14 Accuracy of Personality Trait Judgments Based on Environmental and Social Media Cues

Research investigating personality judgments in face-to-face contexts has revealed that people seem to rely on behavioral cues such as tone of voice (McAleer, Todorov, & Belin, 2014), content/fluidity of speech (Isbister & Nass, 2000), use of compliments (Mairesse, Walker, Mehl, & Moore, 2007), attire (Burroughs, Drews, & Hallman, 1991; Gosling & Standen, 1998), and gait (Thoresen, Vuong & Atkinson, 2012), as well as interaction style (e.g., takes charge, smiles, uses hands while talking or moans; Wall, Taylor, Campbell, Heim, & Richardson, 2018) to make judgments about others. In contrast, much can also be learned about individuals from their physical and virtual worlds, as research suggests such contexts are rich with information about their lifestyles, values, and personalities (Gosling, Augustine, Vazire, Holtzman, & Gaddis, 2011; Gosling, Ko, Mannarelli, & Morris, 2002). Numerous studies have shown that people use the information cues available in everyday physical contexts to form impressions of individuals (Gosling, Gaddis, & Vazire, 2007; Gosling et al., 2002). The utility of exploring cue availability, cue use, and accuracy within physical and virtual contexts is that this work provides us with valuable insight into the ways various aspects of a context might affect trait accuracy. Specifically, it allows us to consider what aspects of a context affect the cues that are available to judges. For example, if people are better at judging personality when interacting over e-mail versus face-to-face, it suggests that language cues alone might be useful cues to another’s personality. It also permits an exploration of the types of cues that may be relevant when judging another’s personality. We propose that trait accuracy is best understood when judgment accuracy is explored both within and across judgment contexts. Before outlining the evidence on judgment accuracy within and across contexts, it is imperative to define what is meant by personality, context, and accuracy.

When conceptualizing personality, the five-factor model (FFM; also known as the Big Five) has gained popularity (John & Srivastava, 1999; Pervin, 1994). The Big Five traits—extraversion, agreeableness, conscientiousness, neuroticism, and openness to experience—emerged from decades of research, and the FFM model has been celebrated for its ability to simplify an otherwise overwhelming number of traits (Hofstee, 1994; John, 1990; McCrae & Costa, 1987). Big Five traits have demonstrated good predictive validity in their ability to predict health-relevant and other outcomes (e.g. Emmons, 1995) and show cross-cultural applicability (McCrae & Costa, 1997). In terms of the traits themselves and their meaning, extraversion is associated with sociability, enthusiasm, talkativeness, and assertiveness (McCrae & Costa, 1992). Individuals high in agreeableness tend to be described as interpersonally oriented, warm, trusting, and compliant. Individuals high in conscientiousness tend to be described as planful, responsible, and able to delay gratification. Individuals high in neuroticism¹ are often described as experiencing psychological distress in the form of anger, depression, anxiety and other negative affect (Costa & McCrae, 1990). Finally, individuals who are high in openness to experience tend to be curious, imaginative, unconventional, and to have wide interests. The majority of the studies outlined in this chapter will refer to the Big Five traits as they represent the common framework in the field and have been well researched from an accuracy perspective. Studies investigating judgment accuracy of other traits will also be included to allow for a full review of the field.

Context is often broadly defined as the environmental setting in which a behavior occurs (Bate, 2012; Cappelli & Sherer, 1991). For example, in this chapter we are investigating the personality cues that influence personality judgments and the contexts that we are exploring are the virtual contexts and physical contexts in which the judgments are made. The physical context that provides cues for a personality judgment could be interpreted as relating strictly to the physical objects in very specific personal spaces, for example, the furniture in a bedroom and how it is arranged, the number of books in an office and how they are stacked, and so on. Alternatively, it may be interpreted more broadly to encompass cues outside specific personal spaces an individual occupies such as the geographical location in which the individual resides. For example, research has indicated that geographical personality variations may be linked to factors such as selective migration and social influences (Rentfrow, 2014; Rentfrow, Gosling, & Potter, 2008). Zimmermann and Neyer (2014) have proposed that patterns of geographic personality variations, such as high aggregate levels of openness to experience, may be manifest in the physical environment through accumulated behavioral tendencies associated with this trait. For instance, expressions of this trait may include participation in cultural activities, formal education, and interest in other cultures, which may be evident in the physical environment through the establishment of theaters, universities, and international restaurants. For the purpose of this chapter, we have defined and operationalized the physical context in line with the former, narrower definition as this permits a focus on the specific environmental settings in which a behavior occurs. The majority of research on judgment accuracy also fits within this narrower category.

Virtual contexts are operationalized as online contexts that the individual operates within. As with the physical context, this may also be defined narrowly to include the social networks, blogs, or websites an individual contributes to or on which they have a profile. However, it may also be defined more broadly to include the social networks, blogs, and websites that an individual consumes, or possibly even search engines or operating systems that they use. For example, it is possible that an individual’s Internet search history may be just as revealing about personality as the number of Facebook friends they have acquired. Both physical and virtual contexts range from the personal (e.g., bedroom or private e-mail) to the public (e.g., offices or Twitter posts). Again, in the context of the present chapter, we focus on the narrow definition of this context.

The empirical investigation of context has played a key role in the development of all subfields in psychology. For example, within cognitive psychology, context effects are fundamental to our understanding of the encoding and retrieval of information in memory (Tulving & Thomson, 1973). In social psychology, context is central to many of the landmark findings ranging from social influence (Asch, 1951) to social group membership (Tajfel & Turner, 1979). Similarly, in personality psychology context is also central to understanding trait accuracy. Specifically, the information contained within e-mails has been shown to allow comparatively accurate judgments for the trait of openness to experience (Markey & Wells, 2002; Vazire & Gosling, 2004), even though judgments for openness to experience tend to be difficult to make accurately when based on face-to-face interactions (Borkenau & Liebler, 1992). Moreover, extraversion is typically the most accurately judged trait in face-to-face contexts (Funder & Dobroth, 1987; Lippa & Dietz, 2000), whereas examinations of judgments based on physical spaces (Gosling, Ko, Mannarelli, & Morris, 2002), e-mail content (Back, Schmukle, & Egloff, 2008), and stream of consciousness writing (Li & Chignell, 2010; Yee, Harris, Jabon, & Bailenson, 2011) tend to find that judgments of conscientiousness are more accurate than judgments of extraversion. Taken together, this differential pattern of trait accuracy highlights the importance of context on accuracy and suggests that much can be learned about the cues used and the types of cognitive processing employed in making personality judgments, and also to consider the implications of each of these for trait accuracy.

The accuracy of personality judgments can be measured in numerous ways. Some studies have used the “realistic accuracy approach,” which relies on a reliable and valid measure of a target’s personality, which may then be compared to the ratings provided by one or more judges to examine how accurate or correct personality judgments are. Here, researchers tend to define an accurate judgment as a rating made about a target’s personality that corresponds with the target’s actual personality (Funder, 1987, 1999; Letzring, Wells, & Funder, 2006; Wall, Taylor, & Campbell, 2016; Wall, Taylor, Dixon, Conchie, & Ellis, 2013). The realistic accuracy approach posits that multiple methods of assessment should be used when measuring personality judgments about a person, such as self-reports and informant ratings (Funder, 1999). A common approach to determining judgment accuracy is through self-other agreement. The self-rating is used as the criterion and the judge is said to be accurate if their rating corresponds with the self-rating. A second approach to measuring the validity of personality judgments is consensus, or the agreement about a target person being judged by two or more judges. This has often been used as a proxy for accuracy, as the benefit of combining several judgments is that errors are likely to cancel each other out (Colvin & Bundick, 2001; Hofstee, 1994). This measure is not without error either, as consensus may not necessarily reflect what the target is actually like, but instead reflect agreement between people about what the target person is like, and people can agree but still be inaccurate. The defining feature of realistic accuracy is the attempt to approach a description of what the target is really like by creating a criterion based on multiple sources of information that includes both self-other agreement, judgment consensus, and ideally behavioral prediction (Webb, Campbell, Schwartz, & Sechrest, 1966). Put another way, if a judge’s assessment of a target person’s personality matches the realistic criterion (i.e., multiple assessments of personality possibly based on self- and informant ratings and direct behavioral observations), then realistic accuracy is obtained. For the purposes of a comprehensive review, studies employing all of these indices of accuracy are reviewed in this chapter.

From Instagram sites, websites, and online social networks (OSNs) to offices, bedrooms, and cars—the range of contexts by which to judge another’s personality is intriguing. Where should we look when we want to get to know a person? Are some traits judged differently depending on the particular context in which we view, or meet, people? As noted by Gosling et al. (2002), people’s judgments are shaped by a number of contextual factors, and he argues it is time to begin finding out exactly how our judgments depend on context.

A number of studies have compared trait accuracy across contexts and revealed a number of interesting and intriguing findings. For example, a study by Wall et al. (2013) compared trait judgments across three increasingly information-rich interaction contexts (that is, contexts that provide an increasing variety of cues on which to base a personality judgment). These contexts were an e-mail communication, a telephone call, and a face-to-face interaction. While there are many personality cues available in an e-mail context (e.g., message content, emojis, spelling errors), there are arguably more in a telephone call as there are additional paralinguistic cues such as tone and pitch of the speaker. Similarly, in a face-to-face interaction, the judge has access to the message and paralinguistic cues, as well as cues related to appearance and nonverbal cues such as facial expression and gestures. The results of this study indicated that the less interpersonal traits² of openness to experience and conscientiousness were rated more accurately in face-to-face interactions than in the information-lean context of an e-mail communication. Such findings indicate that not only are we able to reveal aspects of these traits online, but also that others are able to detect cues relevant to these types of traits and correctly use them. Judgments of extraversion were associated with a linear increase in accuracy as the richness of the context increased; conversely, judgments of conscientiousness increased in accuracy when the context became less rich. These findings suggest that judgment accuracy depends on both the type of trait being judged and the context in which a judgment is made. More specifically, the authors concluded that information-rich contexts are not always preferential when judging personality and that cues relevant to judging traits such as conscientiousness and openness to experience may become less salient in information rich contexts given the breadth of cues that are available (e.g., language cues, tone of voice, nonverbal cues). Put another way, the relevant cues to openness to experience and conscientiousness may be easier to detect via an e-mail relative to face-to-face, as judges are not bombarded by the range of irrelevant cues available in rich contexts (see chapter 8 by Krzyzaniak & Letzring in this handbook for more information on how accuracy differs across traits).

Further support for the differential effect of context on accuracy comes from a study by Holleran and Mehl (2008). Although their focus was to examine accuracy after reading stream of consciousness essays, they summarized findings across a range of studies at a descriptive level. Specifically, they compared studies that had examined accuracy based on a range of contexts (e.g., bedrooms, offices, websites, and stream of consciousness writing). They reported that trait accuracy across different contexts can help to illuminate how personality might be revealed by behavior and judged by others. For example, they theorized that an important dimension by which contexts might vary is in terms of whether they are a public (e.g., face-to-face interactions, handshakes, personal websites) or a private context (e.g., stream of consciousness essay). This study found that in private contexts, such as a person’s stream of consciousness writing, the level of accuracy was substantial and relatively uniform across all Big Five traits. By comparing the findings of this study to other published work investigating public contexts (i.e., Gosling et al., 2002; Mehl, Gosling & Pennebaker, 2006; Vazire & Gosling, 2004), the authors concluded that there were higher levels³ of accuracy for judgments of agreeableness, conscientiousness, and neuroticism and comparable levels of accuracy for openness to experience in private contexts compared to public contexts. Again, such findings advance our understanding of which traits tend to be the most accurately judged and in which contexts, in addition to developing our understanding about why this might be the case.

Although it may seem more intuitive to explore how trait accuracy changes across contexts, we argue that it is also fruitful to consider judgment accuracy within contexts, that is, across traits within the same context. Specifically, comparing judgments across traits in one context permits an exploration of the amount and types of context specific cues that may be relevant to certain traits. For example, instead of focusing on whether extraversion is judged more or less accurately via e-mail relative to face-to-face, one can make trait comparisons within a context and consider whether extraversion was judged more accurately than openness to experience. This notion of context influencing the cues to which judges attend has been supported by Ames and Bianchi (2008), who studied personality judgments made by supervisors and supervisees. They found that when participants were able to choose which traits to judge, the traits that they judged appeared to relate to the characteristics that were important for themselves. Specifically, they found that supervisors judged how hardworking their supervisees would be, whereas the student judged how conscientious and reliable the supervisor was, suggesting that perhaps the aspects of personality that are judged are differentially salient depending on the interpersonal needs of each person making the judgment within that specific context. These findings highlight how a focus on trait differences within contexts can increase our understanding of the cues that judges might focus on and the cues that are available within such contexts, and can tell us more about the way in which context is acting on or interacting with the judgment demands.

Theoretical models have been proposed to try to understand how people make judgments. The Brunswikian lens model (1956) is a general model of the judgment process—not just personality judgment—and is a useful tool for determining the relationship between cues that are available, either through behavior or in the environmental context, and judgment accuracy (see Gosling et al, 2002; Naumann, Vazire, Rentfrow, & Gosling, 2009; Vazire & Gosling, 2004; Vazire, Naumann, Rentfrow, & Gosling, 2008; Wall et al., 2018; and chapter 4 by Osterholz, Breil, Nestler, & Back in this handbook). According to Brunswik, environmental cues can serve as a lens through which people indirectly perceive underlying constructs. As noted by Gosling et al. (2002), these cues can be embedded in the environmental context and can be an artifact of the behavior itself. The lens model, as it applies to the process of personality judgment, begins with the occurrence of a behavior that provides cues to personality, and describes cues relevant for making a judgment of each trait as “valid cues.” For example, neatness of appearance could be perceived as a lens through which to infer that a person is high in the trait of conscientiousness.

In Brunswik’s model, valid cues may be assessed through the coding of each behavior or environmental cue and correlating these aspects with actual personality. As the observer may not necessarily detect (i.e., notice) the relevant cues to each trait, Brunswik’s model is also concerned with “cue utility.” The cues used are established by correlating the cues in behavior or the environmental context with judgments of the trait of interest. Importantly, the cues used may or may not be valid cues for each trait and the final stage of Brunswik’s model addresses the overlap between the valid cues and cues used.

Funder’s RAM model (see chapter 2 by Letzring & Funder in this handbook) has applied the lens model to personality judgments (see Funder, 1999). The RAM (Funder, 1995, 1999, 2012) posits that judgment accuracy is a result of a four-stage process whereby cues (i.e., behaviors) need to first be relevant to the trait being judged. Second, the cues to personality need to be available so that the judge can then detect (i.e., notice) and correctly utilize these cues in order to render an accurate judgment. Gosling et al. (2002) note that “overlap” in Brunswik’s model between “cues used” and “valid cues” should lead to accurate personality judgments according to the RAM, as valid cues have been successfully detected and utilized. For example, if a valid cue to extraversion is talkativeness and a judge notices this cue and then correctly infers that this cue is indicative of the trait of extraversion, this overlap between valid cues and cues used should lead to accurate judgment of this trait. In the context of judgment accuracy based on physical and virtual cues, this poses the question of whether there will be relevant cues of the target’s personality available in physical and virtual contexts; and, if so, whether judges can detect and correctly utilize these cues to render an accurate judgment.

Research to date suggests that as individuals craft and shape their physical personal spaces, they leave traces of their personality, both intentionally and unintentionally. Gosling et al. (2002) applied the lens model and the RAM to judgments of personal spaces and proposed a model of interpersonal perception. Specifically, they proposed mechanisms by which individuals affect their personal environments and leave personality cues in their personal spaces and the processes observers employ to make personality judgments. The two mechanisms outlined by Gosling et al. (2002) by which individuals may affect their personal spaces are through identity claims and behavioral residue.

Identity claims are symbolic statements made by individuals with the aim of shaping how they are perceived by self and others and can be either self-directed or other-directed. Self-directed identity claims are directed toward yourself and serve to reinforce or remind one of important aspects of the self. For example, hanging certificates of professional achievement in one’s office may communicate pride in the accomplishment or educational and professional values. In the context of virtual spaces such as personal websites and OSNs, this may include posting photos with friends that celebrate valued friendships. Other-directed identity claims can be aspirational, strategic, or deceptive attempts to communicate how the individual would like to be perceived by others. For example, an individual may tidy away clutter when expecting visitors in an attempt to appear more conscientious. As Gosling et al. (2002) noted, self-directed and other-directed identity claims may often overlap or be presented in similar ways. For example, if the placement of the books was such so that the individual could look at them and have their view of themselves as an intellectual reinforced, that would be a self-directed identity claim. If the books are placed so that others can also see them in an attempt to create the impression in others that the individual is a professional or an expert, that is an other-directed identity claim.

Behavioral residue differs from identity claims as it refers to personality cues in one’s personal space that are the unintentional remnants of an individual’s behavior. For example, behavioral residue may include dirty dishes left beside a sink reflecting a low frequency of dishwashing, a juicer on the counter may suggest that the individual is conscientious about healthy eating, while a swimming suit on the clothesline may indicate that the individual is physically active. Behavioral residue can offer an insight into an individual’s behaviors conducted within their personal space (interior behavioral residue) or activities performed outside that space (exterior behavioral residue) (see Gosling et al., 2002, for more detail). Behavioral residue can also be found in OSNs, for example, through their use of social media sites individuals may collect a large number of friends on the site, which has been used as a cue when judging extraversion (Gosling et al., 2011).

In Gosling et al.’s (2002) substantive study, they found that a person’s behavioral residue and identity claims affect their physical spaces such as bedrooms and offices; however, subsequent research has also applied this model to more diverse contexts (e.g., Mehl et al., 2006; Rentfrow & Gosling, 2006) including virtual contexts (Vazire & Gosling, 2004). For this reason, we have referred to personal spaces in our description to include both an individual’s physical and virtual contexts. Gosling’s model of interpersonal perception (Gosling et al., 2002) proposed that physical and virtual cues may affect observer judgments through (1) observer’s inferences about the behaviors that may have created the physical evidence, and (2) inferences about the dispositions that may have motivated those behaviors. For example, a long list of Facebook friends may lead an observer to infer that the target is popular online, this in turn may lead the observer to believe that the target is high in extraversion. A study by Back et al. (2010) found that accuracy for extraversion was strongest relative to the remaining Big Five traits. Gosling et al. (2002) also pointed out the possibility that cues in physical and virtual contexts may activate stereotypes, either at the point of making inferences about the behaviors that lead to the presence of these cues, or at the point at which the observers inferred the motivations for the behaviors. For example, objects or decoration in a room that suggests a female occupant may trigger gendered stereotypes that may influence the observers’ trait accuracy. Stereotypes may lead the observers to reach decisions that are not based on evidence in the target’s personal space. If observers use invalid stereotypes to guide their decision-making, it may lead them to form inaccurate impressions of the occupant, however, as we know, there is some merit in some stereotypes (Lee, Jussim, & McCauley, 1995; see chapter 16 by Lee, Stevens, & Honeycutt in this handbook), in which case this may lead the observers to make more accurate judgments.

As aforementioned, research by Gosling et al. (2002) has applied the lens model and the RAM to judgments of personal spaces and determined which behaviors or aspects of an environmental context were valid cues of each personality trait as outlined in the next section. Their study proposed that personal spaces may contain an abundance of potentially informative cues about an individual and therefore provide rich information to observers for personality judgments. In their paper they explored trait accuracy based solely on examination of an individual’s office or bedroom.

In Gosling et al.’s (2002) first study that explored office contexts, judgments of openness to experience showed the strongest consensus, followed by conscientiousness and extraversion. Agreeableness also showed some consensus, while the least consensus was for neuroticism. The pattern of consensus correlations was different from that found in previous zero-acquaintance research,⁴ suggesting that the cues available from photographs or short interactions differ from those available from workspaces. Whereas zero-acquaintance research using the Big Five has found the strongest consensus for extraversion and conscientiousness (e.g., Kenny, 1994; Kenny, Horner, Kashy, & Chu, 1992; Park & Judd, 1989), Gosling’s study found the strongest consensus for openness to experience, although they still found strong consensus for conscientiousness and extraversion. They found that the level of accuracy also varied across traits. Accuracy was highest for openness to experience, followed by extraversion, conscientiousness, and neuroticism. Judgments of agreeableness were not accurate. Again, the pattern of findings differed from the general pattern of findings in zero-acquaintance research, in which extraversion usually yields the highest accuracy, followed by conscientiousness, with little or no accuracy for neuroticism and openness to experience (Kenny, 1994; Watson, 1989). Consensus and accuracy results were found to have been partially affected by the use of sex stereotypes. The cues that were valid and used to make judgments from offices are discussed later in this section.

In a second study, Gosling et al. (2002) explored the role of cues on accuracy in the context of bedrooms. As in the context of offices, openness to experience showed the strongest consensus, followed by conscientiousness and extraversion, while agreeableness and neuroticism showed the least. In terms of accuracy, openness to experience was also the most accurately judged trait and agreeableness the least accurately judged trait. Consensus and accuracy were found to be only partially mediated by sex and race stereotypes. Observers used similar cues to judge extraversion, agreeableness, and conscientiousness across the office and bedroom contexts. For example, a room being inviting was used as a utilized cue for the trait of agreeableness in both office and bedroom contexts, yet this was not a valid cue for this trait in either context. However, Gosling’s findings highlighted that the valid cues to trait accuracy were largely different across contexts and were only similar for the trait of conscientiousness. For example, in comparison to the previous example, many of the cues used to judge the trait of conscientiousness such as a room being in good condition, neat, clean, uncluttered, and organized, were valid cues in both the office and bedroom contexts. This finding highlights the need for research to examine differences across contexts and the ways in which individuals interact with these differences.

In summary, Gosling et al. (2002) employed the Brunswikian lens model to evaluate which cues measured by the researchers were valid cues of each trait and which cues were used by judges when making their personality judgments. This is useful for shedding light on the judgment accuracy for each trait. For example, in the office context valid cues for the trait of openness to experience included how distinctive, stylish, and unconventional the room was. This pattern of accurate detection and utilization of relevant cues has resulted in the comparatively high accuracy for judging this trait in this study. In comparison, in the office context there was only one valid cue measured for the trait of agreeableness, which was whether the office was in a high traffic area. This was not detected by the judges, which can help explain the lack of accurate judgments for this trait. This pattern of the detection and utilization of relevant cues available in the physical context leading to trait accuracy was also found in the bedroom context. For example, while there were only three valid cues for the trait of openness to experience measured in the bedroom context, each of these was utilized by the judges (distinctive room, varied books, varied magazines), which explains the comparatively high level of judgment accuracy for this trait. However, there were no valid cues for neuroticism in the bedroom context and only two for the trait of agreeableness, neither of which were used by judges, which accounts for the poor judgment accuracy for these traits. It is worth noting that a finite number of cues were recorded in this study and it is possible that there were other cues present that were not recorded that informed judges’ decisions. This is supported by the judgment accuracy for the trait of extraversion in the bedroom context despite judges failing to use either of the two valid cues for the trait.

Their study also sheds light on the lay theories that the judges employed when making their judgments, for example, this study illustrated that across both contexts there were many more cues employed to inform personality judgments than there were valid cues. All in all, there were less than 20 valid cues measured that were available in each context and yet participants used more than 50 of the cues measured to make their judgments in each context, suggesting that there is a discrepancy between judges’ perceptions of relevant cues for a trait and valid cues for a trait. This is illustrated when looking at judgments of the trait of conscientiousness in the office context. When judging this trait judges used a number of valid cues to inform their judgments such as the office being in good condition, and whether offices were clean, organized, neat, uncluttered, and had homogeneous books and CDs. However, judges also employed a number of cues that were not relevant to this trait, for example, whether the office was roomy, empty, expensive, comfortable, inviting, large, distinctive, formal, conventional, a good use of space, had matched contents, organized books, homogeneous stationary, and organized stationery. This suggests that the judges’ lay theory of the trait of conscientiousness may be tapping into a broader construct. While the judges were accurate in using cues such as an office being clean, organized, and neat, they seemed to also be associating conscientiousness with something more impressive or possibly luxurious, for example, they employed cues such as an office being expensive, large, roomy, inviting, comfortable, and formal.

It is also interesting that the valid cues to a trait varied across contexts. For example, while being organized, neat, and uncluttered was a valid cue for the trait of conscientiousness in both the office and bedroom contexts, organized books and CDs were valid cues of this trait only in the bedroom context, while homogeneous books and CDs were the valid cues in the office context. Similarly, distinctiveness was a valid cue for the trait of openness to experience across contexts. In the bedroom context this was expressed through cues such varied books and magazines, while in an office context valid cues included the office being perceived as unconventional, stylish, and fresh. While it may be possible to speculate about intuitive explanations about how or why the cues listed here inform judgments of the traits of conscientiousness and openness to experience respectively, it is not wise to do so without first conducting further research. It is unclear why some cues are valid in one context but not another. Again, further research is required to answer this question, particularly as this study is one of a kind.

Although a number of studies have examined first impression accuracy in face-to-face encounters (Bar, Neta, & Linz, 2006; Carney, Colvin, & Hall, 2007; Willis & Todorov, 2006), much less research has examined accuracy in online contexts, which are typically more limited in the number of cues available, particularly regarding nonverbal ones. Knowledge of the specific traits people are more likely to make accurate judgments for and insight into the types of cues that may facilitate or hinder this process is arguably important in a range of contexts such as online dating and job recruitment. When considering the practical means of exploring these issues, the Internet and other virtual settings represents useful contexts in which to explore the expression and perception of personality.

Virtual contexts are an interesting site of research because they offer a context in which to observe an individual engaged in social interactions and also provide an opportunity to access the virtual worlds that they have created for themselves. These virtual spaces are thought to operate in much the same way as physical spaces, displaying the deliberate (e.g., identity claims) and inadvertent (e.g., behavioral residue) cues to an individual’s personality (Vazire & Gosling, 2004).

Numerous aspects of language have been explored in virtual contexts such as the use of text-speak (an informal abbreviated language used in text messaging (Crystal, 2008)), and vocabulary usage, spelling, and pronoun usage (Darbyshire, Kirk, Wall, & Kaye, 2016; Fullwood, Quinn, Chen-Wilson, Chadwick, & Reynolds, 2015; Wall et al., 2013). Writing errors (e.g., grammatical or typography errors) have been found to be related to others’ negative judgments of the sender (Boland & Queen, 2016; Queen & Boland, 2015), specifically judging the sender as less conscientious or attentive (Ellison, Heino, & Gibbs, 2006; Vignovic & Thompson, 2010). However, some of these papers have measured perception rather than judgment accuracy. Among those that have measured the latter, language cues seem to influence judgments of the “less interpersonal” traits such as conscientiousness (Darbyshire et al., 2016; Wall et al., 2013). Although the role of language on perceptions, and to a lesser extent judgment accuracy, have been studied, less research has systematically examined how the quantity of language cues shapes trait perception or accuracy. That is, what happens to perceptions of conscientiousness when different types of errors are present, such as typos and/or grammatical errors? This focus on variations in cue type enables quantity and quality of language cues on perceptions to be investigated. In support of the present focus on variations in cue type, a study by Vignovic and Thompson (2010; see also Scott, Sinclair, Short, & Bruce, 2014) found that people tend to be perceived as less conscientious and trustworthy when their e-mails contained many typos (e.g., “abuot” instead of “about”). However, although the existing literature is relatively consistent on the role of language cues on perceptions of conscientiousness, what is less known is the extent to which amount of usage may be relevant here. This is relevant in respect to the RAM, as the quantity of language cues relates to the availability stage, and as a result, may foster greater capacity to form accurate judgments.

Research has also examined personality expression in virtual contexts via personal websites (e.g., Vazire & Gosling, 2004), e-mail interaction (e.g., Gill, Oberlander, & Austin, 2006), and usernames in online games (e.g., Graham & Gosling, 2012). For example, Back et al. (2008) examined how strangers judged targets’ personality through linguistic features of e-mail addresses, and found that judgments of neuroticism, openness to experience, agreeableness, conscientiousness, and narcissism were reasonably accurate. Neuroticism has been found to be associated with the use of more anxiety words in Facebook profiles (Golbeck, Robles, & Turner, 2011), while agreeable individuals employ more positive emotion words and first-person plural pronouns in blogs (Yarkoni, 2010). People with a higher level of conscientiousness are more likely to discuss others via e-mail communication (Oberlander & Gill, 2006). Taken together, these findings suggest that self-reported personality traits may be differentially associated with specific linguistic patterns in virtual contexts such as microblogs versus nonvirtual contexts such as speech and writing. Importantly, the extent to which judges detect and use these cues is not yet known and is an area for further inquiry.

These text-based symbols provide an online alternative to facial expressions in face-to-face interactions, allowing users to communicate their intended emotions (Derks, Bos, & Grumbkow, 2008). Previous work exploring emoji use on Facebook has found that happy emoji use relates to a range of trait perceptions; specifically agreeableness, conscientiousness, and openness to experience (Wall, Kaye, & Malone, 2016). More recently, studies have examined the role of emojis on perceptions of characteristics such as sincerity and friendliness. The use of smiley emojis in a written conversation between two people has been shown to result in perceptions of the person using smiley emojis as friendly and sincere (Wibowo, Ats-Tsiqoh, Sangadah, Komala, & Utomo, 2017). Although this study only examined the role of emojis on perceptions, not accuracy, their findings highlight the importance of context specific cues on shaping personality judgments and further research is needed to explore how cues are differentially related to perception and accuracy.

Twitter is a social networking microblog service that permits users to broadcast short posts called tweets, which can be up to 140 characters. Members can broadcast tweets and follow other users’ tweets and people can search tweets on Twitter whether or not they are a member. Studies exploring the accuracy of personality judgments based on microblogs on Twitter have revealed that unfamiliar raters can accurately judge neuroticism and agreeableness (Qiu et al., 2012). This is consistent with previous findings that these two dimensions can be judged accurately in self-generated written content (e.g., Holleran & Mehl, 2008).

Webpages are usually maintained by one person or organization and devoted to a single topic or several closely related topics on the World Wide Web. A number of studies have examined trait accuracy based on viewing personal webpages and revealed an interesting pattern of findings. A study by Marcus, Machilek, and Schutz (2006) explored the accuracy of personality impressions after viewing a person’s personal webpage and found valid judgments for all Big Five traits, in particular for the trait of openness to experience. Interestingly, this study also explored the cues available in webpages that were valid indicators of the website owners’ personalities. They found that people high in neuroticism tend to avoid requesting feedback from site visitors, do not reveal their address, and avoid commenting on the external websites linked to their pages. The trait of openness to experience was linked to more use of websites about lyrics and fine arts. Vazire and Gosling (2004) also examined trait accuracy based on personal websites and reported accuracy levels similar to that found in other interpersonal contexts. Specifically, they found that the websites elicited high levels of consensus and accuracy for the two traits of extraversion and agreeableness. They reported accuracy levels higher than those found in zero-acquaintance contexts and suggested that the higher accuracy was due to the identity claims available in this particular context.

Instagram is currently the fastest growing photo-sharing social media platform, with more than 400 million active users, nearly 100 million photos shared on the platform daily, and generation of 1.2 billion likes each day (Lay & Ferwerda, 2018). Ferwerda, Schedl, and Tkalcic (2016) found distinct features within Instagram photos, such as their brightness, hues, and saturation,⁵ were related to personality traits, indicating that users with different personalities make their pictures look different. For instance, openness to experience was positively associated with cold colors such as green and blue, low brightness, high saturation, cold colors, and few faces; individuals high in conscientiousness tended to post images with more variation in color saturation; agreeable individuals were more likely to post images with few dark and bright areas; neuroticism was related to images with high brightness; extraversion was linked with images of green and blue tones, low brightness, and saturated and unsaturated colors (Ferwerda et al., 2016). This finding that features of Instagram pictures are related to personality judgements expands on previous research that has shown consistent links between openness to experience and aesthetic preferences (Ferwerda et al., 2016). The utility of image characteristics has also been found in research by McManus and Furnham (2006), who found image characteristics to be significantly correlated with openness to experience, followed by agreeableness and conscientiousness.

“Selfies” are known as a self-portrait picture taken by oneself, which are then frequently posted on OSN’s (Qui, Lu, Yang, Qu, & Zhu, 2015). Selfies are a distinct subgroup of photographs that can be posted in virtual contexts such as Instagram, Facebook, or other websites. From an accuracy perspective, selfies are a very interesting behavior. When taking a selfie, individuals can view what they look like and decide what they want to show. It has been suggested that they are a new medium for self-expression and self-presentation (Qiu et al., 2015). Selfies contain unique cues that are not typically available in other types of photos such as pouting or “duck face,” which is a lip pouting expression to give the appearance of fuller lips. Some studies have shown that selfies reveal personality-relevant information (Qiu et al., 2015). Selfies would appear interesting to study from a self-presentation perspective because people choose to take selfies. Qiu et al. (2015) were the first researchers to explore how selfies reflect the subject of the photo’s personality and the accuracy with which people make personality judgments based on selfies. They found a number of trait-related cues in selfies, for example, selfies taken in a private location were associated with lower levels of conscientiousness, while emotional positivity was found to predict both openness to experience and agreeableness. Interestingly, although there were relevant cues to personality (as measured using a Brunswikian lens model) and judgment consensus among observers for all traits, only openness to experience was judged accurately. This suggests that observers had difficulty detecting and using relevant cues. Another study found that the selfie owners’ levels of conscientiousness and openness to experience were the easiest traits to judge accurately and noted that when judgments were based on selfies, rather than a full-length photograph taken by the researchers, there were greater perceptions of narcissism but lower perceptions of conscientiousness (Kaurin, Heil, Wessa, Egloff, & Hirschmüller, 2018).

Facebook is a social networking website that allows registered users to upload photos, keep in touch with friends, send messages, and create profiles. The Big Five traits are considered to be expressed differently across a range of “Facebook behaviors” (Amichai-Hamburger & Vinitzky, 2010). For example, traits are expressed in different ways through a number of online behaviors including number of Facebook friends, disclosure of personal information, uploading and modifying photographs, using different spaces on Facebook for uploading information, and the nature of interactions (Amichai-Hamburger & Vinitzky, 2010; Gosling et al., 2011; Hollenbaugh & Ferris, 2014; Moore & McElroy, 2012; Ross et al., 2009). Patterns of trait accuracy in a Facebook context are consistent with other virtual contexts. A number of studies have shown that extraversion tends to be the most accurately judged personality trait when viewing others’ Facebook profiles, and neuroticism the least accurate (Back et al., 2010; Gosling, Gaddis, & Vazire, 2007). This was corroborated in research using personal websites (Vazire & Gosling, 2004) and e-mail correspondence (Gill, Oberlander, & Austin, 2006). Taken together, such trait-specific findings suggest that nuanced context-behavior relations exist.

Interestingly, from a context perspective, these results contrast with those obtained in Qiu, Lin, Ramsay, and Yang (2012) which showed that observers could accurately detect neuroticism and agreeableness from people’s Twitter profiles. This suggests that different social media platforms may afford the exhibition of different personality traits. One possible explanation of the observed personality perception results is that microblogging affects the expression of extraversion, openness to experience, and conscientiousness. For example, Twitter encourages people to disclose their inner feelings and share their social activities with others, meaning that all users will appear extraverted to some extent. Similarly, most people tend to tweet about their new experiences or discoveries, giving others the impression that they are open to new experiences.

There are also substantial differences in the cues available through Facebook interactions and those in face-to-face interactions. Crucially, certain behaviors such as a person’s facial expressions are not available, or are only available as posed photographs, within online platforms (Amichai-Hamburger & Vinitzky, 2010), however, there are other personality cues that are specific to this context, such as number of online friends and status updates (Gosling, Augustine, Vazire, Holtzman, & Gaddis, 2011; Jarvis, 2010). Given the differences in certain types of behaviors able to be revealed online versus offline and the contextual differences in trait-specific accuracy that have begun to emerge, it is important to continue to examine how such judgments are being formed. Although some research has begun to code aspects of behavior in this regard (e.g., Bachrach, Kosinski, Graepel, Kohli, & Stillwell, 2012; Ross et al., 2009), it is also useful to view such coding from the perspective of the target, as Gosling et al. (2011) have done. In this study, Gosling and colleagues found that extraversion was correlated with a large number of self-reported Facebook behaviors, especially those related to maintaining an up-to-date presence and tending to social bonds (e.g., number of Facebook friends and commenting on others’ pages). Agreeableness, the other interpersonal Big Five dimension, was also related to OSN usage; those higher in agreeableness viewed all pages (i.e., any, others, and their own pages) more often than those low in agreeableness. Consistent with the idea that OSNs serve as an opportunity for those low in conscientiousness to procrastinate, participants low on conscientiousness spent more time viewing pages and more time on Facebook than did those high in conscientiousness. Finally, openness to experience was related to adding and replacing photographs, which may reflect the fact that individuals high on this trait tend to engage in a wide range of activities. Neuroticism was not related to any of the self-reported Facebook behaviors (Gosling et al., 2011).

Studies adopting a quantitative approach to this issue have tended to correlate behavioral and/or linguistic cues with scores on trait judgments, with the assumption that significant correlations indicate that judges did actually use these cues. A complementary qualitative approach can examine the cues that people explicitly report they used when judging others. Darbyshire et al. (2016) have explored individuals’ conscious perceptions of the cues used to make personality judgments via the platform of Facebook using thematic analysis of participants’ self-reported accounts of how they made their judgments. This study revealed the following cues: vocabulary (an indication for openness to experience) and occupational status (an indication for conscientiousness). Both of these themes contained all language-based predictors for the traits openness to experience and conscientiousness. For example, subthemes of the vocabulary theme included spelling and use of grammar, which are predictors of intelligence, and intelligence is a facet of openness to experience (Costa & McCrae, 1992). It is suggested that an accurate judgment of openness to experience was formed due to the way the judge detected these cues and then used them. It is important that future research replicate this finding to substantiate this claim. Similarly, subthemes of the occupational status theme included organization levels, which was an indicator of conscientiousness (Costa & McCrae, 1992). Recent evidence has shown that Facebook appears to be the least likely place individuals will express neurotic behaviors when comparing Big Five traits (Back et al., 2010). In line with the previous distinction between more and less interpersonal traits, online contexts appear to differentially influence the expression and perception of personality with contexts such as Facebook, and facilitate greater accuracy for less interpersonal traits. Interestingly, the trait of neuroticism has been shown to be poorly judged in many other contexts, including on Facebook (Back et al., 2010; Gosling et al., 2011; Gosling et al., 2007). Such findings attest to the importance of focusing on the role of environmental contexts on personality judgment accuracy, as it can tell us, descriptively, which traits people are more likely to judge accurately, and can begin to illuminate which cues appear to be relevant and even provide a glimpse into how users are interacting with their environmental contexts and the way in which we present ourselves to the world.

The literature described in this chapter has shed light on the importance of physical and virtual contexts in personality judgments. This pioneering research has shown that by exploring the personal space an individual occupies, whether physical or virtual, it is possible to make relatively accurate personality judgments about individuals without ever having met them. To date, this research has fallen within a narrow definition of what a physical or virtual context entails, however, there is scope to explore a broader conceptualization of these contexts. While acknowledging the need to further explore and replicate many of the findings we have discussed in this chapter, we also wish to highlight areas where there is potential to expand the current definitions of physical and virtual contexts.

To date there have been a handful of studies that have systematically examined how different contexts contribute to judgment accuracy. There is a need for these studies to be replicated and expanded on. As mentioned previously, there is variation in the cues that are valid indicators of personality traits across contexts. For example, for the trait of conscientiousness, being organized and uncluttered served as valid cues across contexts, while other cues such as a room being clean and in good condition were context specific cues to this trait. It is unclear why this is, and further research on this topic is necessary to develop a better understanding of the expression and perception of personality in and across contexts.

There is also a need to expand the range of physical contexts that have been explored. A possible context for future research includes an individual’s vehicle. There is no research to date that has explored the personality cues that an individual’s car may provide about the owner’s personality, however, Alpers and Gerdes (2006) have demonstrated that judges were able to match cars to their owners with above chance level accuracy. This suggests that cars are providing the judges with some salient cues about their owners. Further research is required in this context to determine whether these cues are salient to trait accuracy. However, this seems to be a promising context for research as it is possible to see how behavioral residue could accumulate in a car and also how a car may be used as a means to convey identity claims.

Another aspect of both physical and virtual contexts that warrants further explorations are cues to social group memberships and other similarities that are shared with the judges. For example, an individual’s personal spaces may include many cues to their group memberships such as flags or banners supporting a particular sports team, musical instruments specific to a particular culture, or calligraphy scripts in a specific language. Indeed Gosling et al. (2002) found that cues to age and sex in office and bedroom contexts led to the activation of stereotypes about these groups, which in turn influenced personality judgments. Intragroup and intergroup relationships have also been demonstrated to shape perception in numerous social psychological studies, yet very little research has been conducted to explore how group dynamics influence personality judgments. Letzring (2010) explored the significance of shared gender and ethnicity in trait accuracy. The author hypothesized that trait accuracy would be highest when the judge and target were most similar. The results of this study indicated that female judges were more accurate in rating targets that shared their gender and ethnicity than targets that did not. This effect did not hold for men, who were most accurate when judging female targets. A second study in this area has been conducted by Rogers and Biesanz (2014). This study explored the effect of group membership on interpersonal perception, including trait accuracy. They found that perceptions of in-group members showed higher levels of distinctive accuracy than did perceptions of out-group members. Distinctive accuracy is used to refer to the important aspect of accurately judging how a target person is different from the average person. More specifically, it entails understanding how a given target differs from the average person and others on specific traits. However, the results also indicated that judges also rated ingroup members as higher in distinctive assumed similarity (i.e., a tendency for observers to project their own distinctive characteristics onto others). Overall, this study suggests that perceptions of individuals with shared cultural group membership may be more biased as well as more accurate. These studies highlight that intragroup and intergroup dynamics affect personality perceptions, and as group identification is a pervasive aspect of daily life (Tajfel & Turner, 1979), these dynamics should be considered in personality judgments.

Similarly, there is much research on language use in nonvirtual contexts that warrants further investigation in virtual contexts, for example, the connection between self-reported personality and writing style (Pennebaker & King, 1999). A software program called Linguistic Inquiry and Word Count (LIWC) has been widely used to identify linguistic patterns associated with personality traits by calculating word frequencies in psychologically meaningful categories, such as pronouns, social terms, and affect (or emotion) terms (Pennebaker, Booth, & Francis, 2007). Extraverts have been found to produce fewer large words (Mehl et al., 2006), less complex writings, and more social and positive emotional words than introverts (Pennebaker & King, 1999). Küfner, Back, Nestler, and Egloff (2010) found that raters could judge openness to experience and agreeableness via linguistic cues in creative writing samples, while Mehl et al. (2006) found that raters successfully used the presence of swear words and negative emotion words in everyday speech to judge agreeableness. Holleran and Mehl (2008) found that individuals could accurately judge the Big Five personality traits of unknown others by reading stream of consciousness essays. Future research should explore whether these findings are applicable in a virtual context and whether trait accuracy varies across online platform.

As mentioned earlier in this chapter, the definitions of physical and virtual contexts could be broadened to go beyond an individual’s own personal space and to include much larger areas that may also provide cues to an individual’s personality. In line with research by Rentfrow et al. (2008; see also Zimmermann & Neyer, 2014), which has indicated that there may be personality variation across geographic locations, there is scope for future research to consider how these personality variations shape the local environments, which may in turn influence personality judgments.

Other questions that should be addressed include whether there will be an interaction between the personality cues an individual provides and the cues in the judgment context. That is, will an individual’s personality be judged more accurately in contexts where the individual and the context are consistent? For example, it is possible that individuals high in the trait of openness to experience may be judged more accurately in geographic locations that have higher aggregate levels of openness to experience and therefore more infrastructure (e.g., theaters, universities, international restaurants) that would allow the public expression of these traits. Or more importantly, if a person is available to provide personality cues will the judge still attend to environmental cues available in the context? While it may seem intuitive that more cues, whether from an individual or the context, will lead to more accurate judgments, Wall et al. (2013) have illustrated that this is not always the case.

Physical and virtual contexts offer the possibility to explore personality judgments that occur without any direct interaction with the subject of the judgment. This chapter has outlined a range of literature documenting the role of context, including both physical and virtual, on accuracy. We explored the ways that individuals craft and shape their personal spaces and embed personality cues through identity claims and behavioral residue. Research on judgment accuracy in physical and virtual spaces is valuable as it has shown that in many cases trait specific accuracy (e.g., conscientiousness and openness to experience) is greater in these contexts than in face-to-face interactions. The range of contexts that can be examined is broad and ever-growing, and this chapter has made recommendations about areas where the existing research should be replicated and expanded. Overall, this chapter has shown that context has a differential effect on judgment accuracy in terms of the specific trait, or traits, being judged. For the topic of accuracy, a focus on context is key as it concerns understanding more about the way in which humans navigate social situations to make the complex task of judging manageable (Taylor & Fiske, 1978).