Social–Emotional Expertise (SEE) Scale: Development and Initial Validation

Abstract

Social–emotional expertise (SEE) represents a synthesis of specific cognitive abilities related to social interactions, and emphasizes the timing and synchrony of behaviors that contribute to overall social–emotional ability. As a step toward SEE construct validation, we conducted three experiments to develop a self-report measure that captured key elements of our conceptualization of SEE. In Experiment 1, we generated and tested 76 items for a measure of SEE. The resultant 25-item scale is reliable, test–retest: r(80) = .82, p < .001, and internally consistent (Cronbach’s α = .90). Experiments 2 and 3 examined the relationships between the SEE Scale and related constructs. Convergent constructs, such as emotional intelligence, r(885) = .62, p < .01, and social anxiety, r(885) = −.59, p < .01, and discriminant constructs, such as social desirability, r(885) = .19, p < .01, and self-monitoring, r(885) = .28, p < .01, were found to be related in the expected directions. Additionally, two factors were statistically identified: Adaptability and Expressivity. The items contributing to each factor describe the ability to successfully navigate social environments and the ability to successfully convey affect and ideas to other people, respectively. These factors correlate with related constructs in distinct and theoretically relevant ways.

Keywords

social ability social–emotional interactions affect scale development

In daily life, it is clear that the ease with which people interact with others is subject to individual differences: Some people handle social situations fluently, whereas others are less proficient. One approach to understanding such differences has been to use constructs which attempt to characterize the cognitive and affect-related processes that presumably underlie social behavior. “Emotional intelligence” (EI; Salovey & Mayer, 1990), “social intelligence” (SI; Sternberg & Smith, 1985; R. L. Thorndike & Stein, 1937), “empathy” (Davis, 1983; Lawrence, Shaw, Baker, Baron-Cohen, & David, 2004), and “interpersonal sensitivity” (IS; Hall & Bernieri, 2001) are four such constructs. As they are typically described, these constructs focus on the accurate processing of social cues and the use of those cues to inform future behavior. However, these constructs do not focus on the actual behaviors that facilitate interactions and convey socially meaningful cues.

To bridge this gap in understanding between prosocial cognition and behavior, we propose the construct of “Social–Emotional Expertise” (SEE). As conceptualized here, SEE involves the coordination of affect-related gestures and vocalizations, with an emphasis on the quality and temporal dynamics of these signals. SEE is thought to include automaticity in response to social signals (Spunt & Lieberman, 2013), with little in the way of conscious deliberation required for the smooth navigation of social encounters. The responses themselves are collectively thought of as a “toolkit” of affect-related behaviors that taken together, enhance the quality of interactions. To illustrate, consider your reaction to a joke told by an acquaintance. Determining that what your acquaintance just said was a joke, what the joke means, why your acquaintance just made the joke, and what your acquaintance may be feeling having made this joke are comfortably within the purview of constructs such as EI and IS. The particulars of your response, and the timing in response to your acquaintance, is the essence of SEE. Awareness of this acquaintance’s feelings and knowledge of how to best respond to those feelings is important, but the key component of a high-quality interaction is how that understanding is then translated into prosocial behavior. Knowing how and when to laugh at a good joke, how to respond when the joke is bad or offensive, and timing the response so that the quality of the interaction does not suffer unduly would all be SEE skills. Conversely, poor timing and inappropriate volume of your laughter, inappropriate timing or duration of eye contact, or a lack of acknowledgement of the joke altogether, would represent a lack of SEE skills being utilized in the interaction. In these ways, the timing and use of behavior is directly tied to the quality of the interaction taking place.

Associated Constructs

We expect there to be moderate associations among SEE and conceptually related constructs. In particular, SEE has considerable conceptual overlap with some characterizations of both EI and IS. EI is itself an umbrella concept that encompasses many terms (e.g., “people skills”; Goldenberg, Matheson, & Mantler, 2006), and is most consistently described as being composed of perception and understanding of emotions and emotion-related signals in oneself and others (Mayer, Salovey, Caruso, & Sitarenios, 2003). The original EI model consists of four branches: (a) understanding emotions, emotional language, and emotion-related signals; (b) accurate perception of emotions in self and others; (c) management of emotions for goal attainment; and (d) use of emotions to facilitate thinking (Mayer & Salovey, 1997; Mayer, Salovey, & Caruso, 2008). Of the four branches, it is anticipated that the first two might be most closely related to SEE in that variety of emotionally expressive behaviors and perceptive ability affect the quality of interactions; those who are higher in EI in these two domains may also be higher in SEE.

SI was first described by E. L. Thorndike (1920), along with abstract and mechanical intelligence, as one of three subcomponents of overall intelligence. Due to SI’s strong correlation with verbal intelligence, R. L. Thorndike and Stein (1937) declared that the verbally based attempts to generate the construct were missing key aspects of ecological validity to support the idea of a distinct SI. Since that time, the SI construct has been given a variety of definitions, including the application of general intelligence to social situations (Wechsler, 1958), the ability to understand other people (Barnes & Sternberg, 1989), and the ability to have successful interactions with others (Ford & Tisak, 1983). Similar to EI, the fundamental discrepancy among definitions of SI is whether it is defined as a cognitive ability, like general intelligence, or as a personality characteristic or trait. Measures of SI have been found to be indistinguishable from general intelligence on verbal tasks. The two constructs are, however, distinguishable in nonverbal behaviors (Kihlstrom & Cantor, 2000), an empirical outcome that bears on SI’s relation to SEE. This divergence of SI and IQ in nonverbal behavior is indicative of the key role that behavior, separately from verbal intelligence, plays in the quality of social interactions. Since the early 1990’s, SI has been seen as the overarching construct under which other constructs such as EI are subcomponents (Mayer & Salovey, 1993). From this, we tentatively consider SEE to be related to SI in a similar way to how SEE is hypothesized to be related to EI, with SEE being most closely associated with the nonverbal behavioral component of SI which has been shown to be most distinct from abstract intelligence.

Accounts of SI in the popular press are even closer to our conceptualization of SEE. Goleman’s (2007) account of SI is perhaps the most widely known, and demonstrates this similarity quite nicely. He proposed a dual-factor construction of SI composed of social awareness and social facility. Social awareness is presented as the cognitive component of SI, while social facility is proposed to encompass the behavioral manifestations of social awareness that allows for smooth, effective interactions. Our conceptualization of SEE tracks closely with what Goleman has proposed as the core of SI. However, while the ideas presented by Goleman are logically supported by empirical work in other areas, such as cognitive neuroscience, the constructs that he proposed comprise SI are theoretically but not empirically driven. SEE represents a similar theoretical perspective, but we seek to empirically evaluate the behaviors and cognitive mechanisms that underpin successful social interactions. “Social awareness” closely resembles the cognitive components proposed as part of SEE, and is more reflective of the SI construct as it has been proposed and tested by others in the field, while “social facility” resembles the behavioral components of SEE, but is not at all reflective of SI as it has been empirically defined. SI has been, by the very nature of being labeled an intelligence, restricted to the cognitive abilities involved in understanding social situations. To involve behavioral components that require the translation of understanding into actions moves beyond the realm of an intelligence and into the domain of expertise. SEE is therefore proposed as a construct to account for the transition from understanding to behavior that SI cannot.

Empathy has been a long-standing focus of research on successful interactions (e.g., Davis, 1980; Singer, 2006). Yet as a construct, it has been difficult to define. The contemporary conceptualization of empathy, based on work conducted by Davis (1983, 1994), uses a multidimensional approach, with a cognitive understanding of another’s experience and the emotional reaction to the experiences of others being two key components. The interpersonal reactivity index (Davis, 1980) was a seminal measure in quantifying the multidimensional nature of empathy, with four primary factors identified: perspective taking, fantasy, empathic concern, and personal distress. These four factors encapsulate the qualities typically associated with empathy, and allow for these dimensions to be explored psychometrically (for related work, see Carré, Stefaniak, D’Ambrosio, Bensalah, & Besche-Richard, 2013; Lawrence et al., 2004). However, despite its many iterations and dimensions, conceptualizations of empathy all retain the core attribute of being an internal response to the perception of another’s experience. While empathy is no doubt vital for having smooth and productive social interactions, its various descriptions do not provide a mechanism by which it can be translated into high-quality interactions. SEE offers a hypothesized avenue through which social cognition is realized in the quality of an interaction through behavioral expression.

IS is a fourth construct for which we expect some associations with SEE. IS most typically refers to “the ability to accurately assess other people’s abilities, states, and traits from nonverbal cues” (Kenny, 2004; Montepare, 2004; Snodgrass, Hecht, & Ploutz-Snyder, 1998). To date, most research on IS has been primarily concerned with the accuracy of judgments concerning character attributes and emotional states of others (Kenny, 1994); these judgments presumably then affect how those making the judgments react to others (Gore, 2009). This ability to make accurate inferences about others’ emotional states and characteristics, as well as showing an overall awareness of social situations, could conceivably be related to SEE: Such judgments inform responses to a wide range of emotion-related circumstances with a variety of individuals. Much like EI and empathy, IS has been predominantly defined by the cognitive elements of social interactions. Unlike EI and empathy, IS has also been associated with behavior through constructs such as rapport (Hall & Bernieri, 2001; Tickle-Degnen & Rosenthal, 1987), thereby establishing a link between self-reports of behavior and third-party ratings of that behavior (Bernieri, Gillis, Davis, & Grahe, 1996). A vital test of SEE as a valid construct will be whether self-report ratings of the cognitive elements of SEE are associated with higher interaction quality as rated by third-party observers.

Standard measures of the aforementioned constructs rely heavily on both the accuracy of judgments and the ability to correctly label emotions (Ambady, LaPlante, & Johnson, 2001; Hall, DiMatteo, Rogers, & Archer, 1979; Mayer et al., 2003). Although SEE is expected to have some degree of association with these constructs, our current conceptualization is that SEE, while still incorporating accuracy, is more concerned with the quality and temporal dynamics of nonverbal behaviors during social interactions. Key to this emphasis on quality and temporal dynamics is the concept of fluency. Fluency has many meanings, but typically includes the concepts of ease of use and/or ease of retrieval from memory (Evans, 2008; Jacoby, Kelley, & Dywan, 1989; Kuhn & Stahl, 2003). Applying the definition of fluency in linguistics (i.e., the ability to use language accurately, automatically, and with prosody; Kuhn, Schwanenflugel, & Meisinger, 2010) to social interactions, we find an excellent starting point for how we conceptualize SEE.

Just as in language, social interactions are of the highest quality when the appropriate tone and rhythm are used (dynamics), both people show a level of understanding of one another (accuracy), and responses are delivered without extensive deliberation (automatic). While the accuracy component of SEE is less central to SEE than for EI and IS, we nonetheless consider accuracy to be an important aspect of SEE. Specifically, we posit that accuracy is a necessary but not sufficient condition for SEE. This position concerning accuracy therefore represents a point of potential overlap with associated constructs. What arguably differentiates SEE, however, are the behavioral manifestations of accurate social cognition. Namely, what are the specific behaviors and corresponding timing of those behaviors that are most closely predictive of high-quality social–emotional interactions? This is the question that we hope to answer with SEE. We speculate that SEE might be well-described as a “social–emotional toolkit,” composed of affect-related behaviors that promote higher quality interactions when used with optimal timing. Relative to those lower in SEE, individuals high in SEE should therefore have more flexibility with respect to how they interact with others. This flexibility affords the ability to adapt to a variety of social interactions, make ongoing adjustments throughout the duration of an interaction, and excel in ambiguous or awkward social situations. In other words, the flexibility afforded to someone who has high SEE allows for that individual to adapt to the rhythm of a given interaction.

Additionally, the rhythm of an interaction may vary depending on a host of demographic features, but particularly gender. Many social and affective measures show small but reliable gender differences, and gender has been shown to be an important factor in social interaction and communication (e.g., Baron-Cohen & Wheelwright, 2004; Friedman, Prince, Riggio, & DiMatteo, 1980; Schutte et al., 1998). Examination of the relationship between gender and SEE will be therefore of value.

The studies reported here focus on analyses of self-report data to develop a questionnaire for quantifying SEE through individuals’ self-assessments of SEE attributes. In these studies, we drew from the classic construct-validation recommendations of Cronbach and Meehl (1955) to (a) investigate the internal structure of the scale, (b) establish test–retest reliability, and (c) use other relevant self-report measures to evaluate convergent and discriminant validities of the SEE Scale.

General Method

Three studies were designed to develop and test the psychometric properties of a self-report measure of SEE. All materials and methods were approved by the institutional review board.

Participants

Participants were all 18 years of age or older and compensated for their participation, either through course credit (for undergraduate participants) or monetarily, through Amazon’s Mechanical Turk. For the purposes of scale development, an inclusion criterion was that each participant report that American English was their native language.

Procedure

The self-report surveys administered in each experiment were administered either in person (Experiments 1 and 3) or through Amazon’s Mechanical Turk (Experiment 2). All participants provided written informed consent before beginning the surveys.

Data Analysis

Data were analyzed using SPSS software. Items were generated initially to cover our intuitive conceptual domains of behavior that are involved in social interactions. Experiment 1 was designed as proof of concept and to remove redundant items from the initial item pool. Item removal was done by analyzing internal consistency of the items and eliminating items that had interitem and item-total correlations that were too high or too low, as well as items that did not have variability in participant responses. Experiment 2 was designed to evaluate the dimensionality of the construct. Exploratory factor analysis (EFA) was used to identify factors. Additionally, Pearson correlations were used to evaluate the relationships between the SEE Scale and related constructs. Experiment 3 was designed to measure the test–retest reliability of the SEE Scale, as well as to further evaluate the validity of the construct through its relationship with associated constructs. As a result, Pearson correlations were the primary statistical outcome.

Experiment 1

In this study, we tested our initial item sample of SEE-related attributes.

Method

Two separate groups of psychology undergraduate students participated in exchange for course credit. The first group (n = 55) ranged in age from 18 to 22 years (M = 19.62, SD = .99, data on participant sex were not collected). Data from one participant was excluded because his or her questionnaire was incomplete. Participants in the second group (n = 57) were between the age of 18 and 26 years (M = 19.82, SD = 1.50); 30% male (n = 17) and 70% female (n = 39). Race/ethnicity was not measured in either sample. Signed consent forms were returned alongside the completed questionnaires.

Participants in the first group were asked to complete an initial pool of 76 items. These items were written based on our conceptualization concerning the characteristics and features of the SEE construct. The items assessed various aspects of SEE, including perceived quality and timing of emotion-related signals (e.g., facial expressions, hand gestures), ease of interaction in a variety of emotion-related contexts, and how others might evaluate one’s social skills. Items were judged on a 7-point Likert-type scale with anchors of never, neutral, and always. The instructions were “Please answer each of the following items by circling the response that best describes what’s typical of you.” The initial pool of items contained several redundant items for each domain of behavior, with the intention of identifying the items that had the best psychometric properties within each domain. The resulting data were used in preliminary analyses that led to a 32-item revision of the SEE Scale.

Participants in the second group completed this revised version. Because response distributions for many of the items on the first version of the scale were negatively skewed, responses for the second version were on a 5-point Likert-type scale in an effort to increase the normality of the response distribution. Participants in the second group also completed the 13-item Marlowe–Crowne Social Desirability Scale (MCSDS; Reynolds, 1982); this scale was included in order to assess the extent to which SEE Scale scores were associated with social desirability.

Results and Discussion

First Iteration of the SEE Scale

Data analyses (i.e., Cronbach’s alpha, means, item-total correlations, and interitem correlations) were used to statistically determine which items should be removed from the SEE Scale according to the recommendations made by Clark and Watson (1995).

Due at least in part to the large number of items, the first version of the scale was statistically overdetermined, as indexed by its very high-internal consistency (76 items; α = .98). Ten items were eliminated because they had high means (>5.5 on a 7-point scale) and therefore low-response variability. However, for conceptual reasons, one item with a mean greater than 5.5 was retained. Of the 10 eliminated items, 3 also had low item-total correlations (<.30). Cronbach’s alpha on this shorter version (66 items) was .98. Based on interitem and item-total correlations that were either very low (<.15) or very high (>.70) as well as on conceptual grounds, 26 additional items were then eliminated (see Appendix A). The remaining 40 items had a Cronbach’s alpha of .97. To tighten the scale and further reduce the number of items, some items were reworded and others were eliminated. One additional item regarding laughter was added. The second version of the SEE Scale therefore consisted of 32 items.

Second Iteration of the SEE Scale

Cronbach’s alpha for this version of the SEE Scale (32 items) was .91. Based on interitem and item-total correlations as well as item content, 6 items were removed, leaving 26 remaining (α = .90). Total SEE Scale scores were unrelated to either age, r(54) = .11, p > .20, or gender, t(54) = −.86, p > .20. Importantly, total scores on the SEE Scale and total MCSDS scores had a nonsignificant Pearson correlation of r =.17 (p > .20), indicating that SEE responses are largely independent of social desirability.

Following the analyses described for Experiment 2 and stemming from a recommendation by colleagues, one additional item was removed from the SEE Scale due to its ambiguous nature and poor fit with the factor structure of the scale (see Appendix A). Though the SEE Scale used in Experiments 2 and 3 included this item, responses to this item were not used in statistical analyses. The final version of the SEE Scale (see Appendix A for the removed items and Appendix B for the final scale) therefore consists of 25 items.

Experiment 2

This study was designed to further explore the psychometric properties of the 25-item version of the scale using EFA. Convergent and discriminant validity was also examined by comparing scores on the SEE Scale with scores from measures of related constructs such as EI, IS, and empathy.

Method

Participants were recruited through Amazon’s online “Mechanical Turk” marketplace, in which workers can complete tasks in exchange for small amounts of monetary compensation. For this stage of SEE Scale development, in an effort to reduce potential culture- or language-related confounds, participants were limited to U.S. residents (as determined from the IP address of the computer used to complete the surveys). Participants’ (total n = 1,000) data were therefore excluded if they resided outside of the United States (n = 121), were nonnative English speakers (n = 44), returned invalid responses (i.e., the same choice was selected for every question) or the majority of the questionnaires were incomplete (n = 62). Data from the remaining participants (n = 885) were used in EFA. All participants provided an electronic signature to consent to participation in the study and were compensated with $0.25 for completion of the surveys. Participants were 37.9% (n = 335) male, 61.5% (n = 544) were female, and 0.6% (n = 6) identified as other or preferred not to answer. Participants’ age ranged from 18 to 76 years (M = 33.99, SD = 12.13). The participant sample was well educated: 13.1% (n = 116) had graduate degrees, 41.7% (n = 369) were college graduates, 33.2% (n = 294) had some college, 11.1% (n = 98) completed 11 to 12 years of school, and 0.8% (n = 8) completed 10 or fewer years of school. Most participants identified themselves as White (77.9%, n = 689), with 9.5% (n = 84) identified as Black/African American, 6.2% (n = 55) identified as Asian, 3.1% (n = 27) identified as Hispanic or Latino, and 3.4% (n = 30) identified as other or preferred not to answer.

A link to a secure, online version of self-report measures using REDCap software (Harris et al., 2009) was made available on Mechanical Turk. Participants completed the SEE Scale and several measures expected to provide evidence of convergent and discriminant validity. The MCSDS (Reynolds, 1982), the Interpersonal Sensitivity Measure (IPSM; Boyce & Parker, 1989), and the Self-Monitoring Scale (SMS; Snyder, 1974) were hypothesized to be measures of discriminant validity and therefore unrelated to or negatively correlated with the SEE Scale. The EI Scale (Schutte et al., 1998) and the Social Interaction Anxiety Scale (SIAS; Mattick & Clarke, 1998), on the other hand, were hypothesized to be measures of convergent validity and positively correlate with the SEE Scale.

Results and Discussion

Exploratory Factor Analysis

Internal structure of the final version of the SEE Scale was examined using EFA to statistically discover underlying latent factors. Rather than principal component analysis (which identifies factors based on linear combinations to find optimal groupings for subscales), “true” EFA was used in order to identify factors based solely on common variance. EFA is a desirable approach in this situation because it identifies underlying structures and latent factors (Fabrigar, Wegener, MacCallum, & Strahan, 1999). Principal-axis factoring was used to identify latent factors using oblique (Promax) rotation, which allows for factors to be intercorrelated. To ascertain the number of statistically meaningful factors, the following metrics were considered: (a) the interpretability of each solution, (b) factors with eigenvalues greater than 1, and (c) factor loadings greater than or equal to 0.30. Additionally, because the criterion of using eigenvalues greater than 1 can yield spurious factors (Velicer & Jackson, 1990), a parallel analysis (Horn, 1965) was conducted to determine which factors were ultimately suitable to retain.

All items were highly correlated (i.e., >.40) with the SEE Scale total score (see Table 1). The principal-axis factor analysis revealed four factors with eigenvalues greater than 1.00. The eigenvalue of the first factor (9.35) clearly met that goal, whereas the remaining factors had much lower eigenvalues (1.72, 1.33, and 1.13). The parallel analyses, conducted both with random data and with permutations on the raw data, indicated the possibility of nine factors, two of which had eigenvalues greater than 1 (8.81 and 1.15). Because parallel analyses of this kind tend to identify more factors than are statistically warranted, many of the resultant factors should be disregarded despite their statistical significance (Buja & Eyuboglu, 1992).

Table 1.

SEE Scale Mean Rating (SD), Factor Loading, and Item-Total Scale Correlation for Each Scale Item in Experiment 2.

Factor	SEE Scale items	M (SD) rating (0-4)	Factor I loading	Factor II loading	Item-total scale correlation
Factor I (Adaptability)	I’m good at making eye contact (1)	2.63 (1.13)	.55	−.02	.55
	I can talk easily with people of any level (kids, peers, professors, etc.) (2)	2.65 (1.07)	.65	.001	.65
	In social interactions, my facial expressions are perfectly timed (4)	2.24 (0.98)	.52	.11	.62
	I’m good at confronting people about sensitive situations without making them feel awkward or disrespected (5)	2.21 (1.14)	.79	−.21	.64
	I can express annoyance without putting people off (6)	2.17 (1.02)	.72	−.22	.53
	I use just the right amount and kind of touch in my social interactions (7)	2.36 (1.02)	.64	.004	.66
	In conversations, my hand gestures are helpful, not distracting (8)	2.56 (0.92)	.36	.15	.50
	I can easily draw on my various social skills as situations warrant (9)	2.47 (1.03)	.61	.13	.70
	At social events like parties, people often introduce themselves to me (12)	2.03 (1.08)	.55	.06	.64
	I know just the right things to say and do when someone I know is upset (13)	2.17 (1.07)	.52	.01	.61
	I can make awkward social interactions feel more comfortable (15)	2.19 (1.10)	.62	.12	.71
	I produce the right sorts of smiles at just the right times (16)	2.37 (1.00)	.45	.26	.68
	If people observed me in a group, they would say I’m the most socially gifted (17)	1.52 (1.16)	.61	.08	.68
	I’m a natural at knowing how to coordinate my emotional responses to others’ emotions (18)	2.22 (1.03)	.58	.17	.71
	I have a relaxed, open body posture when I talk with people (19)	2.34 (1.08)	.65	.01	.66
	I know how to calm a heated conversation (24)	2.39 (1.05)	.52	.08	.64
Factor II (Expressivity)	I use people’s body language to help me know how to respond to them (3)	2.91 (0.88)	.16	.43	.53
	I’m animated when I speak (10)	2.37 (1.12)	−.12	.61	.44
	I’m good at reading facial expressions (11)	2.91 (0.90)	.17	.41	.50
	I use the qualities of my voice to influence others (14)	2.27 (1.04)	.15	.54	.65
	Others would say I have an expressive face (20)	2.49 (1.10)	−.12	.77	.59
	People are swayed or influenced by my emotional signals (21)	2.10 (0.99)	.07	.65	.65
	I use my voice to convey my emotions (22)	2.56 (1.00)	−.13	.78	.58
	I nod my head the right amount to let others know that I’m listening (23)	2.89 (0.84)	.03	.45	.45
	I’m good at using laughter to make other people feel good (25)	2.82 (1.02)	.21	.36	.61

Note. SEE = social–emotional expertise. Item numbers are given between parentheses.

Taking this package of statistical outcomes into account in conjunction with our conceptualization of SEE, we arrived at a solution in which the SEE Scale consists of two factors: (a) Factor I, which we have labeled “Adaptability” (eigenvalue = 9.35, 37.38% of variance explained), consists of 16 items that are largely characterized by the ability to easily adapt to a variety of social and emotional interpersonal situations (e.g., “I know just the right things to say and do when someone I know is upset”); and (b) Factor II, labeled “Expressivity” (eigenvalue = 1.72, 6.87% of variance explained), consists of 9 items (e.g., “I’m animated when I speak”) that each reflect the ability to express emotion to others. The two factors correlated with each other, r(883) = .67, p < .001. Cronbach’s alpha for the SEE Scale total score was .92, and values for the Adaptability and Expressivity subscales were also high (i.e., .91 and .82, respectively), with an average interitem correlation of .32. Different patterns of correlations between the Adaptability and Expressivity subscales and measures of convergent and discriminant measures of validity would further support the presence of these two factors (see Experiment 3 for discussion of these results).

Gender Differences

There were significant gender differences for total SEE Scale and Expressivity subscale scores: t(877) = −2.83, p < .01, and t(877) = −5.62, p < .001, respectively. In both cases, females provided somewhat higher self-report ratings than males (see Table 2). The Adaptability subscale scores were not associated with participant gender, t(877) = −1.06, p > .05. Although the gender effects that were obtained were statistically significant, the associated effect sizes were small to medium (see Table 2). Moreover, these gender differences were not unexpected, as measures of many social and affective constructs show higher scores for females than for males (e.g., Baron-Cohen & Wheelwright, 2004; Friedman et al., 1980; Schutte et al., 1998).

Table 2.

SEE Scale Total and Factor Mean Scores (SD) and Gender Differences in Experiment 2.

	Total Sample (N = 879)	Males (n = 335)	Females (n = 544)	t(877)	d
SEE Scale total score	60.79 (15.42)	58.96 (14.84)	61.98 (15.71)	−2.83**	.20
Adaptability	37.14 (10.97)	36.66 (10.40)	37.48 (11.35)	−1.06	.08
Expressivity	23.65 (5.75)	22.29 (5.65)	24.50 (5.66)	−5.62***	.39

Note. SEE = social–emotional expertise.

p < .01 ***p < .001.

Convergent and Discriminant Validity

To evaluate convergent and discriminant validity via self-report, correlations among the SEE Scale and self-report measures of several associated constructs were examined (see Table 3). Convergent validity with related constructs was supported, as SEE had significant negative correlations with both social anxiety, r(883) = −.59, p < .01, and IS, including both the total IPSM score, r(883) = −.23, p < .01, and its subscale scores (see Table 3). These negative correlations were expected because the SIAS and IPSM measures were created to evaluate distress associated with hypersensitivity to negative aspects of social interactions. This hypersensitivity to negative social stimuli is in contrast with SEE, which may incorporate a hypersensitivity to aspects of social interactions, but not a preoccupation with negative experiences. Discriminant validity was supported by social desirability (MCSDS) scores, which were weakly correlated with total SEE scores, r(883) = .19, p < .01. This relationship was largely accounted for by the Adaptability factor scores, r(883) = .28, p < .01. SEE total scores were also correlated with self-monitoring scores, r(883) = .28, all ps < .01.

Table 3.

Associations Among the SEE Scale and Self-Report Measures of Convergent and Discriminant Validity in Experiment 2.

Measure	SEE Scale total	SEE_Adaptability factor	SEE_Expressivity factor	M	SD
Discriminant measures
MCSDS	.19**	.28**	−.01	6.13	3.04
SMS	.28**	.23**	.31**	11.70	4.22
Convergent measures
IPSM	−.23**	−.31**	−.03	53.74	16.69
Interpersonal awareness	−.34**	−.43**	−.10**	10.90	4.96
Need for approval	.30**	.23**	.37**	16.43	3.77
Separation anxiety	−.34**	−.41**	−.12**	10.11	5.10
Timidity	−.06	−.08*	−.01	11.36	4.46
Fragile inner self	−.35**	−.39**	.21**	4.93	3.51
EI	.62**	.57**	.58**	87.84	16.69
SIAS	−.59**	−.65**	−.35**	33.54	15.89
M	60.79	37.14	23.65
SD	15.42	10.97	5.75

Note. N = 885. MCSDS = Marlowe–Crowne Social Desirability Scale; SMS = Self-Monitoring Scale; IPSM = Interpersonal Sensitivity Measure (including subscales); EI = Emotional Intelligence measure; SIAS = Social Interaction Anxiety Scale. The SMS, EI, and SIAS measures were used to test for convergent validity, whereas the MCSDS and IPSM were used to test for discriminant validity.

p < .05. **p < .01.

Also as expected, SEE scores were strongly and positively correlated with self-reports of EI, r(883) = .62, p < .01. Because of the strong correlation with EI, an EFA was conducted using the items from both the SEE Scale and the EI measure. Items from both scales were comparable in that they both measured responses using a 5-point Likert-type scale. Based on the scree plot, five factors were extracted using principal axis factoring with Promax rotation (two SEE factors with factor loadings from 0.28 to 0.74; three EI factors with factor loadings ranging from 0.23 to 0.77). With only four exceptions (out of 58 total items), the items from the two scales loaded on separate factors, indicating that the SEE and EI scales contain items that reflect different latent factors. In other words, these results suggest—at least via these two measurement scales—that SEE and EI are separate constructs. Alternatively, given the strong correlation between the two scales, it is possible that a common higher level factor might occur upstream of the factors revealed by this analysis.

Experiment 3

The purposes of this third study were to (a) assess the test–retest reliability of the SEE Scale and (b) to further assess convergent and discriminant validity using additional scales that measure theoretically related constructs.

Method

Undergraduate students (n = 82) participated for extra credit. Participants ranged in age from 18 to 29 years; 32.9% male (n = 27) and 67.1% female (n = 55) and were 52.4% White, 25.6% Asian/Pacific Islander, 11.0% Hispanic/Latino, 2.4% African American, and 8.5% other.

Participants, who were tested in a psychology laboratory, completed the SEE Scale on two separate occasions, approximately 4 weeks apart (M = 32.02 days), with 70.7% of participants completing both SEE-scale administrations on both occasions. Across the two questionnaire administrations, participants also completed several measures of hypothesized convergent validity: the Basic Empathy Scale–Adult (BES-A; Carré et al., 2013), EI Scale (Schutte et al., 1998), Affective Communication Test (ACT; Friedman et al., 1980), personality traits (NEO-FFI-3; McCrae & Costa, 2010), Berkeley Expressivity Questionnaire (BEQ; Gross & John, 1995), Empathy Quotient (EQ; Baron-Cohen & Wheelwright, 2004), the SIAS, and the SMS. Measures expected to be only weakly associated with the SEE Scale included the Emotional Intensity Scale (EIS; Bachorowski & Braaten, 1994), the IPSM, the MCSDS, the Subjective Happiness Scale (SHS; Lyubomirsky & Lepper, 1999), the Satisfaction With Life Scale (SLS; Diener, Emmons, Larsen, & Griffin, 1985), and the interpersonal reactivity index (empathy and perspective-taking subscales only; Davis, 1980).

Results and Discussion

The SEE Scale exhibited good test–retest validity, with a correlation of r(80) = .82, p < .001. As expected, SEE Scale scores were positively correlated with the measures we included for their potential to provide evidence of convergent validity (see Table 4). Most notably, SEE Scale scores and SEE factor scores were strongly correlated with EI, SEE_Total r(80) = .60, p < .001; SEE_Adaptibility r(80) = .56, p < .001; SEE_Expressivity r(80) = .47, p < .001, and extraversion, SEE_Total r(80) = .60, p < .001; SEE_Adaptibility r(80) = .55, p < .001; SEE_Expressivity r(80) = .49, p < .001. SEE Scale scores also had correlations of moderate strength with EQ scores, SEE_Total r(80) = .43, p < .001; SEE_Adaptibility r(80) = .40, p < .001; SEE_Expressivity r(80) = .33, p < .001, and SHS scores, SEE_Total r(80) = .42, p < .001; SEE_Adaptibility r(80) = .41, p < .001; SEE_Expressivity r(80) = .33, p < .001.

Table 4.

Test–Retest Validity and Correlations With Convergent and Discriminant Validity Measures in Experiment 3.

Measure	SEE Scale total (test)	Adaptability factor	Expressivity factor	SEE Scale total (retest)
SEE Scale total (test)	—	.94**	.81**	.82**
Adaptability factor	.94**	—	.57**	.79**
Expressivity factor	.81**	.57**	—	.60**
SEE Scale total (retest)	.82**	.79**	.60**	—
IPSM	−.13	−.19	.00	−.16
SIAS	−.56**	−.59**	−.39**	−.67**
SMS	−.33**	−.24*	−.40**	−.30*
SLS	.24*	.26*	.15	.29*
ACT	.62**	.54**	.61**	.65**
Empathy questionnaire	.43**	.40**	.33**	.41**
EIS	−.04	−.08	.02	−.09
BES-A	.08	.04	.15	.23
EI	.60**	.56**	.47**	.78**
IRI-empathy	.03	.03	.03	.19
IRI-perspective taking	.10	.07	.09	.17
BEQ	.14	.00	.34*	.16
MCSDS	−.15	−.13	−.14	−.16
NEO subscales
Neuroticism	−.25	−.31*	−.07	−.30*
Extraversion	.60**	.55**	.49**	.61**
Openness	.16	.04	.30*	.16
Agreeableness	−.10	−.07	−.17	.00
Conscientiousness	.02	.09	−.08	.05
SHS	.42**	.41**	.33*	.55**

Note. N = 80. IPSM = Interpersonal Sensitivity Measure; SIAS = Social Interaction Anxiety Scale; SMS = Self-Monitoring Scale; SLS = Satisfaction With Life Scale; ACT = Affective Communication Test; EIS = Emotional Intensity Scale; BES-A = Basic Empathy Scale–Adult; EI = Emotional Intelligence measure; IRI = interpersonal reactivity index; BEQ = Berkeley Expressivity Questionnaire; MCSDS = Marlowe–Crowne Social Desirability Scale; SHS = Subjective Happiness Scale.

p < .05. **p < .01.

As in Experiment 2, SEE Scale scores were strongly and negatively correlated with social anxiety as measured with SIAS scores, r(80) = −.56, p < .001. This global correlation was driven to a larger degree by the SEE_Adaptability factor, r(80) = −.59, p < .001, than the SEE_Expressivity factor, r(80) = −.39, p < .001.

Even though the two factors of the SEE Scale, Adaptability and Expressivity, are themselves highly correlated, r(80) = .57, p < .001, there were some instances in which convergent or discriminant measures support the idea that these factors are conceptually unique (see Table 4). For example, self-monitoring, as measured with SMS scores, had a stronger negative correlation with the Expressivity factor, r(80) = −.40, p < .001, than it did with the Adaptability factor, r(80) = −.24, p < .01. Neuroticism scores, as measured with the NEO, was not significantly correlated with either the Expressivity factor or the SEE Scale total score, although it was correlated with r(80) = −.31, p < .001. Similarly, the personality trait of Openness correlated with the SEE_Expressivity, r(80) = .30, p < .001, but was not related to either the SEE_Adaptability or SEE_Total score. EI was also more strongly correlated with SEE_Adaptibility than it was with SEE_Expressivity, indicating that SEE_Adaptibility may overlap with the cognitive components necessary for SEE more than SEE_Expressivity.

General Discussion

The results from the studies reported here describe the development of a self-report measure of SEE and its basic statistical properties. Through three item iterations and four studies (two studies as part of Experiment 1), the resultant SEE Scale consists of 25 items, each scored with a 5-point Likert-type scale. The scale has both high-internal consistency and test–retest reliability. Factor analyses revealed that the SEE Scale consists of two factors, Adaptability and Expressivity. This two-factor solution is further supported by different patterns of correlations with measures of related constructs.

In conceptually expected patterns, the SEE Scale was also shown to have convergent and discriminant validity—at least in relation to other self-report measures. Although SEE Scale scores were strongly correlated with EI, this correlation was expected in that SEE is thought to build on the two of the four EI branches (i.e., perceiving emotions and expressing emotions; Salovey & Mayer, 1990). An EFA of the SEE and EI measures showed that, for the most part, items from the two scales loaded on different factors. This outcome suggests that although the constructs are conceptually related, they are also statistically separable. In other words, SEE is not best described as a subcomponent of EI. Further work investigating the relationship between SEE and its factors and the four subcomponents of EI may reveal more nuanced associations between the two constructs. Also as anticipated, social anxiety was strongly and negatively correlated with SEE scores, indicating that particular patterns of SEE behaviors could be expressed or notably absent in anxious individuals. For instance, SEE_Adaptability had a stronger negative correlation with SIAS scores than did SEE_Expressivity. This pattern indicates that people with social anxiety may be less flexible in the ways they are able to interact with others. Further investigation in a population with social anxiety could help elucidate the contribution SEE may make to the development or presentation of social anxiety.

There are some key limitations to the experiments presented above. First, the participant samples are not representative of the population at large and were small in Experiment 1, limiting generalizability of the findings. Future studies will benefit from larger, more racially and ethnically diverse samples which may offer insight into differences in SEE across cultures. Second, there are well-documented limitations to data collected from participants on Amazon’s Mechanical Turk, such as paying less attention to experimental materials (e.g., Goodman, Cryder, & Cheema, 2013). Finally, self-reported SEE scores were not compared with ratings of social ability by third-party observers or expert judges, limiting conclusions that can be made about the correlations of SEE Scale total scores with the real-life social abilities. Future studies will benefit from multiple sources of interaction quality rating to examine the validity of the SEE Scale.

Future Directions

Continuing with our process of construct validation (Cronbach & Meehl, 1955), ongoing studies are focused on investigating the behavioral validity of the SEE construct and its associations with other relevant constructs. Our current foci are measuring SEE in naturalistic laboratory-based paradigms, and studying perceptual judgments and psychophysiological responses to individuals who differ in SEE Scale scores. A second key avenue will be to examine the cognitive processes, such as the speed and accuracy of retrieval, that lead to more socially adaptive behavior. Finally, it will be of interest to assess SEE in clinical populations in which social behaviors are typically impaired, such as with social anxiety and autism spectrum disorder.

Footnotes

Appendix A

Appendix B

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: REDCap data capture and management instruments were supported by UL1 TR000445 from NCATS/NIH.

References

Ambady

LaPlante

Johnson

(2001). Thin-slice judgments as a measure of interpersonal sensitivity. In Hall

J. A.

Bernieri

F. J.

(Eds.), Interpersonal sensitivity: Theory and measurement (pp. 89-102). Mahwah, NJ: Lawrence Erlbaum.

Bachorowski

J.-A.

Braaten

(1994). Emotional intensity: Measurement and theoretical implications. Personality and Individual Differences, 17, 191-199.

Barnes

M. L.

Sternberg

R. J.

(1989). Social intelligence and decoding of nonverbal cues. Intelligence, 13, 263-287.

Baron-Cohen

Wheelwright

(2004). The empathy quotient: An investigation of adults with Asperger syndrome or high functioning autism, and normal sex differences. Journal of Autism and Developmental Disorders, 34, 163-175.

Bernieri

F. J.

Gillis

J. S.

Davis

J. M.

Grahe

J. E.

(1996). Dyad rapport and the accuracy of its judgment across situations: A lens model analysis. Journal of Personality and Social Psychology, 71, 110-129.

Boyce

Parker

(1989). Development of a scale to measure interpersonal sensitivity. Australian and New Zealand Journal of Psychiatry, 23, 341-351.

Buja

Eyuboglu

(1992). Remarks on parallel analysis. Multivariate Behavioral Research, 27, 509-540.

Carré

Stefaniak

D’Ambrosio

Bensalah

Besche-Richard

(2013). The Basic Empathy Scale in Adults (BES-A): Factor structure of a revised form. Psychological Assessment, 23, 679-691.

Clark

Watson

(1995). Constructing validity: Basic issues in objective scale development. Psychological Assessment, 7, 309-319.

10.

Cronbach

Meehl

(1955). Construct validity in psychological tests. Psychological Bulletin, 52, 281-302.

11.

Davis

M. H.

(1980). A multidimensional approach to individual differences in empathy. Retrieved from http://www.ucp.pt/site/resources/documents/ICS/GNC/ArtigosGNC/AlexandreCastroCaldas/24_Da80.pdf

12.

Davis

M. H.

(1983). Measuring individual differences in empathy: Evidence for a multidimensional approach. Journal of Personality and Social Psychology, 44, 113-126.

13.

Davis

M. H.

(1994). Empathy: A social psychological approach. Boulder, CO: Westview Press.

14.

Diener

E. D.

Emmons

R. A.

Larsen

R. J.

Griffin

(1985). The Satisfaction With Life Scale. Journal of Personality Assessment, 49, 71-75.

15.

Evans

(2008). The sociology of expertise: The distribution of social fluency. Sociology Compass, 2, 281-298.

16.

Fabrigar

L. R.

Wegener

D. T.

MacCallum

R. C.

Strahan

E. J.

(1999). Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods, 4, 272-299.

17.

Ford

M. E.

Tisak

M. S.

(1983). A further search for social intelligence. Journal of Educational Psychology, 75, 197-206.

18.

Friedman

H. S.

Prince

L. M.

Riggio

R. E.

DiMatteo

M. R.

(1980). Understanding and assessing nonverbal expressiveness: The Affective Communication Test. Journal of Personality and Social Psychology, 39, 333-351.

19.

Goldenberg

Matheson

Mantler

(2006). The assessment of emotional intelligence: A comparison of performance-based and self-report methodologies. Journal of Personality Assessment, 86, 33-45.

20.

Goleman

(2007). Social intelligence: The new science of human relationships. New York, NY: Random House.

21.

Goodman

J. K.

Cryder

C. E.

Cheema

(2013). Data collection in a flat world: The strengths and weaknesses of Mechanical Turk samples. Journal of Behavioral Decision Making, 26, 213-224.

22.

Gore

(2009). The interaction of sex, verbal, and nonverbal cues in same-sex first encounters. Journal of Nonverbal Behavior, 33, 279-299.

23.

Gross

John

(1995). Facets of emotional expressivity: Three self-report factors and their correlates. Personality and Individual Differences, 19, 555-568.

24.

Hall

J. A.

Bernieri

F. J.

(Eds.). (2001). Interpersonal sensitivity: Theory and measurement. Mahwah, NJ: Lawrence Erlbaum.

25.

Hall

J. A.

DiMatteo

M. R.

Rogers

P. L.

Archer

(1979). Sensitivity to nonverbal communication: The PONS test. Baltimore, MD: Johns Hopkins University Press.

26.

Harris

P. A.

Taylor

Thielke

Payne

Gonzalez

Conde

J. G.

(2009). Research electronic data capture (REDCap): A metadata-driven methodology and workflow process for providing translational research informatics support. Journal of Biomedical Informatics, 42, 377-381.

27.

Horn

J. L.

(1965). A rationale and test for the number of factors in factor analysis. Psychometrika, 30, 179-185.

28.

Jacoby

L. L.

Kelley

C. M.

Dywan

(1989). Memory attributions. In Roediger

H. L.

Craik

F. I. M.

(Eds.), Varieties of memory and consciousness: Essays in honour of Endel Tulving (pp. 391-422). Hillsdale, NJ: Lawrence Erlbaum.

29.

Kenny

D. A.

(1994). Interpersonal perception: A social relations analysis. New York, NY: Guilford Press.

30.

Kenny

D. A.

(2004). PERSON: A general model of interpersonal perception. Personality and Social Psychology Review, 8, 265-280.

31.

Kihlstrom

J. F.

Cantor

(2000). Social intelligence. In Sternberg

R. J.

(Eds.), Handbook of intelligence (pp. 359-379). Cambridge, England: Cambridge University Press.

32.

Kuhn

M. R.

Schwanenflugel

P. J.

Meisinger

E. B.

(2010). Aligning theory and assessment of reading fluency: Automaticity, prosody, and definitions of fluency. Reading Research Quarterly, 45, 230-251.

33.

Kuhn

M. R.

Stahl

S. A.

(2003). Fluency: A review of developmental and remedial practices. Journal of Educational Psychology, 95, 3-21.

34.

Lawrence

E. J.

Shaw

Baker

Baron-Cohen

David

A. S.

(2004). Measuring empathy: reliability and validity of the Empathy Quotient. Psychological Medicine, 34, 911-920.

35.

Lyubomirsky

Lepper

H. S.

(1999). A measure of subjective happiness: Preliminary reliability and construct validation. Social Indicators Research, 46, 137-155.

36.

Mattick

Clarke

(1998). Development and validation of measures of social phobia scrutiny fear and social interaction anxiety. Behavior Research and Therapy, 36, 455-470.

37.

Mayer

J. D.

Salovey

(1993). The intelligence of emotional intelligence. Intelligence, 17, 433-442.

38.

Mayer

J. D.

Salovey

(1997). What is emotional intelligence? In Salovey

Sluyter

(Eds.), Emotional development and emotional intelligence: Educational implications (pp. 3-31). New York, NY: Basic Books.

39.

Mayer

J. D.

Salovey

Caruso

D. R.

(2008). Emotional intelligence: New ability or eclectic traits? The American Psychologist, 63, 503-517.

40.

Mayer

J. D.

Salovey

Caruso

D. R.

Sitarenios

(2003). Measuring emotional intelligence with the MSCEIT V2.0. Emotion, 3, 97-105.

41.

McCrae

R. R.

Costa

P. T.

(2010). NEO Inventories: Professional manual. Odessa, FL: Psychological Assessment Resources.

42.

Montepare

J. M.

(2004). Exploring interpersonal sensitivity: What, who, why, and to what end? Journal of Nonverbal Behavior, 28, 143-145.

43.

Reynolds

(1982). Development of reliable and valid short forms of the Marlowe-Crowne Social Desirability Scale. Journal of Clinical Psychology, 38, 119-125.

44.

Salovey

Mayer

J. D.

(1990). Emotional intelligence. Imagination, Cognition and Personality, 9, 185-211.

45.

Schutte

N. S.

Malouff

J. M.

Hall

L. E.

Haggerty

D. J.

Cooper

J. T.

Golden

C. J.

Dornheim

(1998). Development and validation of a measure of emotional intelligence. Personality and Individual Differences, 25, 167-177.

46.

Singer

(2006). The neuronal basis and ontogeny of empathy and mind reading: Review of literature and implications for future research. Neuroscience & Biobehavioral Reviews, 30, 855-863.

47.

Snodgrass

S. E.

Hecht

M. A.

Ploutz-Snyder

(1998). Interpersonal sensitivity: Expressivity or perceptivity? Journal of Personality and Social Psychology, 74, 238-249.

48.

Snyder

(1974). Self-monitoring of expressive behavior. Journal of Personality and Social Psychology, 30, 526-537.

49.

Spunt

R. P.

Lieberman

M. D.

(2013). The busy social brain: Evidence for automaticity and control in the neural systems supporting social cognition and action understanding. Psychological Science, 24, 80-86.

50.

Sternberg

R. J.

Smith

(1985). Social intelligence and decoding skills in nonverbal communication. Social Cognition, 3, 168-192.

51.

Thorndike

E. L.

(1920). Intelligence and its uses. Harper’s Magazine, 140, 227-235.

52.

Thorndike

R. L.

Stein

(1937). An evaluation of the attempts to measure social intelligence. Psychological Bulletin, 34, 275-285.

53.

Tickle-Degnen

Rosenthal

(1987). Group rapport and nonverbal behavior. In Hendrick

(Ed.), Group processes and intergroup relations (pp. 113-136). Newbury Park, CA: Sage.

54.

Velicer

W. F.

Jackson

D. N.

(1990). Component analysis versus common factor analysis: Some issues in selecting an appropriate procedure. Multivariate Behavioral Research, 25, 1-28.

55.

Wechsler

(1958). The measurement and appraisal of adult intelligence (4th ed.). Baltimore, MD: Lippincott Williams & Wilkins.