A primer on studying effects of relationship duration in dyadic research: Contrasting cross-sectional and longitudinal approaches

Abstract

There has been increasing interest in the analysis of dyadic data. While there is ongoing progress in the development of statistically fine-grained methods to model complex dyadic data, many studies examine relationship duration in dyads using cross-sectional data (e.g., when testing the effects of acquaintanceship on the accuracy of personality judgments). We discuss the theoretical and methodological implications of analyzing and interpreting time-based variables in dyadic research on basis of the cross-sectional and longitudinal design approaches. We give examples from the literature on romantic relationships and interpersonal perception. Finally, we derive recommendations on addressing questions of time-based indexes in dyadic research and discuss advantages and disadvantages of each approach.

Keywords

Dyadic data analysis relationship duration relationship length study design cross-sectional longitudinal interpersonal perception romantic relationships

There is growing interest in dyadic research¹ since it allows for conclusions on how people experience and behave in social relationships by simultaneously analyzing data of both dyad members, for example, when studying romantic relationships and the accuracy of interpersonal perceptions—to name but two key domains of dyadic research. Here, we discuss conceptual and methodological merits and disadvantages when testing effects of relationship duration in cross-sectional and longitudinal dyadic research designs. While the discussion on the distinctions and potentially contradictory conclusions drawn from cross-sectional and longitudinal approaches is not unique to dyadic research (e.g., Galambos et al., 2020; Proulx et al., 2007), we argue that the field of relationship research might benefit from considering between- and within-person variation with more precision. We discuss both approaches and give recommendations for designs and analyses for the study of relationship duration. We illustrate our points with examples from studies of romantic relationships and interpersonal perception.

Why relationship duration is of interest in dyadic studies

The study of dyadic relationships often concerns the impact of relationship duration on outcomes such as self-other agreement (e.g., do raters derive more accurate judgments of others with greater acquaintanceship?) and relationship satisfaction (e.g., are couples more satisfied over the course of their relationship?). Early accounts on these questions date back almost a century, when, for example, Terman (1938) studied effects of marital duration and Ferguson (1949) highlighted the role of the degree of acquaintanceship for the accuracy of personality judgments, and to date, the role of relationship duration receives ongoing interest in dyadic research (e.g., Joel et al., 2020; Pusch et al., 2021). The study of relationship duration can be addressed by cross-sectional and longitudinal designs. We want to raise awareness that findings and conclusions should be evaluated with regard to methodological and interpretational limitations of each approach. As we will show, both approaches can provide differing conclusions on the same question and, in some cases, the same data.

The motivation

Across study domains there is consensus that the amount of time two people spend together affects both their individual and their dyadic characteristics. For example, the acquaintanceship effect describes judgments about others becoming more accurate with increasing time that dyad members know each other (Kenny, 2020). Of course, the important criterion is not the passing of time itself, but the processes that occur during that time; specifically, opportunities to collect and accumulate information about the respective other increase with the time spent together (e.g., Beer, 2020; Watson et al., 2000). Among other factors, aggregating such observations and experiences should help in arriving at more accurate inferences about others’ behaviors, preferences, or personality characteristics (Beer, 2020). In romantic couples, relationship duration relates to numerous outcomes, for example, trajectories of relationship satisfaction (e.g., Karney & Bradbury, 1995), individual and dyadic communication styles (e.g., Bierstetel et al., 2020; Weigel & Ballard-Reisch, 1999), and changes in partners’ agreement and similarity (Lenhausen et al., 2021; Luo, 2017), to name but a few. In short, typical research questions in this area examine whether durations of relationships relate to certain characteristics of the dyads and the dyads’ members.

The problem

The description and examination of effects of relationship duration can be tested with cross-sectional and longitudinal data. Both approaches have advantages, disadvantages, and limitations. Researchers should acknowledge these and interpret their data accordingly to avoid misinterpretations; for example, caution is warranted when analyzing and interpreting “effects” (effect used here in its narrow sense) of relationship duration on basis of cross-sectional data (i.e., comparisons between couples) with the aim to derive conclusions of how associations between study variables change over time (i.e., comparisons within couples) as a function of relationship duration, thus, interpreting cross-sectional findings as if they were longitudinal. As we will discuss, using randomly generated data and real-life studies, findings and conclusions on effects of relationship duration can differ depending on the methodological approach (see also Galambos et al., 2020; Proulx et al., 2007).

Also, taking the theoretical and statistical issues aside, we want to highlight the open question of what precisely is understood as “testing effects of relationship duration.” It appears that “effect of relationship duration” is used as an umbrella term because often conceptualizations, hypotheses, and interpretations of studies go beyond its narrow definition of a simple (frequently self-reported) number that characterizes the amount of time for how long two people maintain their relationship. Before we discuss the theoretical and statistical approaches, we will provide an understanding of the statistical status of the dyad-level variable relationship duration.²

Statistical considerations of relationship duration analyses

Statistical status of relationship duration in dyadic research designs

In a standard dyadic design, each member of the dyad provides data. Further, dyads are characterized by interdependence (nonindependence of observations; Kenny et al., 2006), and thus, test scores and error variances between dyad members are correlated. Further, there are different types of variables that differ within and between dyads and their members (Kenny et al., 2006; see also Brauer & Proyer, 2020). First, within-dyad variables capture differences between dyad members but are the same on average for all dyads. For example, in different-sex couples, gender is a within-dyad variable, with one member assuming the role of “man” and one member being assigned the role of “woman.” Second, between-dyad variables are constant for dyad members but differ across dyads. For example, each dyad might be characterized by some degree of similarity, which could be expressed by a profile correlation coefficient and characterizes each dyad but differs between dyads (i.e., some couples are more similar than others). Third, mixed variables describe variation within and between dyads; for example, the scores of personality questionnaires differ within a dyad (i.e., between partners) and between couples. By definition, relationship duration is a between-dyad level variable as each dyad is characterized by a value representing the relationship duration, and dyads differ in how long their relationship lasts.

Study design

When researchers are interested in testing questions of stability, change, or effects of relationship duration, the study design and data analytic approach must be considered. First, the study design can be broadly distinguished between cross-sectional and longitudinal. The study design and available data affect the analytic approach, with cross-sectional data only allowing between-dyad comparisons, whereas longitudinal data allow for within- and between-dyad comparisons (i.e., considering the same participants’ data across time). The latter allows testing mixed-variable level effects and describing changes across time in participants’ scores within dyad members and between dyads.³ Secondly, one should consider whether relationship duration is tested in terms of a main effect or as a moderator because analyses and interpretations differ for the cross-sectional and longitudinal design. We will discuss these points in the following sections.

Illustrative example

For illustration, we consider an example that addresses a comparatively simple question: Is there an association between expressions in variables of relationship satisfaction and relationship duration? Table 1 displays sample data of 10 fictitious dyads (e.g., self-reports from questionnaires regarding satisfaction by partner A and B⁴). On the left-hand side of Table 1, a typical data pattern for cross-sectional designs and information on the dyads’ relationship duration (“Duration”) is displayed. On the right-hand side, a longitudinal design pattern is displayed, which extends the cross-sectional data (collected at T1) by repeated data collection at T2. In our example, data were collected with a 6-month time lag. Note that the longitudinal approach does not necessarily require a dyad-level variable characterizing the relationship duration, as the time lag between T1 and T2 naturally captures the time between the two measurements and reflects the change over time. Also note that the cross-sectional data only allow for estimating between-couple variance (column-wise), whereas longitudinal data can provide between- (column-wise) and within-person (row-wise) variance.⁵

Table 1.

Example data structures of cross-sectional and longitudinal data designs of dyad members (A and B) at Time 1 (T1) and 2 (T2).

	Cross-sectional			Longitudinal
						Partner A			Partner B
Dyad	Duration	A	B	Duration_T1	Duration_T2	Score_T1	Score_T2	d _within _(T2-T1)	Score_T1	Score_T2	d _within _(T2-T1)
1	55	2	3	55	61	2	3	1	3	3	0
2	30	2	2	30	36	2	3	1	2	3	1
3	16	3	5	16	22	3	3	0	5	4	−1
4	12	5	4	12	18	5	4	−1	4	4	0
5	14	5	5	14	20	5	5	0	5	5	0
6	9	1	2	9	15	1	2	1	2	1	−1
7	14	4	4	14	20	4	3	−1	4	4	0
8	22	2	3	22	28	2	3	1	3	3	0
9	52	4	2	52	58	4	3	−1	2	4	−2
10	44	4	5	44	50	4	4	0	5	5	0
σ²_between	σ²_Duration	σ²_A	σ²_B	σ²_Duration	σ²_Duration	σ²_A(T1)	σ²_A(T2)		σ²_B(T1)	σ²_B(T2)
r	—	−.08	−.26	—	—	.81			.75

Note. A = A’s score of dyad k. B = B’s score of dyad k. T1 and T2 denote two occasions of data collection (e.g., within a 6-month interval). In couple research, A and B might denote men and women providing self-ratings in relationship satisfaction; in interpersonal perception studies, A might be the target and provides a self-report while B might be an observer providing judgments of A. σ² = variance. d = mean difference.

Testing the “effects of relationship duration” based on cross-sectional data relies on between-person (or between-dyad) variance (σ²_between; Table 1). This approach examines the associations between relationship duration and A’s and B’s scores, with correlations computed along the columns. In our example, we find that A’s scores are unrelated to duration (r = −.08), whereas B’s scores are higher in dyads that reported less duration of their relationship than members of dyads with greater scores in relationship duration (r = −.26). These associations are also illustrated in the upper diagram of Figure 1. Such findings are often unwarrantedly interpreted as evidence for time-based processes at the within-person level (e.g., “with increasing relationship duration people feel less satisfied”), although interpretations can only be made in terms of between-couple comparisons (i.e., “people from couples who report higher relationship duration feel less satisfied in comparison to those from couples with shorter relationship duration”). In extension to the analysis of main effects of relationship duration, moderator analyses are also frequently employed. Therefore, one would compute a grouping variable on basis of “Duration” (e.g., couples are clustered into the groups of 0 = short relationship duration and 1 = long duration) and one would analyze whether an interaction effect exists between the Duration variable and a predictor variable of interest (e.g., a score in a personality questionnaire) when predicting an outcome such as satisfaction.⁶ A statistically significant interaction term would indicate an effect of “group” (i.e., findings differ depending on whether couples are denoted by short or long relationship durations).

Figure 1.

Illustration of Example Data (see Table 1) of Cross-Sectional (Upper Diagram) and Longitudinal Data (Lower Diagram) of Dyads Members (A and B). Note. In the cross-sectional example data, relationship duration (x-axis) is plotted against the satisfaction scores (y-axis), whereas in longitudinal analyses, changes are depicted with regard to the time points of data collection, here Time 1 (T1; x-axis) and Time 2 (T2; y-axis).

Longitudinal data allow testing effects of time at the within-person level. Therefore, data points that were collected at least on two occasions are correlated and this allows to compute within-person variance over time. As displayed in Table 1 (see “Longitudinal”), we would need multi-occasion data to address questions on the trajectories of scores over time. Testing the within-person correlation of the same participant in the same variable over time allows inferences about the development of expressions in a study variable over time. Therefore, data across rows capture within-person variance (see Table 1; σ²_within). In our example, we see that A’s and B’s scores are comparatively stable across time (r = .81 and .75 for A and B). This is also displayed in the lower diagram of Figure 1. Contrary to the conclusion made based on the cross-sectional data, we would not conclude that B’s ratings of satisfaction decline over time. Note that cross-sectional data are limited in addressing questions of stability and change over time as within-person change is not assessed, whereas longitudinal models are suited to address such questions. Thus, conclusions differ depending on the design used and whether comparisons are made between or within persons (or dyads).

Preliminary conclusions

Our simple example illustrated the differences between the approaches and the conclusions that can be derived. Note that studies often examine far more complex questions regarding relationship duration, for example, how dyadic indexes such as partner similarity or self-other agreement relate to relationship duration. Furthermore, note that changes over time might not only differ within and between persons but also consequently between dyads, thus showing differential trajectories of stability and change (Curran & Bauer, 2011; Lee et al., 2021).

The statistical issues of using cross-sectional data for statistical inferences on longitudinal developments have been extensively discussed over the years (e.g., Curran & Bauer, 2011; Galambos et al., 2020). Overall, formal mathematical evidence and analyses of simulated and empirical data show robust issues when mixing cross-sectional data with longitudinal analysis methods, such as robust parameter misestimation, to the point that findings from cross-sectional and longitudinal data sets contradict each other (e.g., Curran & Bauer, 2011; Maxwell & Cole, 2007; Maxwell et al., 2012). Similarly, this applies to testing longitudinal questions using cross-sectional data in dyadic research, providing statistical misestimation and erroneous conclusions.

While it seems that longitudinal models might be preferred from a statistical perspective, there are also some disadvantages in practice: The level of acquaintance at T1 is usually not the same across dyads, unless participants are sampled according to a criterion of relationship duration, for example, by studying unacquainted dyads (e.g., speed dating studies; Kerr et al., 2020) that get to know each other during the study interval or when recruiting dyads who know each other for a specific amount of time at T1. Hence, there is already some degree of variance across dyads on the baseline which might affect findings on effects of relationship duration, with dyads differing in their “starting points.” Standardizing the time lag between measurement points for all participants leads to the same degree of change in acquaintance across all dyads. While this approach has the merit that the change in acquaintanceship is controlled and the same across dyads, it is assumed that the change and its effects are linear. However, this assumption has not been supported empirically, as we will discuss in the “Empirical Examples” section. Also, it should be considered that the length of the time lag between T1 and T2 plays a role. For example, in the field of interpersonal perception, increases in accuracy (i.e., self-other agreement) are often reported in early phases of relationships, whereas effects are smaller when dyads are well-acquainted (Beer, 2020). Thus, short intervals (e.g., 8 weeks) in later phases might not be long enough to detect change or moderator effects of relationship duration, whereas shorter intervals are needed to detect fine-grained changes in early phases of relationships (e.g., Beer, 2020). Comparatively longer time lags would be needed to identify effects, particularly in dyads that are already well-acquainted. In comparison, cross-sectional studies often provide considerably higher variance in relationship duration than longitudinal studies and thus might be well suited to examine effects of relationship duration between couples. On the other hand, longitudinal studies provide the appropriate data basis for testing effects of relationship duration over time within dyads.

Empirical examples: Relationship satisfaction and judgment accuracy

After considering methodological issues, we will discuss example findings on the development of couples’ relationship satisfaction and interpersonal perception. The examples will highlight further practical issues for both cross-sectional and longitudinal approaches.

Romantic relationships

One of the most frequently studied questions in dyadic research concerns how relationship satisfaction develops over time. Most cross-sectional studies on the development of relationship satisfaction indicate a U-shaped curve “over time,” with a decrease in earlier years and an increase for longer relationship durations (e.g., Heiman et al., 2011). Studies using both cross-sectional and longitudinal analyses replicated the U shape in cross-sectional data, but longitudinal findings revealed a decrease in satisfaction over time (e.g., Mund & Johnson, 2021). Although most longitudinal studies in the field have replicated and supported the latter, alternative trajectories such as stability (e.g., Fallis et al., 2016; Vaillant & Vaillant, 1993) and increase (e.g., Gorchoff et al., 2008) have also been found. Interestingly, studies indicating stability of satisfaction over time have used comparatively smaller intervals between measurements (i.e., 1 to 5 years), whereas studies indicating change over time tracked the couples for longer intervals of up to 20 years. This highlights an important consideration in longitudinal studies: the choice of the measurement interval. It can be argued that when using comparatively short time lags, changes might be small and not detectable (Roberts et al., 2006). As discussed, the effects of change covered during a lag between time points might also depend on the dyads’ initial relationship duration at T1. Thus, longitudinal studies do not unconditionally provide the best estimate of change and its effects, but they do allow for tracking change over time.

Interpersonal perception

Studies on the acquaintanceship effect (i.e., accuracy of judgments increases with greater acquaintanceship) of personality judgments have also provided mixed findings, depending on the study design. Overall, the acquaintanceship effect has been supported by both between- and within-dyad analyses (for overviews see Beer, 2020; Kenny, 2020). First, cross-sectional between-dyad analyses have found positive but comparatively small effect sizes. For example, Lee and Ashton (2017) reported a correlation of r = 0.34 between years of acquaintanceship and self-other agreement in the broad HEXACO personality traits. Also, Biesanz et al. (2007) tested the accuracy of judgments on the Big Five personality traits among self-peer dyads and found a correlation of r = 0.18 between accuracy and relationship duration. And they conclude that, on average, accuracy increases by .09 for every 5-year increase in length of acquaintance. Schneider and colleagues (2010) provided non-linear analyses of associations between acquaintanceship and the self-other agreement on life satisfaction using cross-sectional data and found no robust increases in agreement after around 3 years of acquaintanceship. Secondly, longitudinal within-dyad analyses also revealed more differentiated patterns than the mere increase in accuracy over time. Studies tracking newly acquainted students over the course of several weeks have found substantial increases in accuracy. For example, Kurtz and Sherker (2003) reported a mean increase from r = 0.27 to 0.43 across 13 weeks, and this trend has been corroborated in other longitudinal studies (e.g., comparing 5-minute and 10-week intervals; Brown & Bernieri, 2017). In addition, Park et al. (1997) tested the accuracy of judgments for comparatively longer measurement delays, across 8 months, and did not find support for an association between relationship duration and accuracy in longitudinal data. One might argue that both designs captured a plateau of self-other agreement, and that accuracy does not increase linearly over time after observers have collected sufficient information for stable inferences (e.g., Beer, 2020; Kenny, 2020). Thus, although cross-sectional and longitudinal studies tend to converge concerning the acquaintanceship effect, within-dyad comparisons allow for fine-grained analyses and findings, detecting stagnation after a certain time and indicating that relationship duration becomes irrelevant for the outcome of interest with increasing relationship duration.⁷ These findings from the domains of accuracy and relationships highlight the complexity of the associations between relationship duration and outcomes, and the importance of the study design that is used, but also the selection of the time interval. Thus, there is no universal recommendation for the choice of time lags as the same lags used in accuracy research would not be informative on change in relationship satisfaction research and vice versa.

Be aware of conceptual issues of between-dyad comparisons

While longitudinal and cross-sectional approaches have their unique merits and disadvantages, we also want to discuss conceptual issues based on between-dyad comparisons regarding relationship duration. We discuss this issue using the example of relationship duration and age in romantic relationship research.

We argue that the theoretical status and the implicitly assumed equivalence of the meaning of “relationship duration” between dyads should be critically questioned: Dyads do not change uniformly, but rather they differ in how they change on the within- and between-dyad level, showing intra- and interindividual change patterns (e.g., Curran & Bauer, 2011; Gistelinck & Loeys, 2019; Roberts et al., 2006). These differential changes can be explained by numerous factors such as differences in shared environments (e.g., Caspi et al., 1992). For illustration, one might imagine two newly formed couples that are formally characterized by the same relationship duration (e.g., 6 months). Further, assume that one couple consists of younger partners (e.g., two 18-year-olds) and the other of two older partners (e.g., widowed 67-year-olds). Developmental differences between the dyad members probably distinguish couples whose members are formally at the same stage of their relationship. This notion receives support in that relationship duration is robustly correlated with third variables such as age (e.g., r = 0.49; the findings hold when testing subgroups of the younger and older 50% of the sample, respectively; Latagne & Furman, 2017). In a longitudinal study, Latagne and Furman (2017) disentangled the effects of age and relationship duration for numerous variables (e.g., jealousy, support, and negative interactions) and found the expected strong heterogeneity within and between dyads. They highlighted the complexity of differential developmental trajectories and concluded that “Short relationships in adolescence differ from short relationships in adulthood, but not in the same way that long relationships in adolescence differ from long relationships in adulthood” (p. 1747). Latagne and Furman’s findings show that although the duration of a relationship might be the same between couples, it may have different effects on couples depending on factors such as the dyad members’ age. Hence, relationship duration has no uniform effects that translate to all couples and the assumption that dyads are comparable in how they change over time is not supported (for similar findings in accuracy research, see Beer, 2020). Thus, empirical and conceptual considerations show that between-couple analyses regarding relationship duration should be interpreted cautiously. Irrespective of the implied comparability of the numerical value of a relationship duration variable, conceptual boundaries and the empirical heterogeneity of developments should be considered.

Recommendations and limitations

First of all, we do not suggest choosing either the longitudinal or cross-sectional approach over the respective other. As discussed, there are merits to each type of data collection and analyses; for example, when deriving initial overviews, particularly when exploring new fields, and especially when longitudinal data are not (yet) available, cross-sectional methods should be used. We advocate for deriving sound interpretations and acknowledging limitations (methodological, conceptual, and regarding generalizability) when interpreting findings on relationship duration using cross-sectional data. Such limitations include acknowledging the covered range of relationship durations in the study sample. Since dyadic studies often rely on specific samples with regard to age (e.g., when testing student samples comprising mostly younger participants with regard to age and relationship duration and when testing only elderly couples), variance in relationship duration variables is often low. Further, we encourage collecting sufficiently large subsamples when applying between-group comparisons to allow well-powered analyses. For example, Watson et al. (2000) compared self-other agreement coefficients in personality and affective traits between friendship dyads, dating couples, and married couples. On average, agreement for the Big Five personality traits was numerically highest in married couples (.56), followed by dating couples (.47) and friendship dyads (.41). Thus, differences between groups are comparatively small, and using these coefficients as a guideline, power analyses (G*Power; Faul et al., 2009; see https://osf.io/8w6cv/ for outputs) indicate that large samples are needed to detect the effects with 80% power and 5% type I error rate (one-tailed), namely, 642 dyads (1248 participants) to replicate the difference among married and friendship dyads, and 4468 dyads (8936 participants) to detect the difference between dating couples and friends with null hypothesis significance testing.

Longitudinal designs allow for assessing the variance components needed to model the effects of interest within and between dyad members and within and between dyads over time (e.g., Gistelinck & Loeys, 2019; Heck & Thomas, 2009). Further, numerous approaches to studying change tailored to specific dyadic research designs and questions are available (e.g., longitudinal APIM, Common Fate Model, Latent Change and Growth Models, the Dual Change Score Model, Social Relations Growth Model, or Dyadic Response Surface Analyses; for overviews see Gistelinck & Loeys, 2019; Gray & Ozer, 2019; Lee et al., 2021; Mund & Nestler, 2019; Nestler et al., 2017; Schönbrodt et al., 2018).

As discussed, researchers should critically consider the time interval between measurements, as effects should be expected to also depend on the covered time span. When possible, multiple measurements should be used to track potential changes with fine-grained analyses. The effects of length might depend on the study domain as temporal effects on relationship satisfaction occur in longer terms, whereas changes in accuracy occur in shorter terms, particularly in early phases of the relationship. Additionally, baseline levels at the beginning of data collection might affect the magnitude of change (e.g., Roberts et al., 2006), as differences in acquaintanceship at T1 might affect the magnitude of effects of change since dyads differ in how they change (e.g., Latagne & Furman, 2017; Lee et al., 2021). When researchers are interested in the description of change from a certain starting point, using a standardized entry point has been found fruitful, for example, studying newlyweds (e.g., Karney & Bradbury, 1995) or previously unacquainted persons (e.g., Kenny, 2020) to standardize the baseline of relationship duration. As with cross-sectional designs, aiming for high power is recommended and tools for dyadic power analysis can be used to derive sample size estimates when planning studies (Lafit et al., 2021; Lane & Hennes, 2018). However, known issues of longitudinal designs also exist in dyadic research, including selection effects and selective and natural dropout (e.g., Baltes, 1968; Gistelinck & Loeys, 2019; Heck & Thomas, 2009; Maxwell & Cole, 2007). In dyadic research, this is of course a concern when relationships are dissolved and data are systematically missing across time points. However, note that this might also be an issue in cross-sectional designs, as samples also comprise couples that will break up or “survive” in the long term. Further, methodological effects and artifacts of multiple testing and responding to the same questionnaires might affect responses (Greenwald, 1976; Shrout et al., 2018). Also, longitudinal designs require considerable effort and resources. We encourage researchers interested in longitudinal research to check whether data already exist that might be suited to address research questions with secondary data analyses (e.g., Trzesniewski et al., 2011). For example, data from large-scale studies such as the German Socio-Economic Panel Study and the Inter-university Consortium for Political and Social Research databases provide multi-wave data resources for dyadic data. Those data collections might be a good starting point for standard analyses on the topic of relationship duration.

Conclusion

In conclusion, we raised attention to cross-sectional and longitudinal methods for the analysis and interpretation of relationship duration in dyadic research. Considering and weighing these against each other when planning studies addressing the description and effects of relationship duration might help to extend the knowledge of the role of acquaintanceship and relationship duration for individuals and dyads. While we do not advocate for one approach, we recommend interpreting the findings in the light of each of the approaches’ merits and limitations. Finally, it must be mentioned that this primer mainly focused on the methodological perspective of relationship duration analyses. The literature shows the importance of the temporal dimension of dyadic research, and we encourage researchers to examine the processes that explain the effects of changes over time when drawing conclusions on effects of relationship duration.

Footnotes

Acknowledgment

We are grateful to David A. Kenny for his valuable comments and feedback on earlier versions of this manuscript.

Declaration of conflicting interests

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

Funding

The author(s) received no financial support for 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: This research is a theoretical article and was not pre-registered. No data were used in this research, but power simulations were computed and are available in the Open Science Framework under . No materials were used in this research.

ORCID iDs

Kay Brauer

Rebekka Sendatzki

René T. Proyer

Notes

References

Baltes

P. B.

(1968). Longitudinal and cross-sectional sequences in the study of age and generation effects. Human Development, 11(3), 145–171. https://doi.org/10.1159/000270604

Beer

(2020). Information as a moderator of accuracy in personality judgment. In Letzring

T.D.

Spain

J.S.

(Eds.), The Oxford handbook of accurate personality judgments (pp. 132–148). OUP. https://doi.org/10.1093/oxfordhb/9780190912529.013.9

Bierstetel

S. J.

Farrell

A. K.

Briskin

J. L.

Harvey

M. W.

Gable

S. L.

Ickes

Lin

W.-F.

Orina

M. M.

Saxbe

Simpson

J. A.

V. P.

Slatcher

R. B.

(2020). Associations between language style matching and relationship commitment and satisfaction: An integrative data analysis. Journal of Social and Personal Relationships, 37(8–9), 2459–2481. https://doi.org/10.1177/0265407520923754

Biesanz

J. C.

West

S. G.

Millevoi

(2007). What do you learn about someone over time? The relationship between length of acquaintance and consensus and self-other agreement in judgments of personality. Journal of Personality and Social Psychology, 92(1), 119–135. https://doi.org/10.1037/0022-3514.92.1.119

Brauer

Proyer

R. T.

(2020). Dyadic effects. In Zeigler-Hill

Shackelford

T.K.

(Eds.), Encyclopedia of personality and individual differences (pp. 1222–1226). Springer. https://doi.org/10.1007/978-3-319-28099-8_656-1

Brown

J. A.

Bernieri

(2017). Trait perception accuracy and acquaintance within groups: Tracking accuracy development. Personality and Social Psychology Bulletin, 43(5), 716–728. https://doi.org/10.1177/0146167217695557

Caspi

Herbener

E. S.

Ozer

D. J.

(1992). Shared experiences and the similarity of personalities: A longitudinal study of married couples. Journal of Personality and Social Psychology, 62(2), 281–291. https://doi.org/10.1037//0022-3514.62.2.281

Cook

W. L.

Kenny

D. A.

(2005). The actor–partner interdependence model: A model of bidirectional effects in developmental studies. International Journal of Behavioral Development, 29(2), 101–109. https://doi.org/10.1080/01650250444000405

Curran

P. J.

Bauer

D. J.

(2011). The disaggregation of within-person and between-person effects in longitudinal models of change. Annual Review of Psychology, 62, 583-619. https://doi.org/10.1146/annurev.psych.093008.100356.

10.

Fallis

E. E.

Rehman

U. S.

Woody

E. Z.

Purdon

(2016). The longitudinal association of relationship satisfaction and sexual satisfaction in long-term relationships. Journal of Family Psychology, 30(7), 822–831. https://doi.org/10.1037/fam0000205

11.

Faul

Erdfelder

Buchner

Lang

A. G.

(2009). Statistical power analyses using G* Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41(4), 1149–1160. https://doi.org/10.3758/BRM.41.4.1149

12.

Ferguson

L. W.

(1949). The value of acquaintance ratings in criterion research. Personnel Psychology, 2(1), 93–102. https://doi.org/10.1111/j.1744-6570.1949.tb01673.x

13.

Galambos

N. L.

Krahn

H. J.

Johnson

M. D.

Lachman

M. E.

(2020). The U shape of happiness across the life course: Expanding the discussion. Perspectives on Psychological Science, 15(4), 898–912. https://doi.org/10.1177%2F1745691620902428

14.

Gistelinck

Loeys

(2019). The actor–partner interdependence model for longitudinal dyadic data: An implementation in the SEM framework. Structural Equation Modeling, 26(3), 329–347. https://doi.org/10.1080/10705511.2018.1527223

15.

Gorchoff

S. M.

John

O. P.

Helson

(2008). Contextualizing change in marital satisfaction during middle age: An 18-year longitudinal study. Psychological Science, 19(11), 1194–1200. https://doi.org/10.1111/j.1467-9280.2008.02222.x

16.

Gray

J. S.

Ozer

D. J.

(2019). Comparing two models of dyadic change: Correlated growth versus common fate. Social Psychological and Personality Science, 10(7), 957–965. https://doi.org/10.1177/1948550618799066

17.

Greenwald

A. G.

(1976). Within-subjects designs: To use or not to use? Psychological Bulletin, 83(2), 314–320. https://doi.org/10.1037/0033-2909.83.2.314

18.

Heck

R. H.

Thomas

S. L.

(2009). An introduction to multilevel modeling techniques (2nd ed.). Routledge.

19.

Heiman

J. R.

Long

J. S.

Smith

S. N.

Fisher

W. A.

Sand

M. S.

Rosen

R. C.

(2011). Sexual satisfaction and relationship happiness in midlife and older couples in five countries. Archives of Sexual Behavior, 40(4), 741–753. https://doi.org/10.1007/s10508-010-9703-3

20.

Joel

Eastwick

P. W.

Allison

C. J.

Arriaga

X. B.

Baker

Z. G.

Bar-Kalifa

Bergeron

Birnbaum

G. E.

Brock

R. L.

Brumbaugh

C. C.

Carmichael

C. L.

Chen

Clarke

Cobb

R. J.

Coolsen

M. K.

Davis

de Jong

D. C.

Debrot

DeHaas

E. C

Wolf

(2020). Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies. Proceedings of the National Academy of Sciences, 117(32), 19061–19071. https://doi.org/10.1073/pnas.1917036117

21.

Karney

B. R.

Bradbury

T. N.

(1995). The longitudinal course of marital quality and stability: A review of theory, method, and research. Psychological Bulletin, 118(1), 3–34. https://doi.org/10.1037/0033-2909.118.1.3

22.

Kenny

D. A.

(2020). Interpersonal perception (2nd ed.). Guilford Press.

23.

Kenny

D. A.

Kashy

D. S.

Cook

W. L.

(2006). Dyadic data analysis. Guilford Press.

24.

Kerr

L. G.

Tissera

McClure

M. J.

Lydon

J. E.

Back

M. D.

Human

L. J.

(2020). Blind at first sight: The role of distinctively accurate and positive first impressions in romantic interest. Psychological Science, 31(6), 715–728. https://doi.org/10.1177/0956797620919674

25.

Kurtz

J. E.

Sherker

J. L.

(2003). Relationship quality, trait similarity, and self‐other agreement on personality ratings in college roommates. Journal of Personality, 71(1), 21–48. https://doi.org/10.1111/1467-6494.t01-1-00005

26.

Lafit

Sels

Adolf

J. K.

Loeys

Ceulemans

(2021, June 29). PowerLAPIM: An application to conduct power analysis for longitudinal actor-partner interdependence models that include quadratic effects. PsyArXiv https://doi.org/10.31234/osf.io/mnce4

27.

Lane

S. P.

Hennes

E. P.

(2018). Power struggles: Estimating sample size for multilevel relationships research. Journal of Social and Personal Relationships, 35(1), 7–31. https://doi.org/10.1177/0265407517710342

28.

Latagne

Furman

(2017). Romantic relationship development: The interplay between age and relationship length. Developmental Psychology, 53(9), 1738–1749. https://doi.org/10.1037/dev0000363.

29.

Lee

Ashton

M. C.

(2017). Acquaintanceship and self/observer agreement in personality judgment. Journal of Research in Personality, 70, 1-5. https://doi.org/10.1016/j.jrp.2017.05.001.

30.

Lee

T. K.

Wickrama

K. A. S.

O'Neal

C. W.

(2021). Modeling longitudinal dyadic processes in family research. Journal of Family Psychology, 35(7), 994–1006. https://doi.org/10.1037/fam0000862

31.

Lenhausen

M. R.

van Scheppingen

M. A.

Bleidorn

(2021). Self–other agreement in personality development in romantic couples. European Journal of Personality, 35(6), 797–813. https://doi.org/10.1177/0890207020969011

32.

Luo

(2017). Assortative mating and couple similarity: Patterns, mechanisms, and consequences. Social and Personality Compass, 11(8), Article e12337. https://doi.org/10.1111/spc3.12337

33.

Maxwell

S. E.

Cole

D. A.

(2007). Bias in cross-sectional analyses of longitudinal mediation. Psychological Methods, 12(1), 23–44. https://doi.org10.1037/1082-989x.12.1.23

34.

Maxwell

S. E.

Cole

D. A.

Mitchell

M. A.

(2012). Bias in cross-sectional analyses of longitudinal mediation: Partial and complete mediation under an autoregressive model. Multivariate Behavioral Research, 46(5), 816–841. https://doi.org/10.1080/00273171.2011.606716.

35.

Mund

Johnson

M. D.

(2021). Lonely me, lonely you: Loneliness and the longitudinal course of relationship satisfaction. Journal of Happiness Studies, 22(2), 575–597. https://doi.org/10.1007/s10902-020-00241-9

36.

Mund

Nestler

(2019). Beyond the Cross-Lagged Panel Model: Next-generation statistical tools for analyzing interdependencies across the life course. Advances in Life Course Research, 41, 100249. https://doi.org/10.1016/j.alcr.2018.10.002.

37.

Nestler

Geukes

Hutteman

Back

M. D.

(2017). Tackling longitudinal round-robin data: A Social Relations Growth Model. Psychometrika, 82(4), 1162–1181. https://doi.org/10.1007/s11336-016-9546-5

38.

Park

Kraus

Ryan

C. S.

(1997). Longitudinal changes in consensus as a function of acquaintance and agreement in liking. Journal of Personality and Social Psychology, 72(3), 604–616. https://doi.org/10.1037//0022-3514.72.3.604

39.

Proulx

C. M.

Helms

H. M.

Buehler

(2007). Marital quality and personal well‐being: A meta‐analysis. Journal of Marriage and Family, 69(3), 576–593. https://doi.org/10.1111/j.1741-3737.2007.00393.x

40.

Pusch

Schönbrodt

F. D.

Zygar-Hoffmann

Hagemeyer

(2021). Perception of communal motives in couples: Accuracy, bias, and their associations with relationship length. Journal of Research in Personality, 91, 104060. https://doi.org/10.1016/j.jrp.2020.104060.

41.

Roberts

B. W.

Walton

K. E.

Viechtbauer

(2006). Patterns of mean-level change in personality traits across the life course: A meta-analysis of longitudinal studies. Psychological Bulletin, 132(1), 1–25. https://doi.org/10.1037/0033-2909.132.1.1

42.

Schneider

Schimmack

Petrican

Walker

(2010). Acquaintanceship length as a moderator of self-informant agreement in life-satisfaction ratings. Journal of Research in Personality, 44(1), 146–150. https://doi.org/10.1016/j.jrp.2009.11.004

43.

Schönbrodt

F. D.

Humberg

Nestler

(2018). Testing similarity effects with dyadic response surface analysis. European Journal of Personality, 32(6), 627–641. https://doi.org/10.1002/per.2169

44.

Shrout

P. E.

Stadler

Lane

S. P.

McClure

M. J.

Jackson

G. L.

Clavél

F. D.

Iida

Gleason

M. E. J.

J. H.

Bolger

(2018). Initial elevation bias in subjective reports. Proceedings of the National Academy of Sciences, 115(1), E15–E23. https://doi.org/10.1073/pnas.1712277115

45.

Terman

L. M.

(1938). Psychological factors in marital happiness. McGraw-Hill.

46.

Trzesniewski

K. H.

Donnellan

M. B.

Lucas

R. E.

(2011). Secondary data analysis: An introduction for psychologists. APA. https://doi.org/10.1037/12350-000

47.

Vaillant

C. O.

Vaillant

G. E.

(1993). Is the U-Curve of marital satisfaction an illusion? A 40-year study of marriage. Journal of Marriage and Family, 55(1), 230–239. https://doi.org/10.2307/352971

48.

Watson

Hubbard

Wiese

(2000). Self–other agreement in personality and affectivity: The role of acquaintanceship, trait visibility, and assumed similarity. Journal of Personality and Social Psychology, 78(3), 546–558. https://doi.org/10.1037/0022-3514.78.3.546

49.

Weigel

D. J.

Ballard‐Reisch

D. S.

(1999). The influence of marital duration on the use of relationship maintenance behaviors. Communication Reports, 12(2), 59–70. https://doi.org/10.1080/08934219909367711