Understanding what Predicts Meta-Accuracy: Exploring the Good Metaperceiver,the Good Perceiver,and the Good Trait

Abstract

Knowing how you come across to new acquaintances carries important social implications, especially in first impression contexts. But what factors relate to this capacity? The present research aimed to map out the characteristics of the “good metaperceiver”, “good perceiver” and “good trait”. We explored these links across two large studies involving getting-acquainted interactions in person (N = 860, dyads = 4579; M_age = 20.37, 82.91 % women) and on Zoom (N = 911, dyads = 4897; M_age = 20.45, 82.74% women). Adapting the social accuracy model for metaperceptions, we found consistent results revealing that the good metaperceivers generally viewed themselves more positively (i.e., reported desirable personality traits and psychological adjustment) and were generally viewed more positively by interaction partners (i.e., rated as more likeable). Further, the good perceiver reported being more interpersonally warm (e.g., agreeable). Finally, the good trait was observable and non-evaluative (socially neutral). Together, these findings offer a more nuanced understanding of when and for whom meta-accuracy emerges. By identifying the individual and trait-level correlates that may facilitate accurate metaperception, this work sheds light on how people come to understand the impressions they make and lays the groundwork for improving social functioning in everyday interactions.

Whether navigating a first date, making a new friend, or introducing oneself in a professional setting, people often come to wonder about how others view them. These impressions we have about how others view us are referred to as metaperceptions, and they can be more or less accurate. As such, meta-accuracy captures the extent to which people’s impressions about how they are seen align with others’ actual impressions of them. For example, Mira’s metaperception might be that Paulo, her colleague, sees her as being very reserved and somewhat nervous during their first encounter. If Paulo does, in fact, see her that way, Mira is meta-accurate: her metaperception corresponds to his impression of her. But if Paulo instead sees her as highly talkative and fairly confident, she has misjudged his impression, revealing a misalignment between her metaperception and his actual impression.

Given the example above, then, meta-accuracy is likely a crucial social tool because it can inform how we manage our relationships and interactions with others. When we accurately understand how we are perceived, we can tailor our behaviors to strengthen connections, resolve misunderstandings, and prevent unnecessary conflict. But, if we have missed the mark, we might unknowingly reinforce negative impressions, misjudge social cues, or behave in ways that undermine our relationships. If Mira believes Paulo saw her as reserved and tries to appear more talkative, but Paulo already sees her as quite talkative, her behavior might strike him as over-the-top or overwhelming, potentially undermining the impression she’s trying to manage. As such, meta-accuracy, reflecting metaperceptions grounded in reality, could be helpful for social coordination and relationship development (Carlson, 2016b; Tissera et al., 2021; Tissera et al., 2023).

But what predicts how well we can tell how we come across to others? Although personality meta-accuracy has been examined in prior work, there is less clarity about which predictors consistently predict meta-accuracy. In fact, prior research has used different analytic approaches and examined different predictors in isolation or within narrow conceptual frameworks (e.g., Carlson, 2016a; Carlson & Kenny, 2012; Elsaadawy & Carlson, 2022; Hater et al., 2023; Tissera et al., 2021; Tissera et al., 2023), which makes it challenging to discern systematic patterns or draw broader theoretical conclusions. In response to this gap, and to directly build on the field’s existing findings, we adapt the moderators of accurate judgment proposed by Funder (1995) to the domain of meta-accuracy. In doing so, we take a broader and more systematic approach by exploring the role of three main categories of moderators associated with personality meta-accuracy. First, we examined the role of the “good metaperceiver”, exploring whether some people are better at knowing how they come across to others. Second, we explored the role of the “good perceiver”, investigating whether some interaction partners elicit more accurate metaperceptions. Finally, we examined the role of the “good trait” to test whether some traits are more easily metaperceived than others.

To explore the characteristics of the good perceiver and metaperceiver, we examined a large common set of individual difference indicators across both roles ¹ . These included personality traits (the big five, narcissism), intrapersonal (self-esteem, satisfaction with life) and interpersonal adjustment (loneliness, positive relations with others), and interpersonal appeal (attractiveness, likeability). This parallel approach allowed us to assess whether the same characteristics facilitate meta-accuracy from both the metaperceiver and perceiver perspectives. Of note, while some characteristics have prior empirical or theoretical support for at least metaperceivers (e.g., Carlson, 2016a; Carlson & Kenny, 2012; Elsaadawy & Carlson, 2022; Hater et al., 2023; Tissera et al., 2021, 2023), we opted to take an inclusive, exploratory approach, examining a large set of individual difference variables that were available in both datasets to further extend prior research. Finally, to test the good trait, we explore two of the most commonly examined characteristics of traits (e.g., Funder & Dobroth, 1987; John & Robins, 1993; Leising et al., 2010; Mignault et al., 2023; Rau et al., 2021; Vazire, 2010; Wiedenroth et al., 2024): (1) observability, how easily a trait can be observed by others, and (2) evaluativeness, how socially desirable or undesirable a trait is. In doing so, this research is the first to examine multiple categories of moderators together and multiple characteristics within each category, providing a comprehensive examination of the correlates of meta-accuracy. Figure 1 below provides a conceptual framework that visually summarizes these moderators and their possible influence on the stages of the meta-accuracy process (described below).

Figure 1.

Overview of the Conceptual Framework.

In the present research, we focus on the context of first impressions, which are arguably the moments when we ponder the most about the impressions we leave on others. Moreover, forming accurate metaperceptions in first impressions has been linked to positive social outcomes. For example, personality meta-accuracy among new acquaintances is predictive of being more liked by that acquaintance both concurrently (Tissera et al., 2021; Tissera et al., 2023) and in the long term (Carlson, 2016b), potentially setting the trajectory for relationship development. Thus, getting-acquainted interactions is a context where meta-accuracy about others’ personality impressions is especially important to achieve, making it crucial to understand how to do so.

Indexing Meta-Accuracy

People are meta-accurate when their metaperceptions correspond to others’ impressions of them. For example, Mira is meta-accurate if her colleague, Paulo, actually sees her as being very sociable and somewhat creative. In the context of metaperception, the question of “truth”, often debated in other accuracy research (Funder, 1995; Vazire, 2010), has a more concrete benchmark: the perceiver’s actual rating. That is, what the perceiver says they think of the metaperceiver becomes the criterion for accuracy. This makes metaperception research uniquely well-positioned to assess accuracy, since the target and the criterion both exist within the same interpersonal exchange. Thus, to index Mira’s meta-accuracy, we can compare Mira’s metaperceptions against her colleague’s actual impressions of her. The degree of correspondence between metaperceptions and perceivers’ impressions provides an operational definition of meta-accuracy. Nevertheless, it is important to keep in mind that the perceivers’ and metaperceivers’ impressions may not align with the objective reality of what a metaperceiver is like. For example, a more positive interpersonal exchange may also foster shared positive illusions, such that metaperceptions and perceivers converge on a similarly positive impression profile, but the metaperceiver may not actually possess such a positive profile. However, meta-accuracy is not assessing the objective reality of the metaperceiver’s personality profile; it is assessing whether the metaperceivers’ impressions of how they are seen converge with the perceivers’ impressions of the metaperceiver.

Prior work has distinguished between generalized and dyadic meta-accuracy (Hater et al., 2023; Kenny, 1994; Carlson & Kenny, 2012). Generalized meta-accuracy refers to the extent to which individuals understand how they are typically seen by others across interaction partners, whereas dyadic meta-accuracy reflects the extent to which individuals understand how they are seen by a specific interaction partner. In the present research, we focus on dyadic meta-accuracy, examining how well metaperceivers track the unique impressions formed by specific perceivers within each interaction.

To quantify this correspondence, researchers have adopted several approaches that vary in both focus and granularity. One distinction is between absolute and relative meta-accuracy. Absolute accuracy reflects how close a person’s metaperception is to the perceiver’s impression in raw terms. For instance, if Mira thinks Paulo sees her as a 6 on sociability and he rates her a 6, that is high absolute accuracy. However, due to the challenges in interpreting absolute scores and difference-based indices (e.g., Cronbach, 1955; Furr, 2011), we instead adopt a relative approach to assessing meta-accuracy.

The relative approach looks at whether a person understands the pattern of how they are seen across several traits, rather than focusing on exact numbers. Within this approach, there are two common ways to measure accuracy. Trait-wise accuracy looks at each trait on its own. As such, it is particularly useful when researchers aim to study how accurately people understand their reputations in relation to others on specific personality dimensions (e.g., Carlson & Oltmanns, 2015; Eisenkraft et al., 2017; Levesque, 1997; Mastroianni et al., 2021; Oltmanns et al., 2005). For example, is Mira, the metaperceiver, able to accurately understand whether Paulo sees her as more or less sociable than their other colleagues?

In contrast, the profile-based approach, which we use in the present research, looks at the overall pattern across traits. For instance, does Mira accurately perceive how Paulo views her across multiple dimensions, such as sociability, creativity, and intelligence? A higher profile correlation means that Mira accurately understands how Paulo ranks her traits relative to one another (e.g., she believes he sees her as more sociable than imaginative, and he in fact does). A lower correlation suggests her metaperception misrepresents the relative highs and lows in Paulo’s impression of her. Unlike the trait-wise analyses, this holistic perspective enables us to capture the interplay between the “good metaperceiver,” “good perceiver,” and “good trait” moderators within a single, integrated framework. Moreover, the profile approach can be implemented in a way that makes it equivalent to the average trait-wise accuracy, across all the items that are included within the profile (Biesanz, 2021; see more details in the analytical approach).

The profile approach also offers several statistical advantages. By considering the entire personality profile collectively rather than conducting separate analyses for each trait, the total number of analyses is reduced, thereby decreasing the likelihood of Type I errors. By treating each item as the unit of analysis, it consolidates all data into a single model to maximize statistical power and minimize Type II errors. This approach also allows for the exploration of how specific item characteristics (e.g., observability) may relate to meta-accuracy, which is of interest to the present research.

Achieving Meta-Accuracy

Having defined how meta-accuracy is conceptualized and measured, we now ask: Do people actually achieve meta-accuracy in first impressions? Past work suggests that people can achieve meta-accuracy (Carlson & Furr, 2009; Carlson & Kenny, 2012; Kenny & DePaulo, 1993), even based on brief initial interactions (e.g., Carlson, 2016b; Carlson et al., 2010; Elsaadawy et al., 2021; Stopfer et al., 2014; Tissera et al., 2021; Tissera et al., 2023; Wu & Zheng, 2019). However, even though meta-accuracy tends to be significantly greater than zero, the average correlation between metaperceivers’ average metaperceptions and the average impressions formed by others tends to be relatively low for first impressions of personality (r: .17 – .22; Carlson & Kenny, 2012), suggesting that people may generally have limited insight into how their personalities are initially perceived. Further, levels of personality meta-accuracy in first impressions appears to vary significantly (Elsaadawy et al., 2021), suggesting the presence of meaningful predictors.

The Four-step Process of Achieving Meta-Accuracy

One way to uncover these predictors is to consider how meta-accuracy is achieved in the first place. Funder’s Realistic Accuracy Model (RAM; Funder, 1995) offers a useful framework to understanding how people might achieve meta-accuracy. He outlines four stages in forming accurate judgments (see Figure 1). According to the RAM, accuracy depends on (1) cue relevance (2) cue availability, (3) cue detection, and (4) cue utilization. Although the RAM was initially developed to explain how people form accurate judgments of individual traits, the model is conceptually compatible with profile-based meta-accuracy. As Letzring and Funder (2021) note, real-world judgments often involve integrating information across multiple traits and cues simultaneously. The RAM explicitly accounts for this complexity, acknowledging that people make judgments of many traits at once and that the four stages (relevance, availability, detection, utilization) must be successfully navigated for each of these judgments. For example, even a highly attentive person cannot achieve accuracy if no relevant cues were present to begin with. Thus, the model provides a flexible framework that can be extended to profile-level judgments, particularly when considering how patterns of behavior are observed and interpreted holistically.

Now, we can apply the RAM to meta-accuracy (Elsaadawy & Carlson, 2021). For Mira to realize that Paulo sees her as sociable, she must navigate four stages: (1) cue relevance, where relevant information about how she is perceived must exist; (2) cue availability, where she must have access to this information; (3) cue detection, where she must notice or be aware of the relevant information; and (4) cue utilization, where meta-accuracy depends on interpreting the information correctly (see Figure 1).

Sources of Meta-Accuracy

The first step of the RAM pathway depends on having relevant informational sources (See Figure 1). These sources can be grouped into at least two categories: pre-existing and interaction-specific. We discuss each of these below.

Pre-Existing Beliefs

One primary source of accurate metaperceptions is self-perceptions, as people often rely on their own views to gauge how others see them, with substantial overlap between self-views and others’ perceptions (John & Robins, 1993; Kenny & DePaulo, 1993). For instance, Mira might see herself as generally sociable and creative, and this could align with how others, like Paulo, perceive her. However, people also display meta-insight (Carlson et al., 2011), understanding how their self-views deviate from others’ perceptions, which suggests that other sources of information are also important.

Indeed, other pre-existing sources include normative beliefs (i.e., beliefs about how people generally tend to be) and positive beliefs (i.e., beliefs about what is desirable). Of note, normative beliefs also tend to be highly positive (Edwards, 1957; Rogers & Biesanz, 2015; Wood & Furr, 2016), but it is possible to also examine them separately (e.g., Wessels et al., 2020). Given that perceivers typically tend to view others in normative and positive ways (Biesanz et al., 2007; Borkenau & Leising, 2016; Rogers & Biesanz, 2015; Wood & Furr, 2016), reliance on this information can help people predict how others may perceive them (Elsaadawy et al., 2021; Hater et al., 2023; Tissera et al., 2021; Tissera et al., 2023). ²

Interaction-specific

Other sources of meta-accuracy are those cues available during the actual interaction with a perceiver. First, people might use their own (i.e., metaperceiver’s) behaviors to infer how they are seen. The way they are behaving (e.g., not talking much) or even self-disclosures during the interaction (e.g., explicitly stating “I’m usually shy”) may help metaperceivers to infer others’ impressions of them, as perceivers’ impressions are likely to at least in part correspond to how a metaperceiver behaved during the interaction.

Second, the perceivers’ behavior could also be informative. During the interaction, people might be able to infer stable individual differences in how perceivers tend to view others across interactions or contexts, referred to as perceiver effects. These idiosyncratic judgment styles can shape impressions and are a central component in person perception research (Dufner et al., 2016; Hehman et al., 2017; Heynicke et al., 2022; Kenny, 1994; Rau et al., 2021, 2022; Srivastava et al., 2010; Wood et al., 2010). As such, in the context of meta-accuracy, perceiver effects may be particularly informative, as understanding these idiosyncratic tendencies could help metaperceivers figure out how they are viewed (e.g., Hater et al., 2023; Tissera et al., 2021, Tissera et al., 2023).

Another form of perceiver behavior is feedback, both verbal (e.g., compliments or remarks) and nonverbal (e.g., a surprised look). While potentially useful, feedback can be an unreliable source (Elsaadawy & Carlson, 2021) due to perceivers’ hesitance to provide honest input, especially about less favorable traits (Hebert & Vorauer, 2003), or metaperceivers’ misinterpretations of feedback due to complexities of social interactions or egocentric biases (Gilbert & Osborne, 1989; Gilovich et al., 2000; Lieberman & Rosenthal, 2001; Shectman & Kenny, 1994; Swann et al., 1992).

In sum, meta-accuracy begins with the presence of relevant information, and this information can stem from both pre-existing beliefs and interaction-specific behaviors. By shaping what information is available to be judged, these sources lay the groundwork for the entire RAM pathway to meta-accuracy.

Moderators of Meta-Accuracy

Even when relevant information exists, people may differ in how effectively they navigate the RAM. To be able to better explain this variance, Funder (1995) proposed four different moderators of accurate judgments: (1) good target (i.e., some people are more easily judgeable than others), (2) good judge (i.e., some people are more accurate judges than others), (3) good trait (i.e., some traits are easier to judge accurately than others), and (4) good information (i.e., some contexts offer better quantity and quality information). All of these moderators have been shown to predict the accuracy of personality judgments (Biesanz et al., 2007; John & Robins, 1993; Langlois et al., 2000; Letzring et al., 2006; Letzring, 2008; Ruben & Hall, 2016; also see Letzring & Funder, 2021 for a recent discussion).

Given that meta-accuracy is also a certain type of accurate judgment, each of these moderators also likely applies here. Thus, they can be reframed as follows (1) good metaperceiver (i.e., some people might be more accurate about how they are seen than others), (2) good perceiver (i.e., some people’s impressions might be easier to accurately perceive than others), (3) good trait (i.e., some traits might be easier to know how you are seen on), and, (4) good information (i.e., some contexts might offer more and better quality information). ³ Below we discuss each moderator in more detail, but in the present research we focus on the first three.

The Good Metaperceiver

In the present research, the idea of the good metaperceiver is operationalized as individual differences in the accuracy of metaperceptions. That is, some people might have characteristics that help them more accurately understand the unique ways in which they are perceived by others. Notably, being a “good” metaperceiver in this sense refers to accuracy in metaperceptions and does not necessarily imply that such insight is necessarily prosocial, authentic, or well intentioned.

Past research suggests that meta-accuracy does significantly vary by metaperceiver (Elsaadawy et al., 2021; Hater et al., 2023; Tissera et al., 2021; Tissera et al., 2023), even though these studies have indexed meta-accuracy in slightly different ways. Together, this work indicates that there may be some people who are better at this ability than others. Accordingly, in the present work, we conceptualize good metaperceiver characteristics as individual differences that may serve as antecedents of meta-accuracy.

Our focus on personality traits, intra and interpersonal adjustment, and interpersonal appeal (see Figure 1) was guided by prior theory and empirical precedent (e.g., Carlson, 2016a; Elsaadawy & Carlson, 2022; Mosch & Borkenau, 2016; Tissera et al., 2021; Tissera et al., 2023). At the same time, this framework is exploratory and not intended to be exhaustive. These categories provide a theoretically grounded starting point for identifying predictors of meta-accuracy, but future work may benefit from expanding beyond these domains to consider additional cognitive, motivational, demographic or contextual factors that shape how well people understand the impressions they make.

Personality

We examined whether metaperceiver personality traits would predict meta-accuracy. Although meta-accuracy has not been operationalized in exactly the same way across prior studies, existing work suggests that individual differences in personality may nonetheless relate to how accurately people understand how they are perceived by others. For example, extraverted metaperceivers might be more meta-accurate (Hater et al., 2023), because extraverts tend to express themselves clearly and engage in more social interaction, features that may increase cue relevance and availability. Although there is less empirical evidence that the remaining big five traits would predict meta-accuracy (Hater et al., 2023), it is theoretically plausible that having a desirable personality (e.g., high in agreeableness, conscientiousness, emotional stability and openness) might facilitate meta-accuracy. For example, agreeableness may promote sensitivity to others’ perspectives, and openness may foster curiosity about others’ views and integration of diverse perspectives, supporting more accurate metaperceptions. Additionally, past work has found some evidence that narcissistic people have some awareness of their reputations among new acquaintances and close others (Carlson et al., 2011). All to say, there is reason to believe that the metaperceiver’s personality could relate to their meta-accuracy and we, therefore, examined a range of potentially relevant characteristics.

Intrapersonal Adjustment

Psychological adjustment has been conceptualized as an antecedent of some measures of meta-accuracy (Carlson, 2016a; Mosch & Borkenau, 2016). Specifically, higher intrapersonal adjustment could benefit different stages of the RAM. For example, people higher in self-esteem may be at ease with themselves and others, enabling clear self-expression (cue relevance and availability), facilitating observation of behaviors (cue detection), and interpretation of behaviors as they might be less defensive (cue interpretation). However, past research has found mixed evidence to support this possibility. Carlson (2016a) found that intrapersonal adjustment (e.g., self-esteem, life satisfaction) was related to meta-accuracy, but this link disappeared when controlling for positivity. Mosch and Borkenau (2016) found some evidence that more adjusted people were actually less insightful about others’ unique impressions of them that were different from how they saw themselves. These inconsistencies likely stem from the use of diverse methods to assess meta-accuracy across studies, making it difficult to draw definitive conclusions. Nevertheless, the theoretical plausibility, combined with mixed past research findings, suggests that further research is required to better understand the links between intrapersonal adjustment and meta-accuracy.

Interpersonal Adjustment and Appeal

Instead, there is more empirical evidence that the good metaperceiver might be interpersonally adjusted (e.g., greater relationship well-being) and may have interpersonal appeal (e.g., greater likeability). For example, the interaction-specific sources of meta-accuracy involve paying attention to the interaction and/or the interaction partner. Given the inherent intricacies involved in navigating social interactions, detecting and utilizing relevant cues during an interaction might require high social skills. People with interpersonal skills may also be able to elicit more feedback from perceivers, enhancing cue availability. This also aligns with past work which finds that meta-accuracy is related to greater social (or interpersonal) adjustment (Carlson, 2016a). Although interpersonal appeal (e.g., likeability) is often conceptualized as an outcome of meta-accuracy, many of these studies were cross-sectional and thus it is plausible that likeability also facilitates meta-accuracy. Together these findings converge to suggest that the good metaperceiver might be interpersonally adjusted and might have high interpersonal appeal.

The Good Perceiver

At the outset, it is useful to clarify that the term “good perceiver” has traditionally been used in the accuracy literature to refer to a person who is particularly skilled at judging others accurately (along with the term “good judge”; e.g., Back & Nestler, 2016; Biesanz, 2010; Funder, 1995; Ickes et al., 2000). However, in the present context, the “good perceiver” refers to individuals who promote greater meta-accuracy from others, meaning that the perceiver’s impression of the metaperceiver is more easily readable because of some characteristic(s) of the perceiver. In other words, the good perceiver is also, in a sense, a target of metaperceptions. Despite this potential for terminological ambiguity, we use this term to maintain consistency with prior work (e.g., Elsaadawy et al., 2021; Hater et al., 2023; Porter et al., 2019; Tissera et al., 2023), and it ensures that the perceiver remains the same person across different approaches to accuracy measurement.

Past research suggests that there is a modest degree of perceiver variance around meta-accuracy (Elsaadawy et al., 2021), albeit with a small effect size. The variance distribution was narrower than for the good metaperceiver, indicating that identifying the characteristics of a good perceiver may be challenging. This is also consistent with the accuracy literature, which finds less variance and narrower distributions around accurate perceivers of personality as compared to targets (Biesanz, 2021; Human & Biesanz, 2013). Perhaps for this reason, there has been little work thus far examining the characteristics of a good perceiver, especially in the context of metaperceptions. Nonetheless, from a theoretical standpoint, certain characteristics could contribute to someone being a good perceiver.

Personality

Perhaps people who are more talkative and expressive, such as those high in extraversion, might be better perceivers, as their greater expressiveness could provide metaperceivers with more relevant information and context during social interactions. Further, those higher in agreeableness and openness (Christiansen et al., 2005; Letzring, 2008) tend to be more accurate judges of others’ personalities. These perceivers may, in turn, facilitate greater meta-accuracy for the metaperceivers, as their impressions are likely to be more consistent, coherent, and grounded in observable behavior, making it easier for metaperceivers to accurately infer how they are viewed.

Intrapersonal Adjustment

Might interacting with more intrapersonally adjusted perceivers facilitate meta-accuracy? To our knowledge, only Mosch and Borkenau (2016) explored perceiver-based correlates of profile-wise meta-accuracy (self-esteem and personality disorders), but did not find significant links. However, it remains theoretically plausible that more intrapersonally adjusted perceivers could facilitate meta-accuracy. Such individuals may approach social interactions with greater confidence and ease, resulting in personality impressions that are more easily readable. In other words, just as better adjusted metaperceivers may engage in clearer self-expression, facilitating cue availability and relevance for self-views, so too might well-adjusted perceivers more clearly express their personality impressions, facilitating information availability and relevance for how they view the metaperceiver. This could, in turn, enable metaperceivers to form more accurate impressions about how they are viewed by better adjusted perceivers.

Interpersonal Adjustment and Appeal

Interacting with perceivers who are more interpersonally adjusted may also facilitate meta-accuracy. These perceivers might be better liked by metaperceivers, because they tend to be more socially skilled and have better social interactions (Goldman & Lewis, 1977). In turn, these perceivers might elicit greater attention and engagement from metaperceivers, which could facilitate the detection of feedback and relevant social cues that informs their metaperceptions. Additionally, more likeable perceivers might encourage expressiveness from the metaperceiver (Collins & Miller, 1994), allowing the metaperceiver to convey observable cues that both the perceiver and the metaperceiver can use to form accurate impressions and metaperceptions. Overall, then, while perceiver variance for meta-accuracy is modest, there is reason to believe that some people may consistently help metaperceivers form more accurate metaperceptions.

The Good Trait

Meta-accuracy might also depend on the properties of the traits being judged. Some traits may systematically elicit higher or lower levels of meta-accuracy due to their unique features. Traits can differ on their observability (how easily a trait can be detected from external behavior) and evaluativeness (how socially desirable or undesirable a trait is). These are the two main characteristics we focus on in the present research, as they have received the most attention in the literature on personality accuracy thus far (e.g., Funder & Dobroth, 1987; John & Robins, 1993; Leising et al., 2010; Mignault et al., 2023; Rau et al., 2021; Vazire, 2010; Wiedenroth et al., 2024). To date, these have not yet been directly examined in the context of meta-accuracy.

The profile-based approach is particularly well-suited for identifying “good traits”. Within this approach, the unit of analysis is the personality item rather than the broader trait, allowing us to test whether meta-accuracy systematically differs based on observability or evaluativeness of items. That is, do some types of personality items elicit more/less accurate metaperceptions than others, such that those that are more observability or less evaluative? Although these item-level indicators serve as proxies for broader trait dimensions, patterns of within-profile variation provide insight into which kinds of trait characteristics, on average, are more conducive to meta-accuracy.

Observability

First, observability refers to how directly visible, or external, a specific personality characteristic is to perceivers. For example, being enthusiastic and full of energy are highly observable characteristics in a first impression, whereas whether someone is a deep thinker might be less observable in a getting-acquainted context. Within the RAM, observability primarily influences the cue availability stage by determining how easily a trait can be noticed by others. When traits are highly observable, they are more clearly expressed in behavior during the interaction, providing a shared basis of information that both perceivers and metaperceivers can draw on, thereby increasing the likelihood of meta-accuracy. In contrast, relying on cues related to less observable, internal traits (e.g., being thoughtful) might not help meta-accuracy because perceivers have less access to this information, and therefore, are less likely to base their judgments on it. Past research has observed descriptively higher levels of meta-accuracy for traits such as extraversion (i.e., higher in observability) than for traits like neuroticism (i.e., lower in observability; Carlson & Kenny, 2012; Elsaadawy & Carlson, 2022). As such, the observability of the traits being judged is likely an important moderator of meta-accuracy, with high observability traits facilitating meta-accuracy.

Evaluativeness

Second, evaluativeness refers to the degree to which the trait is seen as socially desirable or undesirable (see Edwards, 1957; Edwards & Horst, 1953, for foundational work on social desirability). It is a measure of how positive or negative a connotation the characteristic carries. Traits with high evaluativeness are seen as either highly desirable (e.g., intelligent) or highly undesirable (e.g., careless). Traits low in evaluativeness are more neutral (e.g., sociability). Evaluativeness can affect multiple stages of the RAM, including availability, detection and utilization. Drawing from the accuracy literature, more evaluative traits have lower self-other agreement (John & Robins, 1993). One reason is that people are likely to have lower self-knowledge of these traits, for ego-centric and motivational reasons (Vazire, 2010). Because people are generally motivated to maintain a positive view of themselves (Greenwald, 1980; Paulhus & John, 1998; Robins & Beer, 2001; Taylor & Brown, 1988), it may be harder for people to objectively utilize information to form judgments on evaluative traits when there is a competing motivation for self-enhancement. Furthermore, evaluative traits might also limit cue availability because perceivers may be less likely to share honest feedback on these traits. Aligning with this possibility, past work has descriptively observed lower meta-accuracy for more evaluative traits than for less evaluative, neutral traits (Carlson & Kenny, 2012).

Additionally, lower meta-accuracy for evaluative traits may also emerge because perceiver impressions on more evaluative items may be more idiosyncratic, as they may be more affected by unique perceiver attitudes than shared metaperceiver characteristics (Leising et al., 2015). The more idiosyncratic perceivers’ attitudes toward a metaperceiver are, the noisier those ratings will be, reducing inter-rater agreement and by extension meta-accuracy, which is a form of inter-rater agreement. As such, evaluativeness may reduce meta-accuracy through both motivational processes of the metaperceiver and more idiosyncratic perceiver impressions. Together, this implies that evaluativeness is likely an important moderator of meta-accuracy.

Good Information

Another moderator of meta-accuracy could be the quantity (e.g., conversation length, acquaintanceship) and the quality (e.g., diagnostic value) of information, which can vary across different interaction contexts. While both information quantity and quality are expected to positively impact meta-accuracy via cue availability (e.g., Elsaadawy & Carlson, 2021), we keep them consistent in this research by focusing on first impressions, with interactions that are standardized in length and very short (2-3 minutes). We, therefore, do not expect much variability in terms of quality or quantity of information. Further, by keeping this aspect relatively constant, we are able to more rigorously examine whether the characteristics of the metaperceiver, perceiver, and traits play a role in meta-accuracy.

Overview of Present Research

Overall, the primary aim of the present research is to explore the role of three moderators of meta-accuracy: (a) the good metaperceiver, (b) the good perceiver, and (c) the good trait. In additional analyses, we also controlled for multiple sources of meta-accuracy (e.g., self-perceptions, positivity, perceiver effects) to better understand the mechanisms through which some moderators operate.

Further, the RAM framework (Funder, 1995) also suggests that each moderator could interact with every other moderator. As such, in supplemental analyses, we explored the interactions between the good metaperceiver and the good trait as well as the good perceiver and the good trait (see Supplemental Online Materials (SOM)). We examined these questions across two large first-impression studies, using both in-person (N = 863, Dyads = 4582) and Zoom interactions (N = 886, Dyads = 4818). While these settings differ in the medium of communication, they allow us to determine the replicability and generalizability of our findings.

Method

Transparency and Openness

All studies were approved by the institutional review board (protocol # 178-1015; title: Social Consequences of First Impressions Study). We adhere to the highest standards of transparency by reporting how sample sizes were determined, detailing all data exclusions, and making all study materials, datasets, and R code publicly available for replication (https://osf.io/ztf8g/). Data analysis was conducted using R (version 4.2.3; R Core Team, 2020) and the lme4 package (Bates et al., 2015). The present research was not preregistered. Portions of the dataset have been used in previous publications to examine different research questions, but the moderators and analytic approach presented here are novel and have not been previously reported. Importantly, we implemented a consistent methodological approach across both samples, using identical measures and analytical strategies. This rigorous and systematic design strengthens the reliability and interpretability of our findings.

Sample

All participants were recruited from a university community. In both studies, participants were compensated with either 2-course credits or $20.00 for their participation.

Study 1: In-Person Study

A total of 863 participants took part in this study across four waves of data collection (wave 1: September 2016 – April 2017; wave 2: September 2017 – April 2018; wave 3: September 2018 – April 2019; wave 4: September 2019 – March 2020). Participants were recruited with flyers and ads stating that the study involved forming first impressions. Occasionally, dyads who happened to already know each other before the study (e.g., from a class) were excluded from analyses (n _dyads = 242), representing 5.3% of the total number of dyadic interactions. One session was excluded from analyses because only 3 participants showed up, indicating that each participant only had 2 interactions, which is insufficient to obtain reliable indices of meta-accuracy. As such, the present analyses involved a final sample of 860 participants and 4579 dyadic interactions (M_age = 20.37, SD_age = 2.18; 82.91 % women; 13.14% men and 1.51% did not identify as either a man or a woman).

Study 2: Zoom Study

In total, 975 participants took part in this study over the course of two waves (wave 1: September 2020 – May 2021; wave 2: September 2021 – August 2022). As in Study 1, participants were informed during recruitment that they would be forming impressions of unfamiliar others. Of these, 39 people did not complete the Zoom interaction portion of the study (i.e., only completed the initial survey). Further, 17 participants did not complete the initial survey (i.e., only completed the Zoom interaction portion). We excluded 4 Zoom sessions where 3 or fewer people were present, resulting in an exclusion of 8 additional participants. We also excluded all dyadic interactions where participants reported knowing each other before the session (n _dyads = 157), representing 3.2% of the total number of dyadic interactions. Thus, our final sample consisted of 911 participants and 4897 dyadic interactions (M_age = 20.45, SD_age = 2.59; 82.74% women, 15.49% men, 1.43% did not identify as either a man or a woman).

Statistical Power

We conducted expected power analyses using the fabs package for R (github\jbiesanz\fabs; also see Biesanz & Schrager, 2017; McShane & Böckenholt, 2016) to ensure these studies provided sufficient power for the present analyses. We based our power analyses on previous research that examined individual differences (social anxiety) in meta-accuracy employing the same studies (Tissera et al., 2021; Tissera et al., 2023). The effect sizes for the association between social anxiety and meta-accuracy ranged between r = .12 and .20. After incorporating the uncertainty of the effect estimates, to detect a similar effect on meta-accuracy, Study 1 provides 84% – 98% power and Study 2 provides 85% – 98% power, indicating that the present samples are likely adequately powered for the present research purposes.

Procedure

Study 1: In-Person Study

Participants came into the lab in groups of four to eight. Participants first provided ratings of their own personalities. Then, following a round-robin design, participants engaged in brief, unstructured one-on-one getting acquainted interactions, each lasting 2-3 minutes, with every other participant who was at that session. Immediately following each dyadic interaction, participants provided their impressions of their partner’s personality, their metaperceptions of personality, as well as the extent to which they thought their partner was attractive and engaging. After having met with every participant at that session, participants completed another survey in which they reported on multiple individual difference measures. Participants were compensated two-course credits or $20.00 for their participation in this study.

Study 2: Zoom Study

This study consisted of two parts. First, two to three days before the Zoom session, interested participants were invited to complete an online questionnaire in which they reported on their personalities and other individual differences. Second, participants joined a Zoom session with three to eight other participants. All participants were requested to have their video and audio on for all interactions. As in Study 1, following a round-robin design, participants met with every other participant at that session for 2 minutes in a break-out room, which was monitored by a research assistant. During the interaction, the research assistant’s video and audio were off to minimize distractions for the participants. For consistency, participants were instructed to set their screen on ‘gallery mode’ and to have the self-view on. Following each dyadic interaction, participants reported on their partner’s personality, their metaperceptions of personality, and their partner’s attractiveness and how engaging they were. Participants were compensated with 2-course credits or $20.00 for their participation.

Measures

All measures below were obtained on a 7-point Likert scale ranging from 1 = disagree strongly to 7 = agree strongly.

Personality Impressions and Metaperceptions

Following each getting-acquainted dyadic interaction (in-person or on Zoom), participants provided ratings of their interaction partner’s personality and metaperceptions of personality. These ratings were obtained using a subset of 10 items from the BFI (see Table 1 for the list of items), which included 2 items for each of the big five traits (also see study codebooks available online on open science framework). In addition, we also included 2 items to index intelligence (for examples of other published work using these items, see Human et al., 2012; 2014; Lorenzo et al., 2010; Kerr et al., 2020; Tissera et al., 2021; Tissera et al., 2023).

Table 1.

Personality Item Means and Standard Deviations

Item	Metaperception		Perceiver impressions		Observability	Evaluativeness
	Study 1	Study 2	Study 1	Study 2	Mean	Mean
Is intelligent	5.11 (0.95)	5.08 (0.97)	5.83 (0.82)	5.82 (0.85)	5.10 (1.09)	2.29 (0.08)
Tends to be quiet	3.44 (1.37)	3.44 (1.40)	3.88 (1.55)	3.97 (1.55)	6.00 (0.89)	0.89 (0.02)
Makes plans and follows through with them	4.81 (0.94)	4.86 (1.00)	5.19 (1.03)	5.23 (1.02)	4.90 (1.60)	2.19 (0.03)
Is outgoing, sociable	5.24 (1.04)	5.27 (1.06)	5.22 (1.20)	5.16 (1.23)	6.27 (0.64)	0.42 (0.003)
Tends to find fault with others	3.18 (1.09)	3.04 (1.14)	3.14 (1.22)	3.04 (1.22)	4.33 (1.95)	6.87 (0.16)
Is depressed, blue	2.59 (1.02)	2.49 (1.16)	2.61 (1.08)	2.75 (1.22)	4.10 (1.86)	7.22 (0.001)
Is original, comes up with new ideas	4.56 (0.95)	4.57 (1.01)	4.94 (1.04)	4.94 (1.08)	4.31 (1.65)	0.89 (0.001)
Is helpful and unselfish with others	4.99 (0.93)	4.98 (0.96)	5.31 (1.00)	5.29 (1.06)	5.03 (1.59)	2.00 (0.01)
Can be somewhat careless	3.64 (1.22)	3.63 (1.29)	3.30 (1.23)	3.24 (1.26)	4.73 (1.44)	7.04 (0.001)
Is relaxed, handles stress well	4.68 (1.06)	4.56 (1.15)	4.89 (1.13)	4.85 (1.16)	5.03 (1.13)	1.64 (0.12)
Is ingenious, a deep thinker	4.43 (1.02)	4.47 (1.08)	4.79 (1.11)	4.86 (1.10)	3.67 (1.63)	0.83 (0.01)
Is bright	5.21 (0.87)	5.17 (0.92)	5.60 (0.87)	5.64 (0.86)	4.87 (1.25)	0.89 (0.0001)

Note. The standard deviations appear within parentheses. Item evaluativeness scores were obtained by first centering desirability ratings provided by coders, then squaring them.

Moderators of the Good Metaperceiver and Good Perceiver

Table 2 provides details regarding all of the measures we used to examine the metaperceiver and perceiver moderators. ⁴ For correlations among the individual-level moderators, see Figure S1 in the SOM. Overall, the moderators showed small-to-moderate associations with one another, indicating some shared variance but not strong redundancy among the constructs (absolute correlations ranged from .01 to .59).

Table 2.

Summary of Descriptive Statistics for the Good Metaperceiver and Good Perceiver Moderators

Measure	Reference/Item	# of items	Descriptives for study 1			Descriptives for study 2
			Mean	SD	Reliability ( $\mathit{\alpha }$ )	Mean	SD	Reliability ( $\mathit{\alpha }$ )
Personality traits	John & Srivastava (1999)
Extraversion		8	4.43	1.04	.87	4.52	1.08	.87
Agreeableness		9	5.23	0.79	.78	5.14	0.83	.78
Conscientiousness		9	4.83	0.85	.80	4.70	0.91	.82
Openness		10	4.94	0.85	.81	4.92	0.81	.78
Neuroticism		8	4.26	1.08	.85	4.43	1.10	.86
Narcissism	Back et al. (2013)
Admiration		4	4.02	1.08	.68	4.15	1.22	.72
Rivalry		2	2.44	1.16	.57	2.50	1.25	.56
Self-esteem	Rosenberg (1965)	10	4.94	1.04	.89	4.76	1.11	.90
Life satisfaction	Diener et al. (1985)	5	4.79	1.18	.84	4.60	1.21	.84
Loneliness	Russell et al. (1978)	20	3.43	0.77	.81	3.50	0.77	.81
Positive relations with others	Ryff (1989)	3	5.31	1.14	.66	5.14	1.14	.59
Perceived physical attractiveness	‘is physically attractive’	1	5.04	0.77	-	5.21	0.73	-
Perceived likeability	‘is likable’	1	5.61	0.53	-	5.65	0.52	-

Note. Study 1 = In-person context, Study 2 = Zoom context. Reliability was indexed by calculating Cronbach’s alpha.

All individual difference measures were obtained using a questionnaire presented at the start of the session in Study 1 and in the online pre-event questionnaires for Study 2, except for two measures, perceived physical attractiveness (Lorenzo et al., 2010; Tissera et al., 2023) and perceived likeability (Mignault et al., 2023), which were indexed by perceiver ratings following each of the interactions. We indexed participants’ physical attractiveness and likeability by saving out the empirical Bayes (EB) estimates derived from linear mixed-effects models that included random intercepts for both the metaperceiver and the perceiver. These models captured impressions made during round-robin. We extracted the EB estimate for each metaperceiver, which reflected how attractive/likeable each metaperceiver was perceived to be across their interactions, while adjusting for perceiver-level biases. For example, Mira’s physical attractiveness was computed by extracting the empirical Bayes estimates capturing the average ratings of physical attractiveness as rated by Paulo, Pam, Pablo, and Priya (Mira’s different interaction partners), accounting for individual differences in their rating tendencies. This reflects the extent to which Mira’s is perceived as attractive by others on average. This approach has been previously used by research examining the effects of physical attractiveness and accuracy of personality impressions (Lorenzo et al., 2010; Tissera et al., 2023). We then merged the EB scores into the full dataset one by metaperceiver ID and once by perceiver ID. In doing so, we were able to obtain perceived physical attractiveness and perceived likeability indices for both metaperceivers and perceivers.

Good Trait Moderators

To establish the observability and evaluativeness of items, all items were rated for how socially desirable they were by a sample of coders (N = 106) who were drawn from the same population as the present samples. Coders rated all items using the same scale as participants, and ratings demonstrated a high degree of interrater reliability (ICC_{observability} (2,k) = .96; ICC_desirability (2,k) = .99). The correlation between the average observability and average social desirability ratings was r = −0.01. We computed an average observability score for each item by aggregating across coders. Evaluativeness was computed by averaging the social desirability ratings and then squaring the grand-mean centered social desirability ratings to remove the directionality of the desirability. As such, high scores reflected greater evaluativeness. See Table 1 for descriptive statistics.

Analytical Approach

Operationalization of Meta-Accuracy

As mentioned above, in the present research, we employ a profile approach to index meta-accuracy, an approach that has been used in several past studies of meta-accuracy (e.g., Carlson, 2016a; Carlson, 2016b; Carlson et al., 2010; Elsaadawy et al., 2021, 2022; Hater et al., 2023; Mosch & Borkenau, 2016; Tissera et al., 2021; Tissera & Lydon, 2022; Tissera et al., 2023).

We ran two separate sets of analyses. First, we examined the moderators of the good metaperceiver, perceiver, and trait for distinctive meta-accuracy only controlling for meta-normativity, following much of the prior work that has focused on meta-accuracy as a unique construct separate from broader normativity/positivity biases (e.g., Carlson, 2016a, 2016b; Elsaadawy et al., 2021; Tissera & Lydon, 2022). This modeling approach is also motivated by evidence that both self- and other-ratings systematically reflect substance-independent evaluative (halo) variance (Anusic et al., 2009), making it important to isolate distinctive meta-accuracy controlling for such normative or positivity-related effects. Halo effects are of particular concern in profile analyses because items across a profile differ in social desirability, which can result in items pulling for perceivers’ tendencies for positive or negative evaluation to varying degrees (Leising et al., 2015). Accordingly, our models aim to account for these evaluative components by isolating distinctive meta-accuracy.

Second, we ran an extended model in which we statistically controlled for multiple sources of meta-accuracy simultaneously: normative profile, socially desirable profile, distinctive perceiver effect profile and metaperceiver self-ratings profile. In this more stringent model, our primary indicator of meta-accuracy is indexed by ‘dyadic distinctive meta-insight’, or meta-perceivers’ insight into others’ unique impressions of them, controlling for other potential sources of meta-accuracy, including how people generally tend to be (meta-normativity), how positively they believe they are seen (meta-positivity), how perceivers typically view others (distinctive perceiver effect meta-accuracy), and metaperceivers’ own self-views (distinctive meta-transparency). See Table 3 for an overview of the terminology.

Table 3.

Overview of Profile Types and Profile Agreement Terms

Term	Notation	Definition
Model 1
Types of profiles
Metaperception profile	M mpi	The metaperceiver’s beliefs about how they are seen by a specific perceiver on each personality item.
Distinctive impressions profile	D mpi	Perceivers’ unique impressions of metaperceiver (subtracting out the normative profile).
Normative profile	N i	The profile representing the average or typical personality profile (average self-report).
Profile agreement
Distinctive meta-accuracy	${\beta }_{1mp}$	How accurately metaperceivers track perceivers’ unique impressions of them, controlling for meta-normativity.
Meta-normativity	${\beta }_{2mp}$	The degree to which metaperceptions align with the average person’s self-rated personality profile.
Model 2
Additional types of profiles
Distinctive perceiver effect profile	PE pi	The perceiver’s general (average) way of viewing others, computed across metaperceivers for each perceiver and item, after subtracting the normative profile.
Socially desirable profile	P i	The profile representing the general desirability of each personality trait using the average of independent coder ratings of how desirable each item is.
Self-ratings profile	${SR}_{mi}$	Metaperceiver unique self-ratings (after subtracting the normative profile).
Profile agreement
Dyadic distinctive meta-insight	${\beta }_{1mp}$	The extent to which metaperceivers track the perceivers’ unique impressions of metaperceivers, controlling for meta-normativity, perceiver effect meta-accuracy, meta-positivity, and meta-transparency.
Meta-normativity (controlling for meta-positivity)	${\beta }_{2mp}$	The degree to which metaperceptions align with the average person’s self-reported personality profile, controlling for meta-positivity.
Distinctive perceiver effect meta-accuracy	${\beta }_{3mp}$	How accurately metaperceivers track perceivers’ general tendencies to see others, after accounting for normative impressions, social desirability, and the metaperceiver’s own self-views.
Meta-positivity	${\beta }_{4mp}$	The association between metaperceptions and the social desirability (positive) personality profile.
Distinctive meta-transparency	${\beta }_{5mp}$	The association between metaperceivers and metaperceivers’ their unique self-ratings, controlling for normativity and positivity.

Note. Subscripts indicate the unit of analysis. m refers to metaperceiver, p refers to perceiver, and i refers to personality item. Profiles indexed by mpi vary across items within metaperceivers and perceivers. Meta-normativity is defined consistently across models as the extent to which metaperceptions align with the normative profile (derived from aggregated self-ratings), but is estimated with different covariates in Model 1 and Model 2.

Indexing Meta-Accuracy Using the Social Accuracy Model

We adapted the multilevel modelling guidelines from the social accuracy model (SAM: Biesanz, 2010, 2021; see also Tissera et al., 2021; Tissera et al., 2023) to account for the dependencies within the data (i.e., multiple ratings by perceivers and metaperceivers of one another). We ran two models. In the first model, we assessed distinctive meta-accuracy while controlling for meta-normativity. Although meta-normativity is a potential source of meta-accuracy, we control for it in Model 1, as this is common practice when assessing profile meta-accuracy. Then, in the second model, we assessed distinctive meta-accuracy controlling for meta-normativity, but also distinctive perceiver effect meta-accuracy, meta-positivity and distinctive meta-transparency. Both models involved estimating a crossed-random effects model designed for datasets such as these to examine different components of (meta-)accuracy. Following the SAM guidelines, items were not reverse-coded prior to analyses to allow an adequate spread in the personality profiles.

In Model 1, as illustrated in Equations 1.1 and 1.2 below, we predicted metaperception profiles ( $M_{mpi})$ from two profiles (see Table 3 for notation definitions):

$M_{mpi}={\beta }_{0mp}+{\beta }_{1mp}{D}_{mpi}+{\beta }_{2mp}{N}_i+{\epsilon }_{mpi}$

(1.1)

${\beta }_{1mp}=b_{10}+u_{1m}+u_{1p}+u_{1mp}$

(1.2)

${\beta }_{2mp}=b_{20}+u_{2m}+u_{2p}+u_{2mp}$

As such, ${\beta }_{1mp}$ reflects distinctive meta-accuracy while ${\beta }_{2mp}$ reflects meta-normativity. The coefficient $b_{10}$ reflects the average levels of distinctive meta-accuracy and $b_{20}$ reflects average levels of meta-normativity across the sample (also see Table 3).

In the second, expanded model, we predicted metaperception profiles ( $M_{mpi})$ from five profiles, as illustrated in equations 2.1 and 2.2:

$M_{mpi}={\beta }_{0mp}+{\beta }_{1mp}{D}_{mpi}+{\beta }_{2mp}{N}_i+{\beta }_{3mp}{PE}_{pi}+{\beta }_{4mp}{P}_i+{\beta }_{5mp}{SR}_{mi}+{\epsilon }_{mpi}$

(2.1)

${\beta }_{1mp}=c_{10}+u_{1m}+u_{1p}+u_{1mp}$

(2.2)

${\beta }_{2mp}=c_{20}+u_{2m}+u_{2p}+u_{2mp}$

${\beta }_{3mp}=c_{30}+u_{3m}+u_{3p}+u_{3mp}$

${\beta }_{4mp}=c_{40}+u_{4m}+u_{4p}+u_{4mp}$

${\beta }_{5mp}=c_{50}+u_{5m}+u_{5p}+u_{5mp}$

Here, the coefficient $c_{10}$ reflects the average levels of dyadic distinctive meta-insight, but for simplicity and for ease of comparison with prior work, we use distinctive meta-insight interchangeably with dyadic distinctive meta-insight. Further, $c_{10}$ reflects average levels of meta-normativity (controlling for meta-positivity), the coefficient $c_{30}$ reflects average levels of distinctive perceiver-effect meta-accuracy, $c_{40}$ reflects average levels of meta-positivity, and $c$ ₅₀ reflects average levels of distinctive meta-transparency (also see Table 3). Finally, to take into account the crossed nature of the present data, due to the round-robin design, we allowed all random slopes to vary by metaperceivers ( $u_{1m},u_{2m},\dots$ ), perceivers ( $u_{1p},u_{2p},\dots$ ), and dyads ( $u_{1mp},u_{2mp},\dots$ ), when possible. At times, the random effects structure was simplified to address convergence issues.

To examine the moderators of the good metaperceiver, good perceiver, and good trait, we added these as a moderators of each slope. For instance, a grand-mean centered moderator of the good metaperceiver (e.g., self-reported extraversion) would be included at the person-level of the model, as illustrated in the equation 3 below which builds on equation 2.2 above.

${\beta }_{1mp}=c_{10}+c_{11}{Moderator}_m+u_{1m}+u_{1p}+u_{1mp}$

(3)

${\beta }_{2mp}=c_{20}+c_{21}{Moderator}_m+u_{2m}+u_{2p}+u_{2mp}$

${\beta }_{3mp}=c_{30}+c_{31}{Moderator}_m+u_{3m}+u_{3p}+u_{3mp}$

${\beta }_{4mp}=c_{40}+c_{41}{Moderator}_m+u_{4m}+u_{4p}+u_{4mp}$

${\beta }_{5mp}=c_{50}+c_{51}{Moderator}_m+u_{5m}+u_{5p}+u_{5mp}$

Within this modelling approach, $c_{11}$ reflects the fixed association between the moderator and dyadic distinctive meta-insight, $c_{21}$ reflects the fixed association between the moderator and meta-normativity, $c_{31}$ reflects the fixed association between the moderator and perceiver effect meta-accuracy, $c_{41}$ reflects the fixed association between the moderator and meta-positivity, and $c_{51}$ reflects the fixed association between the moderator and distinctive meta-transparency.

Of note, across all moderator models, given our primary interest in identifying individual differences in the good metaperceiver and perceiver, we estimated random slopes at the metaperceiver and perceiver levels only. This approach reduces model complexity and mitigated convergence issues while still capturing the sources of variance central to our research question. This decision is consistent with previous work using similar modeling strategies (Hater et al., 2023; Rogers & Biesanz, 2019). In this way, in each study, we estimated two baseline models. Then, for each variation, we estimated 13 models testing moderators of the good metaperceiver, 13 models testing moderators of the good perceiver, and 2 models testing moderators of item features.

Of note, while meta-normativity was estimated in both models, the meaning does change, given that normative and positive profiles tend to be very highly correlated (e.g., Edwards, 1957; Wood & Furr, 2016). Thus, in the first model, meta-normativity includes meta-positivity, whereas in the second model, meta-normativity is independent of meta-positivity. Accordingly, we report and interpret correlates of meta-normativity only from the expanded model, where the overlap with the socially desirable profile is separated out.

Results

We first examined the baseline social accuracy model to assess the presence of distinctive meta-accuracy and how it varies across metaperceivers, perceivers, and dyads. As shown in Table 4, in both studies, people displayed significant levels of distinctive meta-accuracy, indicating that people were able to recognize how others viewed them, controlling for the normative profile. These results align with past work examining distinctive meta-accuracy (e.g., Carlson, 2016b; Elsaadawy et al., 2021; Tissera et al., 2021; Tissera et al., 2023).

Table 4.

Baseline Levels

Model Parameter	Symbol	Model 1
		Study 1		Study 2
		Estimate	SE	Estimate	SE
Fixed Effects
Distinctive meta-accuracy	${\beta }_{1mp}$	0.22**	0.01	0.22**	0.01
Meta-normativity	${\beta }_{2mp}$	0.85**	0.02	0.93**	0.02
Random Slopes by Metaperceiver
Distinctive meta-accuracy	$u_{1m}$	0.14		0.13
Meta-normativity	$u_{2m}$	0.46		0.49
Random Slopes by Perceiver
Distinctive meta-accuracy	$u_{1p}$	0.07		0.08
Meta-normativity	$u_{2p}$	0.02		0.05
Random Slopes by Dyad
Distinctive meta-accuracy	$u_{1mp}$	0.12		0.12
Meta-normativity	$u_{2mp}$	0.004		0.03

Note. The random slopes are standard deviations of the respective random slope (i.e., the random variable around the fixed effect that is attributable to the metaperceiver, the perceiver, or the dyad). **p < .001. Study 1 = In-person; Study 2 = Zoom.

To better understand the mechanisms that contribute to distinctive meta-accuracy, we also fit the extended social accuracy model to examine (dyadic) distinctive meta-insight. In this model, we additionally controlled for the distinctive perceiver effect profile, socially desirable profile, and metaperceivers’ self-ratings (as detailed above in the analytical approach). In this extended model (Table 5), distinctive meta-insight remained significant, suggesting that people are able to gauge how the perceiver viewed them even after controlling for their distinctive perceiver effects, meta-normativity, meta-positivity and distinctive meta-transparency. Further, we also observed significant levels of meta-normativity, distinctive perceiver effect meta-accuracy, and distinctive meta-transparency as well.

Table 5.

Baseline Levels

Model Parameter	Symbol	Model 2
		Study 1		Study 2
		Estimate	SE	Estimate	SE
Fixed Effects
Distinctive meta-insight	${\beta }_{1mp}$	0.09**	0.01	0.09**	0.01
Meta-normativity	${\beta }_{2mp}$	0.35**	0.01	0.39**	0.01
Perceiver effect meta-accuracy	${\beta }_{3mp}$	0.12**	0.01	0.07**	0.01
Meta-positivity	${\beta }_{4mp}$	0.64**	0.01	0.70**	0.01
Distinctive meta-transparency	${\beta }_{5mp}$	0.25**	0.01	0.24**	0.01
Random Slopes by Metaperceiver
Distinctive meta-insight	$u_{1m}$	0.13		0.13
Meta-normativity	$u_{2m}$	0.31		0.31
Perceiver effect meta-accuracy	$u_{3m}$	0.11		0.09
Meta-positivity	$u_{4m}$	0.29		0.32
Distinctive meta-transparency	$u_{5m}$	0.20		0.20
Random Slopes by Perceiver
Distinctive meta-insight	$u_{1p}$	0.002		0.06
Meta-normativity	$u_{2p}$	0.10		0.06
Perceiver effect meta-accuracy	$u_{3p}$	0.13		0.11
Meta-positivity	$u_{4p}$	0.03		0.03
Distinctive meta-transparency	$u_{5p}$	0.004		0.13
Random Slopes by Dyad
Distinctive meta-insight	$u_{1mp}$	0.12		0.00
Meta-normativity	$u_{2mp}$	0.11		0.11
Perceiver effect meta-accuracy	$u_{3mp}$	0.11		0.16
Meta-positivity	$u_{2mp}$	0.03		0.06
Distinctive meta-transparency	$u_{3mp}$	0.04		0.04

Note. The random slopes are standard deviations of the respective random slope (i.e., the random variable around the fixed effect that is attributable to the metaperceiver, the perceiver or the dyad). **p < .001. Study 1 = In-person; Study 2 = Zoom.

To better understand the sources of variability in distinctive meta-accuracy components, we examined the random slope standard deviations. Following prior published work (Elsaadawy et al., 2021), we interpret these SD estimates as effect sizes. We follow the interpretive guidelines proposed in prior work (Biesanz, 2021; Elsaadawy et al., 2021), which classify SDs of ∼.05 as small, ∼.10 as moderate, and ∼.24 or higher as large. Applying these guidelines, we observed moderate variance across metaperceivers for distinctive meta-accuracy (see Table 4). Further, perceiver-level variance in distinctive meta-accuracy was small-to-moderate, indicating limited variability across perceivers, aligning with prior work (Elsaadawy et al., 2023). Finally, dyadic variance was also moderate in both studies, suggesting that meta-accuracy may also vary across specific dyads in virtual contexts. These effects remained largely similar in the extended model as well, with a couple of exceptions: the perceiver variance in Study 1 and dyadic variance in Study 2 were observed to be very small/negligible (see Table 5).

Results From Model 1: Moderators of Distinctive Meta-Accuracy

The Good Metaperceiver

What characterizes a good metaperceiver? In both studies, we observed a consistent pattern (see Table 6): higher levels of self-reported extraversion, agreeableness, conscientiousness, self-esteem, life satisfaction, positive relations with others, as well as perceiver-rated physical attractiveness and likeability during interactions were all significantly linked with greater distinctive meta-accuracy. Lower self-reported neuroticism, narcissism and loneliness were also consistently linked with greater distinctive meta-accuracy. The only indicators that were not significantly associated with distinctive meta-accuracy in both studies were narcissistic admiration and openness, though the association with openness was in the expected direction in Study 2, despite not reaching statistical significance.

Table 6.

Summary of Associations With Distinctive Meta-Accuracy, Controlling for Meta-Normativity (Model 1)

Moderator	Study 1			Study 2
	b	r	t	b	r	t
Metaperceiver
Extraversion	0.05**	.30 [.23, .36]	8.81	0.04**	.26 [.18, .34]	6.13
Agreeableness	0.05**	.22 [.15, .28]	6.25	0.07**	.32 [.24, .39]	7.51
Neuroticism	−0.04**	−.22 [-.28, −.15]	−6.42	−0.04**	−.25, [-.33, −.17]	−5.84
Openness	0.03**	.12 [.06, .19]	3.53	0.02†	.08 [.00, .17]	1.86
Conscientiousness	0.04**	.17 [.10, .24]	4.99	0.04**	.23 [.14, .31]	5.22
Narcissistic admiration	0.01	.04 [-.03, .11]	1.17	−0.003	−.02 [-.11, .07]	−0.48
Narcissistic rivalry	−0.02**	−.12 [-.19, −.05]	−3.47	−0.03**	−.22 [-.30, −.13]	−5.08
Self-esteem	0.05**	.27 [.21, .33]	8.06	0.05**	.29 [.21, .36]	6.82
Life satisfaction	0.03**	.20 [.13, .26]	5.63	0.04**	.26 [.18, .34]	6.17
Loneliness	−0.04**	−.16 [-.23, −.09]	−4.59	−0.06**	−.25 [-.33, −.17]	−5.83
Positive relations with others	0.03**	.21 [.15, .28]	6.22	0.05**	.30, [.22, .37]	7.00
Perceived physical attractiveness	0.06**	.26 [.19, .33]	7.34	0.04**	.13 [.05, .22]	3.03
Perceived likeability	0.12**	.35 [.28, .41]	10.08	0.08**	.21 [.13, .30]	4.89
Perceiver
Extraversion	−0.001	−.01 [-.08, .06]	−0.25	0.01	.05 [-.04, .14]	1.06
Agreeableness	0.02**	.13 [.04, .20]	3.09	0.02*	.11 [.02, .20]	2.30
Neuroticism	−0.01*	−.08 [-.15, −.01]	−2.12	−0.01†	−.09 [-.18, .00]	−1.91
Openness	0.004	.02 [-.05, .10]	0.63	−0.002	−.01 [-.11, .08]	0.78
Conscientiousness	0.01	.04 [−.03, .12]	1.18	0.003	.02 [-.07, .11]	0.42
Narcissistic Admiration	−0.003	−.02 [-.10, .05]	−0.57	−0.01†	−.08 [-.17, .01]	−1.67
Narcissistic Rivalry	−0.01	−.06 [-.13, .02]	−1.51	−0.01*	−.11 [-.19, −.02]	2.35
Self-esteem	0.01	.05 [-.03, .12]	1.24	0.003	.02 [-.07, .11]	0.47
Life satisfaction	0.002	.02 [-.06, .09]	0.44	0.01†	.09 [-.01, .18]	1.83
Loneliness	0.001	.00 [-.07, .08]	0.12	0.001	0.00 [-.09, .10]	0.06
Positive relations with others	0.01*	.09 [.01, .16]	2.32	0.01*	.10 [.01, .20]	2.20
Perceived physical attractiveness	0.001	.00 [-.07, .08]	0.12	0.01†	.09 [.00, .17]	1.92
Perceived likeability	0.03*	.10 [.03, .17]	2.77	0.02	.07 [-.02, .16]	1.56
Trait
Observability	0.09**	.09 [.08, .10]	20.52	0.04**	.04 [.03, .05]	6.72
Evaluativeness	−0.02**	−.07 [-.08, −.06]	−17.06	−0.02**	−.06 [-.07, −.05]	−11.14

Note. **p < .01, *p < .05, †p < .10. Study 1 = In Lab, Study 2 = Zoom.

Accounting for General Metaperceiver Positivity

Given that all positive moderators were consistently positively related to being a good metaperceiver, while negative moderators showed the opposite pattern, we conduct post-hoc analyses to determine whether these associations could be explained by general positivity in self-views. As such, we conducted a one-factor exploratory factor analysis (EFA) on all the self-reported metaperceiver moderators (i.e., we excluded physical attractiveness and likeability, which were rated by perceivers). Most moderators had moderate to strong loadings (>.40) on the single factor, consistent with a general positivity factor. Narcissistic admiration, narcissistic rivalry, and openness showed weak loadings (<.40) and, as such, were excluded from the factor. After excluding these moderators, the single factor accounted for approximately 35% of the total variance across the remaining moderators. We then exported factor scores per participant and entered them into Model 1 as a moderator. General metaperceiver positivity was related to distinctive meta-accuracy in both Study 1 (b = .06, t = 9.74, r = .33 [.26, .38], p < .001) and Study 2 (b = .07, r = .39 [.31, .45], t = 9.37, p < .001).

Next, we tested the effect of each metaperceiver moderator again but this time with general metaperceiver positivity also included in the model (see Table S1-a in the SOM). In Study 1, self-reported extraversion, openness, and loneliness, as well as perceiver-rated attractiveness and likeability were positively related to distinctive meta-accuracy after controlling for general metaperceiver positivity, while self-reported life satisfaction was negatively related to distinctive meta-accuracy. In Study 2, self-reported agreeableness and perceiver-rated likeability were positively related to distinctive meta-accuracy after controlling for general metaperceiver positivity, while self-reported narcissism-rivalry was negatively related to distinctive meta-accuracy. In sum, most of the effects appear to be explained by general metaperceiver positivity: a general tendency to view the self positively. The only moderator that had a unique effect across both studies after accounting for general metaperceiver positivity was perceiver-rated likeability.

The Good Perceiver

Do some perceivers facilitate greater meta-accuracy? Across both studies, we found that metaperceivers displayed greater distinctive meta-accuracy when interacting with perceivers who reporting being higher on agreeableness and having more positive relations with others (see Table 6). Other perceiver traits, such as self-reported neuroticism, narcissistic rivalry and likeability were positively associated with distinctive meta-accuracy in one of the two studies (see Table 6), making it unclear whether these effects were reliable.

Accounting for General Perceiver Positivity

Paralleling the post-hoc analyses for the metaperceiver moderators, we used the positivity scores that were extracted per participant and included them as a moderator in Model 1. General perceiver positivity was unrelated to distinctive meta-accuracy in both Study 1 (b = .01, r = .07 [-.01, .14], t = 1.74, p = .08) and Study 2 (b = .01, r = .08 [-.01, .17], t = 1.78, p = .08). As such, we did not go on to test whether the effects of the perceiver moderators remained after controlling for general perceiver positivity.

The Good Trait

Are there specific types of items that might facilitate meta-accuracy? In both studies, we found that people displayed significantly greater distinctive meta-accuracy on items rated as more observable and less evaluative (see Table 6), suggesting that people may be more accurate about how they come across on more observable and less evaluative personality items.

Results From Model 2: The Expanded Model

Note that in this section, we only focus on the results that were significant in both samples, but see Figure 2 for all the significant effects observed in these models (also see Table S6 in the SOM).

Figure 2.

Effect Sizes of the Results From the Expanded Model (Model 2).

The Good Metaperceiver

In the expanded model, only two predictors were reliably associated with dyadic distinctive meta-insight: perceived physical attractiveness and likeability. Likeable metaperceivers also displayed greater meta-normativity, meta-positivity, and distinctive perceiver effect meta-accuracy. Nearly all metaperceiver reported characteristics were significantly related to meta-positivity in both samples, except for openness. Finally, metaperceivers who reported being higher on more desirable personality traits (with the exception of openness, conscientiousness, and both narcissism dimensions) and being better adjusted both intra- and interpersonally were associated with greater distinctive meta-transparency.

The Good Perceiver

In the expanded model, relatively few predictors emerged as consistent across both studies. Perceived likeability stood out as the most robust perceiver characteristic, predicting higher meta-normativity, meta-positivity, and distinctive meta-transparency. Perceived physical attractiveness also had a consistent positive association with distinctive meta-transparency, suggesting that more attractive perceivers might lead metaperceivers to believe they are seen line with their own self-views.

The Good Trait

Trait characteristics showed two unique patterns across components of metaperceptions. For observability, higher observability was consistently associated with higher distinctive meta-insight and higher distinctive meta-transparency across both studies. The associations with meta-normativity and meta-positivity were mixed: both were negatively related to observability in Study 1 but positively related in Study 2.

For evaluativeness, higher evaluativeness was consistently associated with lower dyadic distinctive meta-insight, lower distinctive perceiver effect meta-accuracy, and lower distinctive meta-transparency in both studies. In contrast, evaluativeness was positively associated with meta-normativity and meta-positivity across both studies. Together, these findings suggest that while high observability generally facilitates accuracy in distinctive judgments, evaluativeness is linked to reduced meta-accuracy but may promote meta-normativity and meta-positivity.

Accounting for General Positivity

As a robustness check, we applied the same general positivity analyses to Model 2. In contrast to Model 1, general metaperceiver positivity was unrelated to dyadic distinctive meta-insight in both studies, but was positively associated with meta-positivity (see Table S1-b in the SOM). This suggests that metaperceivers’ tendency to rate themselves positively (general meta-perceiver positivity) was related to their tendency to believe they were viewed positively by others but not their ability to track specific perceivers’ distinctive impressions of them. In contrast, general perceiver positivity was unrelated to both distinctive meta-insight and meta-positivity. Thus, perceivers’ general positivity was unrelated to both how positively meta-perceivers thought perceivers saw them and meta-perceivers’ ability to track perceivers’ impressions of them. As such, we did not go on to test the associations between dyadic distinctive meta-insight and metaperceiver or perceiver moderators controlling for general metaperceiver or perceiver positivity.

Discussion

Do characteristics of the metaperceiver, perceiver, and the trait being judged facilitate or hinder the ability to know how other people view us? In two large studies, encompassing nearly 9,500 getting-acquainted interactions both in person and via Zoom, we examined multiple moderators within each category. Overall, we found that characteristics of the metaperceivers, perceivers, and traits being judged were related to the accuracy with which people recognized how they come across to others. In essence, by taking a comprehensive approach, this research provides insight into who is more likely to be meta-accurate, who is more adept at enabling meta-accuracy for their interaction partners, and which attributes facilitate more accurate metaperceptions. Overall, this work advances the literature by integrating multiple moderators within a unified framework that extends the Realistic Accuracy Model (Funder, 1995) and related meta-accuracy research (Carlson & Kenny, 2012; Elsaadawy & Carlson, 2021).

Who Is a Good Metaperceiver?

Converging with past research (Elsaadawy et al., 2021), we found that meta-accuracy largely varied across different metaperceivers. This suggests that some people tended to be better at understanding how they come across to others. What might help explain this variability? We found that the good metaperceiver generally reported more desirable personality traits (e.g., higher extraversion, lower neuroticism), greater intra and interpersonal adjustment, and exhibited higher interpersonal appeal in first impression contexts. Notably, general metaperceiver positivity (i.e., the global positivity across self-reported metaperceiver variables) was strongly associated with distinctive meta-accuracy. Moreover, when we controlled for this general metaperceiver positivity, we found that only a few metaperceiver characteristics had unique associations with distinctive meta-accuracy, effects that were inconsistent across studies. In fact, perceiver-rated likeability was the only metaperceiver moderator for which we observed a unique association with distinctive meta-accuracy beyond general metaperceiver positivity. Taken together, these results indicate that metaperceivers who saw themselves positively and metaperceivers who were liked by perceivers were good metaperceivers. Perhaps metaperceivers who see themselves positively and/or are liked by perceivers are more likely to self-disclose and share more information about themselves with perceivers, allowing perceivers to form impressions that are more aligned with the metaperceivers’ self-views (which may or may not be accurate in an objective sense). Another possibility is that metaperceivers who see themselves positively and/or are liked by perceivers might behave more in line with their self-perceived personality traits, allowing perceivers to see them in line with how they see themselves. Both possibilities could facilitate meta-accuracy by allowing metaperceivers to use their self-perceptions of personality and/or their self-observations of behaviour with perceivers to more easily form accurate metaperceptions in first impression contexts.

How Do Good Metaperceivers Achieve Meta-Accuracy?

Pre-existing Sources of Meta-Accuracy

The expanded model revealed that for most metaperceiver moderators, the strongest and most consistent associations were with meta-positivity and distinctive meta-transparency, and, to a lesser extent, meta-normativity, all components that largely reflect pre-existing knowledge rather than unique information obtained from a specific interaction. For example, across both samples, a wide range of desirable personality traits predicted believing others saw the self positively (meta-positivity), and many also predicted believing others saw the self in line with one’s own self-views (distinctive meta-transparency). Thus, good metaperceivers in our studies appeared to achieve accuracy not by detecting idiosyncratic signals from each perceiver in a given interaction, but by drawing on broad, pre-existing information: (a) their general expectation of being seen positively, (b) how they tend to see themselves, and, to a lesser extent, (c) how they believe most people are seen (the normative profile).

Interaction-specific Cues

Our findings also point to another pathway through which metaperceivers can achieve accuracy: picking up on interaction-specific cues available in the moment. This alternative route to meta-accuracy was most evident among metaperceivers with particularly high interpersonal appeal, namely, those perceived as especially likeable or physically attractive. In both studies, these qualities reliably predicted components that depend on detecting cues unique to a given perceiver, such as dyadic distinctive meta-insight (knowing how a specific perceiver uniquely sees you) and perceiver effect meta-accuracy (calibrating to that perceiver’s general rating tendencies). Using interaction-specific cues of meta-accuracy likely requires the ability to attend closely to moment-to-moment social cues, such as subtle changes in facial expressions, body language, or tone of voice. In this context, it is perhaps unsurprising that the people who emerge as good metaperceivers are those who are socially skilled and interpersonally appealing. These are people who tend to be more attuned to social dynamics and who are better able to notice, interpret, and respond to interpersonal feedback as it unfolds (Murphy & Hall, 2021).

However, because our measure of interpersonal appeal came from ratings made during the same interactions in which we measured meta-accuracy, it is possible that we are capturing an outcome of meta-accuracy rather than a true predictor. People who are more meta-accurate may be liked more by their partners (e.g., Carlson, 2016b; Tissera et al., 2021; Tissera et al., 2023), which would then be reflected in higher ratings of likeability and physical attractiveness. Past research has found that meta-accuracy is linked to being liked and to romantic interest, which are both related to interpersonal appeal. This means the connection we observed could work in both directions. Future studies could address this by measuring interpersonal appeal or social skill before the interaction takes place, or by experimentally manipulating the level of social skill displayed during interactions (e.g., by using trained confederates) to see whether it leads to greater use of interaction-specific cues.

Who is a Good Perceiver?

Next, we examined whether certain interaction partners, or perceivers, might facilitate more accurate metaperceptions, enabling their impressions of the metaperceiver to be more easily read. While meta-accuracy varied across perceivers, it was smaller than metaperceiver variance, consistent with prior findings (Elsaadawy et al., 2021). This suggests that while some perceivers may be more likely to promote more accurate metaperceptions than others, the perceivers’ role is relatively modest compared to that of a metaperceiver.

Nevertheless, some predictors of the good perceiver emerged. In both studies, people displayed greater meta-accuracy when interacting with perceivers who were reportedly higher in agreeableness and who reported more positive relations with others. Other self-reported traits, such as neuroticism, the rivalry dimension of narcissism, and perceived likeability, were positively associated with meta-accuracy in only one of the two studies, making it unclear whether these effects are reliable. Overall, the most consistent predictors suggest that perceivers who report being more interpersonally warm and having more positive social relationships may create interaction contexts that facilitate meta-accuracy.

Why might agreeable perceivers with positive social connections facilitate greater meta-accuracy? Agreeable people tend to be warmer, more cooperative, and more empathetic in their interactions (Graziano & Eisenberg, 1997; Jensen‐Campbell & Graziano, 2001). These qualities are linked to greater willingness to provide social feedback and to respond in ways that are clear, consistent, and non-threatening (Letzring, 2008). Similarly, people who report more positive relations with others are likely to have strong interpersonal skills and to engage in behaviors that promote mutual understanding (Reis & Patrick, 1996). Such behaviors, whether verbal affirmations or clear nonverbal cues, may make a perceiver’s impressions easier to detect and interpret. In the context of first impressions, these perceivers may offer richer or more interpretable behavioral cues that allow metaperceivers to calibrate their judgments more accurately. In line with the Realistic Accuracy Model (Funder, 1995), these traits may increase the availability and detectability of relevant cues, thereby enhancing the likelihood that metaperceivers form accurate beliefs about how they are seen.

When examining the expanded model, relatively few perceiver-rated moderators consistently predicted metaperceivers’ judgments across both studies. Perceived likeability emerged as the most robust predictor, associated with higher meta-normativity, meta-positivity, and distinctive meta-transparency. Perceived physical attractiveness also consistently predicted distinctive meta-transparency. It is possible that these ratings of perceived likeability and attractiveness capture outcomes of the interaction itself, rather than purely stable perceiver characteristics, particularly given that the expanded model includes more predictors and leaves less variance attributable to perceivers. In fact, in the expanded model there was even less perceiver variance to explain than in the simpler models, making it more challenging to detect stable individual differences when their effects are relatively small. Nevertheless, the pattern that emerges suggests that good perceivers may be those who are interpersonally appealing and socially engaging. Perhaps these perceivers create social environments that encourage metaperceivers to feel understood and to believe their self-views are aligned with how they are perceived. Future research is needed to establish the robustness of these results.

What is a Good Trait?

We also investigated whether people were meta-accurate about certain traits more than others. Specifically, we examined two dimensions of traits: observability and evaluativeness. Notably, in our first model, we found that the good trait is more observable and less evaluative, consistent with previous observations (e.g., Carlson & Kenny, 2012; Elsaadawy & Carlson, 2022). These results suggest that meta-accuracy also varies depending on the specific item being judged.

The Role of Trait Observability

Observable traits, such as extraversion, are characterized by external manifestations that are easily perceived through cues like behavior and mannerisms. These cues provide tangible and concrete information that can be readily observed by both the metaperceiver and the perceiver. As such, people can more easily detect them and utilize them in their judgments (both metaperceptions and perceiver impressions). In the expanded model, we observed this pattern for distinctive meta-insight and distinctive meta-transparency, which were consistently higher for more observable traits. The utilization of the same observable cues by both the metaperceiver and the perceiver enhances the alignment between metaperceptions and the impressions formed by others. Furthermore, the association between observable traits and high meta-accuracy reinforces the significance of observation of behaviors during the interaction as a crucial route to accurate self-perception (Albright & Malloy, 1999). This process of self-observation allows individuals to align their metaperceptions with the impressions they make available to others, thereby enhancing meta-accuracy. Thus, these findings also provide further insight into the processes that fuel meta-accuracy. Of note, the role of observability was less consistent for the other components of meta-accuracy in the expanded model, making the robustness of these effects uncertain.

The Role of Trait Evaluativeness

Moreover, meta-accuracy was lower for more evaluative traits. This finding converges with research suggesting that people possess less accurate self-knowledge on these dimensions (Vazire, 2010). One reason for this could be rooted in ego-centric motives and self-serving biases (Greenwald, 1988; Robins & Beer, 2001; Taylor & Brown, 1988). People generally strive to maintain positive self-views, and accurately perceiving others’ less positive impressions on socially evaluative traits may pose a threat to one’s self-esteem. This notion suggests that people engage in self-enhancement on more evaluative traits.

At the same time, drawing on more recent developments in the accuracy literature, lower meta-accuracy for evaluative traits may also arise because such items increasingly reflect perceivers’ idiosyncratic attitudes toward the target rather than shared trait information. When perceivers’ attitudes toward a given target are weakly correlated with one another, ratings on highly evaluative items become noisier, reducing inter-rater agreement and, in turn, limiting the extent to which metaperceivers can accurately infer how they are seen (Leising et al., 2015). Thus, evaluativeness may undermine meta-accuracy not only through motivational processes on the part of the metaperceiver, but also through greater perceiver-level heterogeneity in impressions.

Aligning with both accounts, in the expanded model, we found that people displayed greater meta-normativity and meta-positivity on more evaluative traits. In other words, while people were less able to track their unique standing on these more evaluative traits, they tended to see themselves as being viewed in ways that were more typical and more favorable overall.

Overall, then, good traits of metaperceptions appear to be those that are more observable and less socially evaluative. This finding carries important interpretive implications. In particular, variation in item observability and evaluativeness may help explain why some studies report weaker or null effects. For example, studies relying more heavily on less observable or highly evaluative traits may be more susceptible to restricted variance or floor effects, which can limit the detection of individual differences and their predictors. Considering item-level properties alongside broader, representative item sets may therefore help clarify inconsistencies in prior findings.

Limitations and Future Directions

The present research is not without its limitations, warranting careful consideration and future investigation. For example, the cross-sectional design of our studies limit the ability to establish causal relationships between associations. While we have attempted to map out the predictors of the good metaperceiver, the good perceiver, and the good trait, the directionality of these relationships remains speculative. For example, it is unclear whether being interpersonally appealing enhanced people’s meta-accuracy in conversation or whether those who were more meta-accurate were rated as more interpersonally appealing by new acquaintances (e.g., more likable and more physically attractive) due to their ability to accurately read the social dynamics that unfolded during the interaction, or both. Future research employing experimental designs could shed light on the causality underlying these associations.

Importantly, our samples were predominantly women who were recruited from a university community. Incorporating more diverse samples across various demographic and cultural contexts would enhance the generalizability of findings. Furthermore, while our study focused on first impressions in a lower-stakes context, the extent to which our findings generalize to other contexts remains an important question. Specifically, higher-stakes first impressions, such as those encountered in dating or job interviews, may involve different dynamics and considerations. Moreover, it is unclear whether the results would generalize to interactions with close others (e.g., romantic partners). Indeed, past research finds that meta-accuracy processes may differ across levels of acquaintanceship (Elsaadawy & Carlson, 2022). Thus, future research could explore whether similar patterns emerge in interactions among acquainted dyads.

Another related limitation concerns the nature of the getting-acquainted context itself. Such contexts may impose relatively strong normative pressures on behavior (e.g., to be polite, engaged, and socially appropriate), which can constrain behavioral variability. To the extent that behavior is more homogeneous across individuals, this may attenuate the strength of associations and lower the upper bound for detecting meta-accuracy correlates. In weaker situations, where behavioral expression is less constrained (e.g., casual interactions among friends), individual differences may be more pronounced, potentially leading to stronger associations with meta-accuracy. Future research should examine how the strength of the situation moderates these associations across different interaction contexts.

The present research also focused on key moderators assessed via self-report and therefore reflect participants’ subjective self-views rather than behavioral or observer-based indicators of personality. As such, this approach limits inferences about personality per se. Future research should incorporate behavioral indicators, informant reports, or experience sampling methods (ESM) to examine whether similar patterns emerge when personality is indexed using multimethod and repeated-measures approaches.

Moreover, our study primarily examined meta-accuracy in the domain of personality impressions, reflecting the predominant focus of the meta-accuracy literature. However, meta-accuracy extends beyond personality traits to encompass various other domains, such as attitudes (e.g., Boothby et al., 2018; Mastroianni et al., 2021) and emotions (Tissera et al., 2023; Tissera & Lydon, 2022). Future research could investigate whether the patterns observed in our study extend to these other domains.

While our study mapped out the three main categories of moderators, an intriguing avenue of investigation lies in the exploration of the good information moderator. While we anticipate both quantity and quality of information to positively relate to meta-accuracy, we maintained consistency in these aspects by focusing on standardized, brief interactions typical of first impressions. Future research could delve deeper into how variations in information quantity and quality impact meta-accuracy across different contexts and interaction durations, providing valuable insights into the dynamics of accurate metaperceptions.

Finally, although our study focused on profile-based meta-accuracy, it is worth noting that different analytic approaches could yield a different pattern of predictors of meta-accuracy. Some researchers have argued that trait-wise and profile-wise approaches reflect separable abilities, with the former indexing the accuracy of specific, trait-level inferences and the latter capturing a broader awareness of patterns across traits (Hall et al., 2018). However, others have shown these approaches to be mathematically and empirically comparable (Biesanz, 2010, 2021), suggesting that they may largely reflect the same underlying ability. In the present research, we adopted a profile-based approach because it offers a holistic assessment of how people are perceived and allows for the simultaneous modeling of perceiver, metaperceiver, and trait-level moderators within a unified framework. Nonetheless, future work could directly compare trait- and profile-based approaches to examine whether they yield unique predictors or offer complementary insights.

While the present studies considered multiple sources of meta-accuracy, such as perceiver effects, meta-normativity, and meta-positivity, we did not directly assess all possible sources of information. For instance, meta-accuracy may also emerge from self-observation, self-disclosures or explicit feedback from others. Future research could systematically isolate these mechanisms to test how strongly each source influences meta-accuracy. For example, meta-perceiver self-disclosure could be coded for, or increased by instructing participants to make statements about how they typically behave (such as saying “I’m usually shy”), thereby assessing or amplifying one source to determine its implications for meta-accuracy. Another approach would be to code perceiver behavior for cues that might serve as real-time feedback (e.g., affirmations, agreement, facial expressions), allowing researchers to quantify the role of perceiver feedback in shaping metaperceptions. These approaches would clarify the unique contributions of each source and shed light on the individual characteristics associated with relying on each type of information.

Conclusion

In the research, we provide insight on who and what might contribute to personality meta-accuracy. In a nutshell, our findings revealed that people tend to be more accurate about how their personality comes across when they see themselves positively or are liked by others, when they are interacting with more physically attractive interaction partners, and when they make judgments on more observable and less socially evaluative aspects. Overall, the present findings offer valuable insights into the multifaceted nature of meta-accuracy, shedding light on the moderators of the good metaperceiver, the good perceiver, and the good trait. This comprehensive approach helps map out the nomological network of meta-accuracy and lays a solid foundation for future explorations in this domain.

Supplemental Material

Supplemental Material - Understanding wWhat Predicts Meta-Accuracy: Exploring the Good Metaperceiver, the Good Perceiver, and the Good Trait

Supplemental Material for Understanding wWhat Predicts Meta-Accuracy: Exploring the Good Metaperceiver, the Good Perceiver, and the Good Trait by Hasagani Tissera, Norhan Elsaadawy, & Lauren J. Human in Personality Science