established boundary. Much of the cited evidence comes from samples or relationships outside that period. Testing specificity requires comparison with other age groups and attention to education, work, resources, and relationship history. University convenience samples cannot establish the framework’s generality to all emerging adults.
Several mechanisms remain unresolved. Compatibility could reflect projection, perceived reciprocity, shared circumstances, or interaction processes that develop jointly. Similarity in a trait, perceived similarity, and preference matching are different claims and require different tests. No general recommendation to prioritize value similarity over personality similarity follows from the present synthesis. Relevant coordination demands should be measured directly rather than assumed from a broad similarity score.
The account also leaves open whether continued search causes dissatisfaction or dissatisfaction prompts search. Experiments can identify the effects of assigned search rules, but observational trajectories require stronger assumptions. Partner selection is partly an evolving mutual process, so it may not have a single optimal endpoint. Decision satisfaction, reduced regret, relationship formation, and later relationship quality should remain separate evaluative criteria.
Conclusion Partner selection under abundant options can be studied as the interaction of screening, updating, and reciprocal feasibility. The same additional candidate can offer an opportunity, consume comparison effort, or provide useful information, depending on how the search is organized and what has already been learned. The framework specifies these possibilities without equating early chemistry with durable compatibility or treating larger pools as uniformly beneficial or harmful. Its central empirical challenge is to show that acquiring information within dyads and managing comparison across candidates explain distinct outcomes beyond initial attraction and simple exposure. Meeting that challenge requires repeated dyadic measurement, behavioral follow-up, and experiments that isolate search architecture from opportunity. Until those tests are completed, bounded comparison and attention to compatibility trajectories remain research hypotheses rather than established prescriptions for daters.
References 1. Rosenfeld MJ, Thomas RJ. Searching for a mate: The rise of the Internet as a social intermediary. American Sociological Review. 2012;77(4):523–547. https://doi.org/10.1177/0003122412448050
2. Devenport S, Davis-McCabe C, Winter S. A critical review of the literature regarding the selection of long-term romantic partners. Archives of Sexual Behavior. 2023;52:3025–3042. https://doi.org/10.1007/s10508-023-02646-y
3. Baxter A, Maxwell JA, Bales KL, Finkel EJ, Impett EA, Eastwick PW. Initial impressions of compatibility and mate value predict later dating and romantic interest. Proceedings of the National Academy of Sciences. 2022;119(45):e2206925119. https://doi.org/10.1073/pnas.2206925119
4. Devenport S, Phillips MJ, Mullan B, Winter S, Davis-McCabe C. Exploring lay understandings of romantic chemistry using inductive and deductive content analysis. Behavioral Sciences. 2025;15(11):1565. https://doi.org/10.3390/bs15111565
5. Eastwick PW, Finkel EJ. Sex differences in mate preferences revisited: Do people know what they initially desire in a romantic partner? Journal of Personality and Social Psychology. 2008;94(2):245–264. https://doi.org/10.1037/0022-3514.94.2.245
6. Eastwick PW, Luchies LB, Finkel EJ, Hunt LL. The predictive validity of ideal partner preferences: A review and meta-analysis. Psychological Bulletin. 2014;140(3):623–665. https://doi.org/10.1037/a0032432
7. Buss DM. Sex differences in human mate preferences: Evolutionary hypotheses tested in 37 cultures. Behavioral and Brain Sciences. 1989;12(1):1–49. https://doi.org/10.1017/S0140525X00023992
8. Walter KV, Conroy-Beam D, Buss DM, Asao K, Sorokowska A, Sorokowski P, et al. Sex differences in mate preferences across 45 countries: A large-scale replication. Psychological Science. 2020;31(4):408–423. https://doi.org/10.1177/0956797620904154
9. Li NP, Meltzer AL. The validity of sex-differentiated mate preferences: Reconciling the seemingly conflicting evidence. Evolutionary Behavioral Sciences. 2015;9(2):89–106. https://doi.org/10.1037/ebs0000036
10. Li NP, Yong JC, Tov W, Sng O, Fletcher GJO, Valentine KA, Jiang YF, Balliet D. Mate preferences do predict attraction and choices in the early stages of mate selection. Journal of Personality and Social Psychology. 2013;105(5):757–776. https://doi.org/10.1037/a0033777
11. Meltzer AL, McNulty JK, Jackson GL, Karney BR. Sex differences in the implications of partner physical attractiveness for the trajectory of marital satisfaction. Journal of Personality and Social Psychology. 2014;106(3):418–428. https://doi.org/10.1037/a0034424
12. Eastwick PW, Sparks J, Finkel EJ, Meza EM, Adamkovič M, Adu P, et al. A worldwide test of the predictive validity of ideal partner preference matching. Journal of Personality and Social Psychology. 2025;128(1):123–146. https://doi.org/10.1037/pspp0000524
13. Eastwick PW, Finkel EJ, Mochon D, Ariely D. Selective versus unselective romantic desire: Not all reciprocity is created equal. Psychological Science. 2007;18(4):317–319. https://doi.org/10.1111/j.1467-9280.2007.01897.x
14. Joel S, Eastwick PW, Allison CJ, Arriaga XB, Baker ZG, Bar-Kalifa E, et al. Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies. Proceedings of the National Academy of Sciences. 2020;117(32):19061–19071. https://doi.org/10.1073/pnas.1917036117
15. Luo S, Klohnen EC. Assortative mating and marital quality in newlyweds: A couple-centered approach. Journal of Personality and Social Psychology. 2005;88(2):304–326. https://doi.org/10.1037/0022-3514.88.2.304
16. Weidmann R, Schönbrodt FD, Ledermann T, Grob A. Concurrent and longitudinal dyadic polynomial regression analyses of Big Five traits and relationship satisfaction: Does similarity matter? Journal of Research in Personality. 2017;70:6–15. https://doi.org/10.1016/j.jrp.2017.04.003
17. Weidmann R, Ledermann T, Grob A. The interdependence of personality and satisfaction in couples. European Psychologist. 2016;21(4):284–295. https://doi.org/10.1027/1016-9040/a000261
18. Hu JM, Markowitz DM. Re-examining relational pursuit and mate selection in online dating. Journal of Social and Personal Relationships. 2026;43(8):2262–2285. https://doi.org/10.1177/02654075261429319
19. Iyengar SS, Lepper MR. When choice is demotivating: Can one desire too much of a good thing? Journal of Personality and Social Psychology. 2000;79(6):995–1006. https://doi.org/10.1037/0022-3514.79.6.995
20. Scheibehenne B, Greifeneder R, Todd PM. Can there ever be too many options? A meta-analytic review of choice overload. Journal of Consumer Research. 2010;37(3):409–425. https://doi.org/10.1086/651235
21. Chernev A, Böckenholt U, Goodman J. Choice overload: A conceptual review and meta-analysis. Journal of Consumer Psychology. 2015;25(2):333–358. https://doi.org/10.1016/j.jcps.2014.08.002
22. Wu PL, Chiou WB. More options lead to more searching and worse choices in finding partners for romantic relationships online: An experimental study. CyberPsychology & Behavior. 2009;12(3):315–318. https://doi.org/10.1089/cpb.2008.0182
23. D’Angelo JD, Toma CL. There are plenty of fish in the sea: The effects of choice overload and reversibility on online daters’ satisfaction with selected partners. Media Psychology. 2017;20(1):1–27. https://doi.org/10.1080/15213269.2015.1121827
24. Schwartz B, Ward A, Monterosso J, Lyubomirsky S, White K, Lehman DR. Maximizing versus satisficing: Happiness is a matter of choice. Journal of Personality and Social Psychology. 2002;83(5):1178–1197. https://doi.org/10.1037/0022-3514.83.5.1178
25. van de Wouw DS, McKay RT, Furl N. Biased expectations about future choice options predict sequential economic decisions. Communications Psychology. 2024;2:119. https://doi.org/10.1038/s44271-024-00172-8
26. Finkel EJ, Eastwick PW, Karney BR, Reis HT, Sprecher S. Online dating: A critical analysis from the perspective of psychological science. Psychological Science in the Public Interest. 2012;13(1):3–66. https://doi.org/10.1177/1529100612436522