Interpersonal Perception of Adult Playfulness at Zero‐Acquaintance: A Conceptual Replication Study of Self‐Other Agreement and Consensus, and an Extension to Two Accuracy Criteria

ABSTRACT

Objective

We replicated and extended previous research examining the accuracy of judgments of four facets of adult playfulness (Other‐directed, Lighthearted, Intellectual, and Whimsical; OLIW) at zero‐acquaintance.

Method

We conducted a conceptual replication study. One hundred sixty targets provided self‐ratings for the OLIW facets, textual self‐descriptions (≤ five sentences), daily self‐ratings of playfulness for 14 consecutive days, and ratings by knowledgeable others. Six unacquainted judges provided rated targets' playfulness based on their self‐descriptions. We replicated findings on trait‐wise self‐other agreement (SOA) and consensus and extended prior research by testing SOA for profiles of the four facets and two distinct accuracy criteria (i.e., targets' diary data and aggregates of targets' self‐reports and those from knowledgeable others).

Results

All interpersonal perception indicators showed that facets of playfulness can be perceived above chance (SOA ≥ 0.26; consensus ≥ 0.29, accuracies ≥ 0.16). SOA extends from single facets to profiles, also when controlling for stereotype effects.

Conclusions

Playfulness can be accurately observed from minimal textual information at zero acquaintance. Our study highlights the robustness of findings on the interpersonal perception of playfulness across samples and methods, and degrees of acquaintanceship. We discuss implications for playfulness in social relationships.

1. Introduction

There is increasing interest in the study of adult playfulness (Bittermann et al. 2021), particularly with an emphasis on its role in social relationships (see Brauer, Proyer, and Chick 2021, for an overview). A crucial question to consider is whether people can accurately infer others' playfulness. Initial evidence supported the notion that playfulness can be accurately perceived at different degrees of acquaintanceship (Brauer, Sendatzki, and Proyer 2024b), including at zero‐acquaintanceship. To our knowledge, only one study addressed the latter. Proyer and Brauer (2018) asked 144 targets to provide short textual self‐descriptions, which were then presented to unacquainted judges to assess their playfulness. Playfulness was accurately perceived in terms of self‐other agreement and inter‐rater agreement (consensus). In the present study, we aimed to replicate these findings and extend the study of interpersonal perception of playfulness in a conceptual replication study in two ways. First, we expanded our analyses to include two additional accuracy criteria: (a) aggregates of targets' and knowledgeable others' reports of playfulness, and (b) targets' aggregated reports of daily playful behaviors (aggregated over 14 days). Second, we extended the analytic approach from the single facets of playfulness to the profiles of the four facets by using variable‐ and person‐centered approaches, allowing to consider a more comprehensive view of judges' accuracy.

1.1. Adult Playfulness

Playfulness as a personality trait has received attention in classics of personality psychology such as in Murray (1938; the need for play) and in Cattell (1950; playfulness is represented in his principal clusters of personality). More current approaches have acknowledged the importance of intellectual types of playfulness in adults and see playfulness as an individual differences variable that allows people to (re)frame situations in a way such that they are experienced as personally interesting, and/or entertaining, and/or intellectually stimulating (Proyer 2017). A frequently used structural model is the so‐called OLIW model—an acronym of four facets of playfulness: Other‐directed (i.e., using one's playfulness in social relationships; e.g., using nicknames for one's partner), Lighthearted (i.e., seeing life as a game rather than a battlefield; e.g., preferring to improvise over planning ahead), Intellectual (i.e., using one's playfulness to solve problems; e.g., researchers liking to play around with data), and Whimsical (i.e., enjoying extravagant hobbies and people; e.g., interests that are considered unusual or flamboyant; Proyer 2017).

Research using this model has extended knowledge of the differential associations between facets of playfulness and outcomes such as physical and mental health, well‐being, and creativity, to name a few (e.g., Brauer et al. 2022; Parker et al. 2023; Proyer et al. 2018). Also, there is increasing evidence that the findings concerning mental health, satisfaction across life domains, and well‐being extend to middle‐ and older age when testing individuals and couples between 50 and 98 years of age (Brauer et al. 2023; Brauer, Stumpf, and Proyer 2024). Additionally, a randomized placebo‐controlled study found that playfulness can be enhanced through targeted interventions (Proyer et al. 2021). Moreover, increases in playfulness were associated with enhanced well‐being and reduced depressive symptoms, and these positive effects persisted for up to 12 weeks after the training intervention.

Playfulness is important for the formation and maintenance of social relationships and in sexual selection processes (see Brauer, Proyer, and Chick 2021, for an overview). The “signal theory of adult playfulness” suggests that playfulness in men signals nonaggressiveness to women whereas playfulness in women signals youth (fecundity) to men and thereby contributes to partner choice for long‐term (heterosexual) romantic relationships (Chick 2001). Studies testing preferences for traits in romantic partners show that playfulness is among the most desired traits in US and German‐speaking samples (Chick, Yarnal, and Purrington 2012; Chick et al. 2020; Proyer and Wagner 2015), thus supporting the notion that playfulness could signal underlying qualities that people seek in a partner. Finally, Actor‐Partner Interdependence Model analyses of 211 couples showed partners to be similar in their playfulness, and that playfulness showed robust associations to facets of relationship satisfaction in actors. In some cases, this also translated into partner effects (e.g., for Other‐directed playfulness). These findings replicated well in 116 middle‐ and older‐aged couples (Brauer et al. 2023), showing that the role of playfulness remains relevant at older age. Recently, findings on the similarity have been replicated and extended to same‐gender couples (Brauer, Sendatzki, and Proyer 2024b).

1.2. Interpersonal Perception of Adult Playfulness

Accurately perceiving others' traits plays a crucial role in building and maintaining positive social relationships. Such perceptions enable people to form expectations about others' behaviors and tailor their own interactions accordingly. Interpersonal perception describes the process of judging a target person's expressions in traits such as intelligence, personality, and depressiveness and is essential for navigating the complexities of human interactions (see Kenny 2020, for an overview). Research has demonstrated that most traits can be perceived accurately above chance, even when only minimal information is available (Kenny 2020). The implications of accurate personality trait perceptions extend to real‐life outcomes, including the formation of long‐term relationships and romantic attraction (e.g., Human et al. 2013; Kerr et al. 2020). Moreover, although perceptions by knowledgeable others do not match perfectly with targets' self‐perceptions, they often hold incremental validity to predict important life outcomes (Luan et al. 2019; see Connelly and Ones 2010, for an overview) because others' perceptions provide external insights into targets' personality traits that are less accessible to self‐views. Considering the role of playfulness in social relationships, efforts to extend the knowledge about the accuracy of judgments of adult playfulness have been made.

Funder and West (1993) proposed three approaches to analyzing interpersonal perceptions: self‐other agreement (SOA; i.e., a correlation between self‐reports and judgments), consensus (inter‐judge agreement between k ≥ 2 judgments), and accuracy (a correlation between judgments and criterion variables such as aggregates of targets' self‐ and peer reports or behavioral criteria such as grades or behavioral data). Research on playfulness has primarily focused on SOA. For the OLIW facets, SOA coefficients ranged between 0.44 (Other‐directed/Intellectual) and 0.57 (Whimsical) among 226 well‐acquainted dyads composed of couples and family members (Proyer 2017), between 0.46 (Intellectual) and 0.55 (Whimsical) in 141 self‐peer dyads (Proyer et al. 2018), between 0.33 (Intellectual) and 0.58 (Lighthearted) among 77 couples (Proyer, Brauer, et al. 2018), and between 0.27 (Other‐directed) and 0.52 (Whimsical) among 166 dyads who completed the 12‐item brief OLIW questionnaire (Proyer, Brauer, and Wolf 2020). Although these findings suggest that playfulness can be accurately perceived regardless of the level of acquaintanceship, Brauer, Sendatzki, and Proyer (2024a) systematically tested whether length of acquaintanceship and type of acquaintanceship (i.e., family vs. couples vs. friends) mattered for the accuracy of the judgments in the four facets. In short, they found robust SOA coefficients between 0.40 (Intellectual) and 0.59 (Lighthearted) and negligible to minor effects of acquaintanceship among 658 dyads. Moreover, they extended their analysis to testing SOA across the full profiles of the OLIW facets. They found profile agreement correlations of 0.50 when analyzing raw profiles and a distinctive profile SOA correlation of 0.38 after adjusting for stereotype effects. Taken together, the findings show that fine‐grained facets of playfulness are on average accurately perceived at different degrees of acquaintanceship and that SOA coefficients align with findings on other personality traits (Connelly and Ones 2010).

1.2.1. The Zero‐Acquaintanceship Approach to Interpersonal Perception

Zero‐acquaintance studies examine interpersonal perception between unacquainted pairs of targets and judges (Albright, Kenny, and Malloy 1988; Kenny 2020). This approach allows for high standardization regarding the duration of acquaintanceship (i.e., zero; only unacquainted dyads of targets and perceivers) and the quantity and quality of information that judges receive to base their inferences on. For example, the interpersonal perception of personality traits has been studied when judges based their inferences on information such as targets' photographs, social media profiles, and video‐taped self‐introductions, to name a few (e.g., Borkenau and Liebler 1993; Osterholz, Mosel, and Egloff 2023; for overviews, see Connelly and Ones 2010, and Kenny 2020). Even under the availability of minimal information, observers' judgments typically show some degree of validity exceeding zero (Kenny 2020). For example, Connelly and Ones' (2010) meta‐analysis of zero‐acquaintance studies of the Big Five personality traits found SOA coefficients between 0.08 and 0.22, and consensus coefficients between 0.23 and 0.40 for the average agreement between two judges. It has been argued that targets produce information—so‐called thin slices of behavior—that allows judges to provide somewhat accurate inferences, even at zero‐acquaintance (Kenny 2020).

Among others, zero‐acquaintanceship research has used targets' textual self‐descriptions as a source of information for personality judgments because there is robust evidence that individual differences in personality traits are reflected in language use (Pennebaker and King 1999; Tong et al. 2020; see Tausczik and Pennebaker 2010, for an overview). As in other “thin slices” studies, observers' judgments based on targets' textual self‐descriptions are accurate above chance, and this has been shown for numerous traits. For example, Borkenau et al. (2016) found positive SOA and consensus when judges based their ratings on targets' self‐descriptions regarding their family, hobbies, friends, future plans, and academic studies; Brauer and Proyer (2020) found that targets' short self‐descriptions provided information for accurate judgments of three dispositions toward ridicule and being laughed at in the sense of SOA, consensus, and overlap between judgments and targets' diary data; Lau et al. (2021) showed that targets' textual self‐descriptions contained information to allow for accurate judgments (SOA and consensus) of cheerfulness; and Körner et al. (2024) showed that dominance, prestige, and power can be inferred from targets' short self‐descriptions.

Beyond examining levels of consensus, agreement, and accuracy, it is also interesting to examine the types of cues that people use to make those judgments. Brunswik's (1956) lens model provides a framework for examining the extent to which specific cues are “valid” reflective of a given trait, and the extent to which judges utilize valid or invalid cues when judging that trait. When using targets' textual self‐descriptions as source of information for external judgments, the texts can be analyzed with the Linguistic Inquiry and Word Count (LIWC) software (Pennebaker and King 1999). The LIWC scans texts per 90 categories describing basic language features (e.g., word count, use of pronouns) and psychological categories (e.g., positive emotions; Pennebaker and King 1999). The LIWC‐based count data for the word categories are considered as cues and are used to compute lens model analyses. For example, Hirsh and Peterson (2009) reported an average correlation of 0.23 between LIWC counts and Big Five personality traits, on average. Rodriguez, Holleran, and Mehl (2010) found that targets' self‐reported depressiveness related to using less positive emotion words (r = −0.38; cue validity) and a lower frequency of using positive emotion words also related to judgments of targets' depressiveness (r = −0.48; cue utilization). While in this example, there is robust overlap between the validity and utilization of cues (so‐called sensitivity), the lens model also allows to identify if judges misinterpret cues by utilizing cues that are not valid or dismiss valid cues (see also Funder 1995, and Breil et al. 2021, for an overview). Thus, in our study we will examine all three types of interpersonal perceptions as well as the cues that are used making some information more concise, specifically in the context of playfulness.

1.2.2. Impressions of Playfulness at Zero‐Acquaintance

To our knowledge, only one study had investigated the interpersonal perception of adult playfulness in a zero‐acquaintance setting so far. In this study, 144 targets provided short textual self‐descriptions (≤ five sentences) and self‐reports of playfulness (Proyer and Brauer 2018). The self‐descriptions provided by the participants were presented to five judges, who independently evaluated each participant's playfulness. SOA analyses showed robust target‐judge agreement with coefficients between 0.31 (Intellectual) and 0.50 (Lighthearted) when analyzing the SOA between targets' self‐reports of playfulness and aggregated data of the five judges. It is important to note that SOA coefficients, as well as consensus and accuracy correlations, are overestimated when using aggregated judgments based on their comparatively high reliability. To address the issue that aggregated judgments robustly exceed the reliability of single judgments, one also computes separate analyses for each judge and then averaged the results. In Proyer and Brauer, the average single‐rater SOAs were between 0.21 (Intellectual) and 0.37 (Lighthearted). Furthermore, they found robust inter‐judge agreement (consensus), between 0.69 (Intellectual) and 0.77 (Other‐directed and Lighthearted) among the five judges, and coefficients between 0.31 (Intellectual) and 0.40 (Other‐directed) between two random judges. In conclusion, Proyer and Brauer's findings showed that the OLIW facets can be perceived comparatively well from targets' short self‐descriptions.

Lens model analyses (Brunswik 1956) of the linguistic cues that were extracted with LIWC software showed robust overlap between cue validity (i.e., correlations between LIWC counts and targets' self‐reports of playfulness) and cue utilization (i.e., correlations between LIWC counts and judges' ratings of playfulness); the coefficients were between 0.42 (Whimsical) and 0.62 (Lighthearted). The only exception was Intellectual playfulness with lower sensitivity (0.19), suggesting that there was only a minor relationship between the validity of cues and the judges utilizing them. Analyses of selected LIWC categories showed only minor relations between targets' playfulness and word use (e.g., descriptions of those high in Other‐directed contained more first‐person plural words). Findings on cue validity have replicated well when analyzing descriptions of playfulness by 264 adults with the LIWC (Brauer, Sendatzki, and Proyer 2022). In addition, analyses of cue utilization showed what cues judges used to infer playfulness. For example, they perceived targets to be more playful when they used positive emotion words, although the linguistic cue was invalid (i.e., the positive emotion word use did not relate to targets' playfulness).

1.3. The Present Study

Proyer and Brauer's (2018) study provided initial evidence for the notion that facets of adult playfulness can be accurately perceived from textual self‐descriptions at zero‐acquaintance, but replication and extension are desirable. To address these aims, we conducted a conceptual replication study of Proyer and Brauer. As in the latter study, we asked targets to provide short textual self‐descriptions that we presented to judges who provided their impressions of targets' playfulness expression. We extended the design from the Proyer and Brauer's study by additionally collecting (a) ratings about targets' playfulness from their knowledgeable others and (b) daily diary ratings about playful behaviors from the targets.

This approach allowed us to replicate and extend the analyses from the original study in three major directions, namely, (1) replicating findings of SOA and consensus, (2) expanding the knowledge about agreement from facets to profiles and beyond normativeness, and (3) testing two accuracy criteria by means of (a) target‐peer report aggregates and (b) aggregated diary data concerning their overlap with ratings of playfulness at zero‐acquaintance.

1.3.1. Conceptual Replication

We tested whether Proyer and Brauer's (2018) findings of the SOA and consensus for the four OLIW facets would replicate in independent samples of targets and judges using the same instructions and materials. Also, we employed the LIWC software for lens model analyses (Brunswik 1956) and tested whether sensitivity coefficients (i.e., overlap between cue validity and cue utilization of linguistic cues) were in the same range as in Proyer and Brauer. Contrary to the original study, we asked a sample of judges to provide ratings for all targets using a brief questionnaire of the four facets (OLIW‐S; Proyer, Brauer, and Wolf 2020). Replication of effects using the approach of conceptual replication increases trust in the findings and supports their generalizability.

1.3.2. Expanding Prior Findings on Judgments of Playfulness at Zero‐Acquaintance

1.3.2.1. Extending Agreement From Facets to Profiles

We extended the analysis of SOA from testing agreement for the single OLIW facets to testing the target‐judge agreement regarding the full profiles of the OLIW facets. The OLIW model assumes within‐person variability between the four facets—to the degree that computing a total score out of the four facets is not recommended. For example, someone might show high expressions in Other‐directed playfulness, average expressions in Lighthearted and Intellectual playfulness, and low Whimsical playfulness. Accordingly, there are individual differences in profiles of expressions in the four facets and it is of interest whether judges' accuracy extends from single facets to the full profiles of expressions in the four facets. Further, the literature has shown that profile‐based agreement and profile‐similarity indexes are more robust predictors of outcomes in dyadic research than trait‐wise indexes, reflecting the notion that holistic agreement is based on a more comprehensive understanding of others than considering only single traits or facets (Rogers, Wood, and Furr 2018). Hence, we aim to describe the degree of profile agreement for the OLIW facets at zero‐acquaintance in the present study.

Hall et al. (2018) discussed that trait‐wise versus profile agreement capture different aspects and processes of interpersonal perception: While trait‐wise SOA is person‐centered and describes how well judges can discriminate between targets on a certain trait, profile analyses are variable‐centered and describe the “ability to distinguish the relative levels of different traits within a target person” (Hall et al. 2018, 221). Hence, testing both approaches allowed us to evaluate whether judges' accurate perceptions extend from single facets to the full profiles of the OLIW facets. Additionally, the profile‐approach allowed us to control for stereotypical responses and desirability by decomposing each profile into a normative component (reflecting the stereotypical profile of an average person) and a distinctive profile (Furr 2008; Wood and Furr 2016). The latter controls for stereotype effects, as it describes how an individual profile deviates from the average profile. While Brauer, Sendatzki, and Proyer (2024a) found profile SOA coefficients for the OLIW facets of 0.50 (raw profiles) and 0.37 (distinctive profiles) in 658 well‐acquainted dyads, the present study was the first to examine profile SOA for playfulness at zero‐acquaintance. Typically, distinctive profile SOA for narrow traits (e.g., dispositions toward ridicule and being laughed at; Brauer and Proyer 2020) is in the range of 0.10 to 0.20 at zero‐acquaintance.

1.3.2.2. Accuracy Criteria

Finally, we explored Funder and West's (1993) accuracy criterion, which describes the analysis of the overlap between observers' judgments and targets' trait expressions from other sources than self‐reports. This approach aims to mitigate potential biases, such as self‐serving bias, that may influence self‐reported trait expressions and might include external indicators such as school grades, supervisor ratings, and diary data. However, implementing this accuracy criterion necessitates collecting additional data beyond self‐reports and observer ratings, which contributes to its relative rarity in research (Connelly and Ones 2010).

A frequently used approach to accuracy is computing a composite of targets' self‐reports and reports from knowledgeable others, which is then correlated with observers' ratings (e.g., Borkenau et al. 2016; Osterholz, Mosel, and Egloff 2023). There is solid evidence that such aggregates provide a good estimate of the “true” expression of targets' personality traits (Hofstee 1994). In the present study, we analyzed two distinct accuracy criteria to provide estimates of targets' playfulness beyond their self‐reports: (1) aggregating their self‐ratings with those of knowledgeable others, and (2) collecting targets' daily diary data of their playful behaviors across 14 days using a previously tested behavior record (BR) of playfulness (Proyer 2017). In addition to providing an accuracy criterion for studying interpersonal perception, using a BR addressed the call for studying behavior‐based trait estimates that go beyond cross‐sectionally assessed self‐reports to contribute to increase confidence in findings on individual differences (see Furr 2009). Connelly and Ones' (2010) meta‐analysis showed that accuracy correlations (i.e., correlations between other‐ratings and targets' trait expression according to external criteria such as school grades and job performance) are typically relatively low when considering the broad Big Five traits (e.g., between 0.06 and 0.23 in relation to job performance), and research using a similar design testing the overlap between observer reports and targets' aggregated diary data has also found numerically small coefficients (e.g., between 0.09 and 0.17 for dispositions toward ridicule and being laughed at; Brauer and Proyer 2020).

2. Method

2.1. Participants

2.1.1. Targets

Our target sample comprised 160 German‐speaking participants (136 women, 22 men, and two did not indicate their gender). Their mean age was 23.9 years (SD = 14.6, median = 21.0). Most (95.0%) were university students from numerous fields, and the remainder were employed (n = 6), in vocational training (n = 1), or job‐seeking (n = 1). Their educational level was high, as 72.5% held a university‐qualifying high school diploma (an “Abitur”), 19.4% held a bachelor's degree, 3.1% a master's degree, 3.1% completed vocational training, and 1.8% held a high school diploma.

2.1.2. Knowledgeable Others

One hundred fifty knowledgeable others provided informant ratings of targets. Of these, 42.0% were partners, 37.3% friends, 10.6% parents, 6.3% siblings, and 2.5% extended family. Seventy‐seven informants were women, 72 were men, and one did not indicate their gender. Their mean age was 27.1 years (SD = 10.9, median = 23.0). The targets and informants knew each other for an average of 8.3 years (SD = 7.8, median = 5 years). Also, informants indicated how well they knew the target on a 10‐point scale (1 = not at all; 10 = very well), showing high acquaintanceship (M = 9.0, SD = 1.09; all ≥ 6.0).

2.1.3. Judges

We recruited six judges (three men, three women) between 18 and 67 years of age (M = 34.0, SD = 20.1, median = 23.0). Their educational status was high; three held a bachelor's degree, two held a university diploma (a “Diplom,” equivalent to a master's degree), and one held an Abitur. Three were students from fields other than psychology, one was a working professional, one was a person who did voluntary service, and one was a retiree.

2.2. Instruments

2.2.1. OLIW Questionnaire

We used the OLIW questionnaire (Proyer 2017) to assess four facets of playfulness. Targets completed the full 28‐item self‐report version that assesses each facet with seven items, and knowledgeable others and judges provided their ratings with the 12‐item brief informant rating version (three items per scale; OLIW‐S; Proyer, Brauer, and Wolf 2020). In the informant rating form, items are worded in the third‐person (e.g., self‐rating: “I use my playfulness to cheer others up”; peer rating: “He/she uses their playfulness to cheer others up;” see ESM A for sample items of all scales). Participants provide their ratings on a 7‐point Likert‐type rating scale (1 = strongly disagree; 7 = strongly agree). There is robust evidence for the reliability of the OLIW and OLIW‐S (e.g., retest‐correlations ≥ 0.67 for 3‐month and ≥ 0.74 for 1‐month intervals), and the validity has been supported by showing the expected overlap with self‐reported outcomes in the nomological net, self‐peer overlap, and measurement invariance, and a well replicable four‐factorial solution across self‐ and other reports (e.g., Brauer, Sendatzki, and Proyer 2024a; Proyer 2017; Proyer, Brauer, and Wolf 2020; Rubinstein et al. 2023; for IRT analyses, see also Davis and Boone 2021). In the present study, the internal consistencies were α/ω = 0.73/0.74 (Other‐directed), 0.77/0.77 (Lighthearted), 0.73/0.74 (Intellectual), and 0.83/0.84 (Whimsical) for the targets. Judges' inter‐rater agreement reliabilities were between 0.71 (Whimsical) and 0.78 (Lighthearted; see Table 1).

TABLE 1.

Interpersonal perception indexes of facets of adult playfulness at zero acquaintance.

	Consensus		Self‐other agreement		Accuracy
					Self‐peer aggregate		Behavior record
	Aggregated	Single	Aggregated	Single	Aggregated	Single	Aggregated	Single
Other‐directed	0.73 [0.64, 0.79]	0.31 [0.23, 0.39]	0.33 [0.18, 0.46]	0.26 [0.20, 0.32]	0.31 [0.17, 0.44]	0.24 [0.18, 0.30]	0.35 [0.19, 0.49]	0.27 [0.21, 0.34]
Lighthearted	0.78 [0.70, 0.84]	0.37 [0.28, 0.47]	0.51 [0.38, 0.62]	0.37 [0.28, 0.45]	0.44 [0.29, 0.57]	0.32 [0.24, 0.40]	0.30 [0.15, 0.44]	0.22 [0.16, 0.28]
Intellectual	0.72 [0.64, 0.78]	0.30 [0.23, 0.37]	0.44 [0.30, 0.55]	0.27 [0.14, 0.35]	0.41 [0.28, 0.53]	0.25 [0.12, 0.32]	0.33 [0.18, 0.47]	0.21 [0.11, 0.28]
Whimsical	0.71 [0.62, 0.78]	0.29 [0.21, 0.37]	0.38 [0.23, 0.51]	0.26 [0.21, 0.30]	0.38 [0.23, 0.50]	0.25 [0.21, 0.30]	0.16 [−0.01, 0.32]	0.10 [0.02, 0.18]

Note: Aggregated = aggregated judgments, Single = average single judge. Bootstrapped 95% confidence intervals (k = 5000 samples) in brackets.

2.2.2. Behavior Record

Targets provided daily ratings of their playful behaviors during the past 24 h based on Proyer's (2017) playfulness BR. The BR contains statements of 22 behaviors that indicate playfulness in adults. The playfulness BR was developed in line with Wu and Clark's (2003) recommendations and psychometrically examined before undergoing validity analyses in independent samples, which showed robust evidence for the convergent and discriminant validity regarding self‐report questionnaires and established behavioral records of playfulness, aggression, impulsiveness, exhibitionism, and narcissism (see Proyer 2017). A sample item is “Today, I have engaged in a playful activity,” and participants indicate whether they showed the behavior during the past 24 h (1 = no, 2 = yes). As recommended by Wu and Clark, we collected these daily ratings on 14 consecutive days and included the data of participants who provided ratings on at least 10 days and aggregated the ratings to a mean score. The daily ratings converged well across the 14 days (ICC [1,14] = 0.89, 95% CI [0.86, 0.92]), supporting the reliability of the BR measure. We replicated and extended Proyer's (2017) findings regarding the BR's validity, as we found robust correlations between the BR and (1) targets' self‐reports in the OLIW questionnaire (rs between 0.34 and 0.56, ps < 0.001), (2) peer reports for the OLIW scores (rs between 0.21 and 0.26, ps ≤ 0.015; exception: Whimsical playfulness, r = 0.05, p = 0.545), and (3) self‐ and peer reports for the five‐item Short Measure of Adult Playfulness (SMAP; Proyer 2012), which assesses an easy onset and frequent display of playfulness (r _{self‐report} = 0.52, r _{peer report} = 0.56, ps < 0.001). All coefficients of these supplementary analyses can be found in ESM B.

2.3. Procedure

2.3.1. Targets and Knowledgeable Others

We recruited targets through online advertisements on our academic department's website and billed the study as “research on personality in daily life.” Inclusion criteria were speaking German fluently, being ≥ 18 years of age, and being willing to complete short questionnaires every evening for 14 consecutive days. Targets completed an online questionnaire where they provided demographic information, wrote a short self‐description containing five or less sentences in accordance with Proyer and Brauer (2018), and completed the OLIW (Proyer 2017) and SMAP (Proyer 2012) questionnaires. The instruction to the self‐description was “In this first section, we would like to ask you to describe yourself freely using up to five sentences. There are no guidelines for this task, except for the maximum number of five sentences.” Finally, targets were asked to forward the link to the informant rating questionnaire to one knowledgeable other, who completed the informant rating version that included only the OLIW‐S and demographic information about them and about their relationship with the target (i.e., length and type of acquaintanceship). The baseline questionnaire was hosted during April and May 2021 online (all online questionnaires in this research were hosted by www.soscisurvey.de).

We collected data on the daily BRs of playfulness between May 31 and June 13, 2021. There were no COVID‐19‐related isolation measures in place during data collection. We sent the link to the daily questionnaire to targets at 6 p.m. on each of the 14 consecutive days. Participants did not receive financial compensation, but psychology students could earn course credit for participating.

2.3.2. Judges

We recruited the judges by leaflets on campus and advertised this part of the study as “research on judgments of personality from textual self‐descriptions.” We sent each judge a link to an online questionnaire that contained basic demographic questions and targets' self‐descriptions. For each target, judges provided their impressions on how playful the target is by completing the informant rating form of the OLIW‐S (Proyer, Brauer, and Wolf 2020). The targets' self‐description was displayed on separate pages with the OLIW‐S informant forms below the targets' descriptions. Judges were allowed to take breaks and complete the questionnaire within 1 month after receiving the link. Each judge received €30 for participating.

2.4. Data Analysis

We computed all the interpersonal perception analyses for (a) aggregated judgments across judges and (b) the average single judge. The aggregation of judgments increases the reliability of scores and thus overestimates the interpersonal perception coefficients. To address this issue and provide more realistic estimates of SOA, consensus, accuracy, and sensitivity, we additionally computed all analyses for each judge separately and computed the mean coefficients across the judges using the Fisher r‐to‐z transformation. The latter estimates the accuracy of interpersonal perceptions of the typical single judge. We computed bootstrapped 95% confidence intervals (CIs; 5000 bootstrap samples) for all coefficients.

2.4.1. Consensus

We computed inter‐judge agreement as intra‐class correlations (ICC; Shrout and Fleiss 1979), with ICC (2,1) providing the average agreement between two randomly chosen single judges, and ICC (2,6) the agreement among the six judges. Note that the latter can also be interpreted as a measure of reliability of the aggregated judgment scores.

2.4.2. Self‐Other Agreement

We computed the trait‐wise SOAs as bivariate correlations between targets' self‐reports and judge reports separately for each OLIW facet. In line with Furr (2008), we computed the profile SOA on the basis of two types of profiles: raw profile SOA was computed as a correlation between targets' and judges' OLIW item scores, resulting in a profile correlation coefficient for each target‐judge dyad. We then computed the average raw profile SOA by transforming the profile correlation coefficients to z values with the Fisher r‐to‐z transformation, averaging the z values, and computing the back transformation (z‐to‐r) of the mean value. We tested the statistical significance of the raw profile SOA by computing a one‐tailed t‐test (test value = 0). To control for stereotype effects, we also computed distinctive profiles that describe deviations from the average person. Therefore, we mean‐centered targets' scores and judges' scores on their respective sample means. We then correlated the distinctive profiles for each target‐judge dyad and computed the average distinctive profile SOA, along with a t‐test to test the statistical significance. In line with the literature, we report both the raw profile agreement and distinctive profile agreement coefficients, for transparency and to evaluate the magnitude of the difference between raw and distinctive profile agreement, which reflects the effect of controlling for stereotype effects (Furr 2008; Wood and Furr 2016).

2.4.3. Accuracy

We computed bivariate correlations between the judgments by the unacquainted judges and (a) the aggregate of ratings by targets and their knowledgeable others and (b) the playfulness BR.

2.4.4. Lens Model Analyses and Sensitivity

In line with Proyer and Brauer (2018), we used LIWC (Pennebaker and King 1999) to analyze the targets' textual self‐descriptions. Contrary to Proyer and Brauer, we used the updated version 18 along with the updated German dictionary by Meier et al. (2018). After computing the cue validity (i.e., the correlation between LIWC‐extracted word frequencies and targets' self‐reports of playfulness) and cue utilization (correlation between LIWC word frequencies and judges' reports of targets' playfulness), we computed vector correlations between the cue validity and cue utilization correlations (e.g., Borkenau and Liebler 1993). The vector correlations were computed after transforming the cue validity and cue utilization correlation coefficients with the Fisher r‐to‐z transformation. Higher sensitivity indicates greater overlap between the validity and utilization of linguistic cues.

2.4.5. Statistical Power

Post hoc power analyses (type = sensitivity) with G*Power (Faul et al. 2009) showed that our sample size allowed to detect typical effect sizes for interpersonal perception studies (rs ≥ 0.22) with a type‐I‐error rate of 5% and 80% power for the variable‐centered analyses. Note that the person‐centered analyses (profiles) are based on the average of 160 correlations (one for each target‐judge dyad) that were each computed on the basis of correlating targets' and judges' responses to the set of questionnaire items; thus, the average profile agreement correlation shows increased power and reduced standard errors (Borkenau and Leising 2016).

3. Results

3.1. Preliminary Analyses

The descriptive statistics for targets' and judges' OLIW scores were comparable to prior research analyzing self‐ and other reports (e.g., Brauer, Sendatzki, and Proyer 2024a; Proyer and Brauer 2018; see ESM C for all coefficients). Similarly, ratings by targets and their knowledgeable others showed the expected agreement, with correlations of 0.47 (Other‐directed), 0.42 (Lighthearted), 0.30 (Intellectual), and 0.46 (Whimsical; ps < 0.001), thus, supporting the aggregation of the ratings by targets and their knowledgeable others for the accuracy criterion. Furthermore, targets reported an average 10.1 (SD = 3.50) playful behaviors each day. Finally, the targets' textual self‐descriptions contained on average 63.2 words (SD = 27.6). The essays were longer than those in Proyer and Brauer's sample (M = 42.3 words, SD = 27.7).

3.2. Interpersonal Perception of Adult Playfulness

Table 1 provides an overview of all indices of interpersonal perception for the trait‐wise analyses. Note that we provide estimates for aggregated judgments of the six judges (labeled average) as well as the accuracies of the typical single rater (labeled single).

3.2.1. Consensus

The inter‐judge agreement was high for the aggregated judgments, with ICCs ≥ 0.71. Compared to Proyer and Brauer (2018), differences were ≤ 0.04 between studies. The inter‐judge agreement between two random single judges was between 0.29 (Whimsical) and 0.37 (Lighthearted), thus also comparable to Proyer and Brauer (maximum difference of 0.09).

3.2.2. Self‐Other Agreement

In line with Proyer and Brauer (2018), we found that the SOA coefficients exceeded 0.30 for all OLIW facets. As expected, the coefficients were lower for single judges, but nevertheless substantial, with rs ≥ 0.26. Extending Proyer and Brauer's analyses, we computed SOAs for the full profiles of the OLIW facets. For the aggregated judgments, the raw profile agreement was r = 0.32 (95% CI [0.26, 0.37]), and the distinctive profile agreement was r = 0.17 (95% CI [0.11, 0.24]; ps < 0.001). For the average single rater, we found the expected lower coefficients, which still exceeded zero (raw: r = 0.15, 95% CI [0.09, 0.20], p = 0.008; distinctive: r = 0.08, 95% CI [0.04, 0.11]; p = 0.011). In conclusion, SOA extended from single facets to full profiles of adult playfulness at zero acquaintance.

3.2.3. Sensitivity

The overlap between cue validity and cue utilization correlations exceeded zero for all facets (ps < 0.001). While the coefficients for Lighthearted (r = 0.66, 95% CI [0.54, 0.75]) and Whimsical (r = 0.48, 95% CI [0.31, 0.63]) replicated well when using the updated LIWC dictionary, sensitivity was numerically slightly lower than reported in Proyer and Brauer (2018) for Other‐directed playfulness (r = 0.33, 95% CI [0.11, 0.55]), whereas the sensitivity was higher than previously reported for Intellectual playfulness (r = 0.41, 95% CI [0.23, 0.55]). All cue validity and cue utilization correlations are provided in ESM D.

3.2.4. Accuracy (Self‐Peer Aggregate)

When testing the correlations between judges' ratings and self‐peer aggregates of playfulness as an accuracy criterion (cf. Funder and West 1993), we found comparable findings to those based on targets' self‐reports only. Hence, supplementing targets' self‐ratings with those by knowledgeable others did not robustly affect SOA, as pairwise comparisons between the SOA correlations with and without supplements by knowledgeable others demonstrated (all ps ≥ 0.211). Similarly, the profile agreement coefficients for aggregated judgments (r _raw = 0.32, 95% CI [0.26, 0.37], r _distinctive = 0.18, 95% CI [0.11, 0.24]; ps < 0.001) and single judgments (r _raw = 0.15, 95% CI [0.08, 0.21], p = 0.013; and r _distinctive = 0.08, 95% CI [0.04, 0.11], p = 0.010) were unaffected by the inclusion of ratings by knowledgeable others.

3.2.5. Accuracy (Diary Data)

We found considerable overlap between the judgments at zero‐acquaintance and targets' aggregated diary data from the playfulness BR (Proyer 2017). Correlations between judge ratings and targets' playful behaviors were ≥ 0.30 (aggregated judgments) and ≥ 0.21 (single judgments) for Other‐directed, Lighthearted, and Intellectual playfulness. Concerning Whimsical playfulness, we found numerically lower associations between external judgments and targets' diary data, with r = 0.16 (aggregated) and 0.10 (single; ps < 0.069).

4. Discussion

The present study aimed to contribute to knowledge of the interpersonal perception of adult playfulness by replicating and extending prior research and by employing diverse methodological approaches (Proyer and Brauer 2018). How accurately do judges infer fine‐grained facets of playfulness from targets' short textual self‐descriptions? In short, our findings provide strong evidence for accuracy and consensus in perceiving adult playfulness at zero‐acquaintance. Also, the sensitivity, measured by the overlap between cue validity and cue utilization, exceeded zero for all facets—and approaches. This means that even with very limited information provided, people are able to accurately infer targets' playfulness at the facet level. This suggests an intriguing interplay between how we communicate playfulness through writing and how others perceive it, and this may have real‐world implications for various aspects of our daily lives.

Our study is a conceptual replication of earlier research (Proyer and Brauer 2018); a major variation was that we used a sample of six judges who provided perceptions of all targets as opposed to a subset of targets. To address increased efforts in providing judgments of the 160 targets, judges completed the 12‐item brief version of the OLIW questionnaire (Proyer, Brauer, and Wolf 2020). Also, we extended the original study in various ways: We tested the SOA for profiles in addition to single facets, which allowed us to examine whether SOA extends to the full profile of OLIW facets and to control for stereotype effects that otherwise overestimate profile agreement (Furr 2008; Wood and Furr 2016). We extended the research by using two accuracy criteria that allow conclusions about whether external judgments of playfulness relate to targets' playfulness beyond what self‐reports provide (Funder and West 1993). First, we supplemented targets' self‐reports with those of their knowledgeable others and aggregated them to derive a more precise estimate of targets' playfulness by reducing blind spots (e.g., favorable self‐ and other views; Hofstee 1994). Second, we collected daily ratings with a BR of playfulness across 14 consecutive days (Proyer 2017) and examined correlations with the judgments that were inferred from targets' self‐descriptions. To our knowledge, this was the first study that addressed all of Funder and West's (1993) proposed criteria of interpersonal perception for the study of playfulness, and, overall, one of only a few studies that has examined an accuracy criterion besides self‐peer aggregates (Connelly and Ones 2010).

The findings on the SOA and consensus criteria and sensitivity well‐replicated those found in Proyer and Brauer (2018). Contrary to the initial study, we found a robust increase of about 0.15 in SOA for Intellectual playfulness. Considering that the reliability of the judgments in terms of consensus was comparable to Proyer and Brauer, it could be argued that this finding relates to the fact that the present target sample provided longer self‐descriptions (despite working with the same instruction). While there is no evidence that Intellectual playfulness relates to writing longer texts according to the present data and earlier research (Brauer et al. 2022; Proyer and Brauer 2018), it is possible that the longer descriptions were more saturated with information that allowed our judges to derive more information regarding targets' intellectual expressions of playfulness. Funder's (1995) Realistic Accuracy Model (RAM) assumes that the availability, detectability, and utilization of trait‐relevant information is the basis of accurate judgments. In line with the RAM, our interpretation is that the relatively greater availability of trait‐relevant information provided by targets could account for the higher SOA. Alternatively, one might argue that our sample of judges was “better” in utilizing this information, but we suggest that this is comparatively less likely as this would arguably also reflect in increases in between‐judge agreement, and, thus, higher consensus coefficients. However, these came out almost identical to Proyer and Brauer's.

At the same time, the sensitivity coefficient that describes the overlap between the cue validity and cue utilization correlations was markedly higher in this study, indicating that the judges correctly utilized and rejected linguistic cues. Nevertheless, the lack of comparability between the LIWC dictionaries across studies and recent research questioning the robustness of LIWC correlations across different samples and versions warrant caution in interpreting these results (Koutsoumpis et al. 2022; Martínez‐Huertas et al. 2022). Additional research on the target × information × judge interaction is needed to elucidate the factors moderating the accuracy of perceptions of playfulness.

On average, we found slightly reduced numerical coefficients in our study, which aligned with our expectations. The deviation can be attributed to differences in judgment reliability compared to the initial study. Two factors likely contributed to lower reliability in this study: First, our judge sample was smaller, with 6 instead of 10 judges, and thus we collected fewer observations (see Walker 2008, for a discussion); second, our judges completed the brief 12‐item version of the OLIW questionnaire instead of its full 28‐item version, to reduce completion time. Hence, our findings support the notion of accurate perceptions in the sense of agreement between unacquainted judges (i.e., consensus) about targets' playfulness and between targets' self‐reports and judgments at zero‐acquaintance. The latter finding remained robust when supplementing targets' self‐reports with those by their knowledgeable others. This approach reduces the potential influence of targets' biases in self‐views (Hofstee 1994) and provides a more accurate estimate of targets' playfulness. The fact that our findings on SOA did not depend on whether ratings by knowledgeable others were included can increase trust in the validity of self‐perceptions as well as those by the judges. Lending further support to the notion that playfulness can be comparatively accurately judged, independently of typical moderators of SOA such as length, type, and intensity of acquaintanceship (Brauer, Sendatzki, and Proyer 2024a).

Extending Proyer and Brauer's (2018) work, we examined SOA for the full profiles of the four facets. As expected, compared to well‐acquainted dyads (Brauer, Sendatzki, and Proyer 2024a), we found lower SOA for the profiles of the OLIW facets. The coefficients reflected above‐chance agreement, even after controlling for stereotype effects (i.e., distinctive agreement; Furr 2008). Hence, our judges were able to identify targets' deviations from the average target person when it came to the OLIW facets, when basing their impressions on targets' short textual self‐descriptions. Although the findings indicate small but above‐chance profile agreement for distinctive profiles, they are lower than those reported for other narrow traits such as dispositions that describe individual differences in dealing with laughter (Brauer and Proyer 2020). Since facet‐wise SOA reflects judges' ability to discriminate between targets on a certain trait, and profile agreement reflects how well judges discriminate between relative levels of a set of facets within a target (Hall et al. 2018), it could be argued that perceptions of playfulness at zero‐acquaintance are comparatively accurate when it comes to single facets but less so when it comes to the full set of the OLIW facets. Our findings indicate that judges appear to be more accurate in estimating levels of playfulness between targets regarding the same facet compared to distinguishing between facets within the same target. In accordance with the RAM (Funder 1995), there is increasing evidence that supports the notion that judges differ in how well they utilize trait‐relevant information and that individual difference variables such as judgment styles (e.g., positivity bias and dissimilation) contribute to the understanding of moderators of agreement (e.g., Carlson et al. 2024). Hence, after replicating basic findings on the interpersonal perception of playfulness at zero‐acquaintance, additional research on the roles of quantity and quality of information and judges is desirable.

Finally, we examined the diary data of playful behaviors across 14 days with the playfulness BR (Proyer 2017). We used the behavior‐like data as a second accuracy criterion to investigate the validity of judgments at zero‐acquaintance beyond questionnaire data. To our knowledge, this was the second study using the playfulness BR and one of the few that have examined accuracy beyond self‐other aggregates (Connelly and Ones 2010). In a first step, we replicated correlations between the BR and targets' self‐reports of playfulness (cf. Proyer 2017) and provided additional evidence for the reliability and validity of the BR in terms of overlap with other ratings of playfulness. We found that the judgments made at zero‐acquaintance related robustly to the aggregates of the longitudinally collected behavior‐like data. This further supports the notion that there is validity in judgments of facets of playfulness at zero‐acquaintance. When comparing the accuracies between facets, judgments of Whimsical playfulness showed the numerically lowest, yet positive, overlap with the behavioral data. Our preliminary analyses showed that while self‐reports of Whimsical playfulness related robustly with the playfulness BR, the reports of knowledgeable others were unrelated. This might hint at a divergence in self‐ and other views, which should be examined in future research regarding the playfulness BR. Future research might also use the approach of plural data sources to examine potential moderators of accuracy. For example, consistencies between self‐report questionnaires and longitudinal diary data might contribute to accurate judgments because targets' self‐reports might be more precise estimates of their typical behavior over time in comparison to targets who show greater variability across time and situations (see also Borkenau et al. 2004; Wiedenroth et al. 2024).

Taken together, our findings were in line with expectations and increase trust in the conclusion that playfulness can be accurately perceived at different degrees of acquaintanceship, including when using short textual self‐descriptions as “thin slices” of behavior at zero‐acquaintance. On a broader level, our findings add to the body of literature showing that self‐descriptions offer information about both broad and narrow personality traits (e.g., Borkenau et al. 2016; Körner et al. 2024) and highlight the importance of written communication in interpersonal processes as a source of information about others. The role of self‐descriptions as a reflection of personality traits has been studied frequently across contexts such as standardized lab conditions (e.g., Borkenau et al. 2016; Körner et al. 2024), regarding personnel selection and application success (e.g., Brandt and Herzberg 2020), and impressions of online dating profiles (e.g., Tong et al. 2020). At the same time, it is questionable whether the type and extent of information is a moderator of accuracy, when considering Allik's (2018) summary of the literature concerning zero‐acquaintance studies as “The general impression seems to be that it does not matter into how many thin slices the experience is divided. Personality information is still present and can be used for accurate judgments” (p. 115). This sentiment also fits with the notion that acquaintanceship plays only a minor for accurate judgments (Allik 2018) because most studies, including those of playfulness (Brauer, Sendatzki, and Proyer 2024a), imply that after a certain threshold of acquaintanceship is surpassed, observing additional acts of behaviors does not contribute robustly to judgment accuracy. While research on the interpersonal perception of playfulness is currently limited to standardized settings, we recommend extending the study to real‐life scenarios. For example, we expect that accurate impressions of playfulness relate to various indicators and processes of social relationships such as work relationships, friendships, intra‐familial relations, and couples. This line of research also extends our approach from describing agreement to testing consequences of agreement.

The role of playfulness in romance has gained strong interest (e.g., Proyer et al. 2019; see Brauer, Proyer, and Chick 2021, for a discussion and overview), and one fruitful future research direction might involve the question of whether assortative mating, which requires deriving accurate inferences about potential partners' playfulness (e.g., initial impressions of online dating profiles), contributes to relationship formation and whether co‐development in actual and perceived playfulness expressions among partners relates to indicators such as relationship satisfaction and stability. Also, other sources of information at the juncture of zero‐acquaintance might be examined, and we might study whether findings extend from self‐descriptions to other media such as physical appearance cues contained in photos like “selfies” and portraits (e.g., Borkenau and Liebler 1993; Osterholz, Mosel, and Egloff 2023).

4.1. Limitations and Future Research

The available findings on the perception of playfulness from language are limited to German‐speaking targets, judges, and texts. Considering that cross‐language differences might play a role (Barnett 2017), replication in non‐German‐speaking countries is desirable to examine the generalizability of our findings. Although the playfulness BR allowed us to use a measure that provides longitudinal data on items that were designed to assess behavioral acts (see Wu and Clark 2003), responses are still based on self‐reports, and although the longitudinal collection of data and subsequent aggregation of data should reduce momentary biases, using a measure of playfulness that is completely independent from self‐perceptions is desirable. We addressed this concern by also investigating ratings by targets' knowledgeable others and using a multimethodological approach to reduce common method biases (Campbell and Fiske 1959).

Our research design followed the classical approach to study interpersonal perception of personality traits, using data from a small sample of judges who provide ratings about a comparatively large sample of targets. Hence, we cannot extract target‐level estimates controlling for perceiver and dyadic variance and did not account for the individual judges' rating styles. Future research should consider two alternative designs. First, using a round‐robin design, in which participants act as targets and perceivers simultaneously, would allow to examine target‐, perceiver‐, and relationship effects that contribute to accurate judgments of playfulness. Analyses with the Social Accuracy Model (Biesanz 2021) or Social Relations Model (Kenny 2020) would allow to learn more about the unique effects that explain accuracy, with regard to the role of targets, judges, and unique dyadic effects. Secondly, it is important to increase the sample size of judges to increase the generalizability and to derive reliable accuracy coefficients. Further, a design focusing on a large sample of judges (i.e., k ≥ 150) who rate a smaller set of targets (e.g., n ≤ 50) would allow studying the role of judges' individual differences as predictor of agreement (Nestler and Back 2017). Considering that initial evidence shows that judges' personality traits and cognitive abilities (e.g., reading comprehension) are a source of variation in the accuracy of judgments when making inferences about personality from text‐based sources (Hall et al. 2016), it would be worthwhile to examine whether playful judges can assess playfulness more accurately than less playful judges.

In conclusion, the present study supports the robustness of findings on the interpersonal perception of fine‐grained facets of playfulness at zero‐acquaintance found in Proyer and Brauer (2018) and contributes to the knowledge of the validity of social judgments of playfulness. We hope that our study helps to advance and stimulate research on the consequences and correlates of accurate judgments of playfulness in applied contexts and life domains such as social relationships using more naturalistic study designs (e.g., speed‐dating studies and studying perceptions at the workplace) to learn more about the short‐ and long‐term consequences of accurately perceiving single facets and profiles of playfulness.

Author Contributions

Kay Brauer: data collection, study design, data analysis, data interpretation, and writing – original draft. René T. Proyer: study design, data interpretation, and writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1.

Data Availability Statement

All data and materials are openly available in the Open Science Framework under https://osf.io/v4ytk/