Abstract
This Special Issue of the Journal of Psychoeducational Assessment offers a critical appraisal of the validity, applied utility, and limitations of self-report assessments of emotional competencies. Using self-concept theory as an integrative theoretical framework, this introductory editorial highlights key methodological and validity issues raised in the contributing articles: (a) distinction between emotional competence self-perceptions and objectively measured abilities, (b) effects of response biases and respondents’ age on the psychometric properties of self-reports, (c) importance of adopting a multi-dimensional assessment strategy, and (d) various aspects of construct validity (conceptual definitions and paradigms, gender differences, relationships with basic personality, mechanisms and scope of prediction). The added value of conceptualizing emotional competence self-reports as self-concepts (as proposed in this article) is illustrated in the discussion of practical implications, outstanding questions, and directions for future research on the meaning and uses of these assessments.
A great deal of research shows that learning to identify, express, empathize with, and regulate emotions constitutes an essential aspect of human development, linked to successful adaptation, social integration, goal achievement, and overall health and well-being (Denham, 1998; MacCann, Lipnevich, & Roberts, 2012; Saarni, 1999; Trentacosta & Fine, 2009). Helping children and youth develop their emotional competencies has become a core mandate of educational systems around the world, with many secondary and post-secondary institutions implementing programs aimed explicitly at enhancing students’ socioemotional learning (Dacre Pool & Qualter, 2012; Durlak, Weissberg, Dymnicki, Taylor, & Schellinger, 2011; Greenberg et al., 2003). The efficacy and sustainability of these efforts depend squarely on having a solid conceptual and assessment base from which emotional competencies can be monitored and understood.
I use the term “emotional competencies” loosely in this editorial to represent a varied hierarchy of constructs permeating the literature, from specific emotion-related skills (e.g., recognition and labeling of facial emotional expressions), to more general abilities and dispositions (e.g., emotional understanding, attention to emotions), to broadband meta-constructs (e.g., emotional intelligence [EI]; Bar-On & Parker, 2000). Complementing the diversity of constructs is the diversity of approaches used to measure these variables, which range from simple laboratory tasks and self-report questionnaires to complex situational judgments and 360° collateral ratings by professionals and close others (Stough, Saklofske, & Parker, 2009). This conceptual and methodological heterogeneity makes it difficult to integrate research findings in this area or to select appropriate measurement tools for specific applications (R. D. Roberts, Matthews, & Zeidner, 2010). Although combining multiple methods of assessments is considered to be the gold standard of evidence-based practice, different tools and procedures invariably produce inconsistent results (Meyer et al., 2001). How exactly these cross-method discrepancies should be interpreted and dealt with remains a topic of continuous research and discussion (De Los Reyes, 2011; Ganellen, 2007).
The focus of this Special Issue of the Journal of Psychoeducational Assessment (JPA) is on the self-report method in particular, which is by far the most commonly used, yet also arguably the most contentious approach to assessing emotion-related abilities (Mayer, Salovey, & Caruso, 2008). In this introductory article, I briefly outline the perennial dilemmas associated with competence self-reports and the key questions put forth before the contributors of this issue. I then present an integrative theoretical framework to organize the various themes that recur throughout this issue’s articles and throughout the corresponding literature as a whole. I conclude with a discussion of the outstanding questions, future research directions, and practical implications arising from the proposed integrative framework and from the leading theories, state-of-the-art tools, and novel validity findings showcased in this issue’s contributions. The ultimate goal of this volume is to critically re-appraise the meaning, utility, and limitations of self-report scales of emotional competencies; stimulate further discussion and innovative research on these topics; and promote most informed interpretations and applications of self-report measures, whether used alone or as part of a multi-method assessment strategy.
The Precarious Nature of Competence Self-Reports
Of all the available methods, self-report is perhaps the most intuitively appealing mode of assessment for emotion-related constructs. After all, in the famous words of Gordon Allport (1942), “if we want to know how people feel: what they experience and what they remember, what their emotions and motives are like, and the reasons for acting as they do—why not ask them?” (p. 37). Coupled with their ease of administration, self-report scales have long been, and will continue to be widely used in psychological research and practice. At the same time, most of us have been trained to approach self-report data with a healthy dose of caution, acknowledging that people often have limited insight into their mental processes and capacities, or else they may be motivated to misrepresent themselves, knowingly or not (Dunning, Heath, & Suls, 2004). Take the classic case of intellectual abilities, for example. Some people believe that intelligence is a fixed entity that cannot be changed much throughout life, whereas others believe that intelligence can be enhanced via learning and experience—irrespective of what the actual nature of intelligence is (Dweck, 1999; Molden & Dweck, 2006). Likewise, people’s self-estimates of intelligence show little congruence with their performance on objective IQ tests (Freund & Kasten, 2012), and more often than not, people are motivated by implicit self-enhancement biases leading them to overestimate their intelligence and other positive attributes (Bouffard & Narciss, 2011).
If people’s perceptions of their abilities are so inaccurate, why not disregard them and focus only on actual abilities? Herein lies the precarious nature of competence self-reports: On the one hand, they may correspond little to the “objective” reality. On the other hand, decades of research convincingly demonstrate that people’s subjective competence beliefs act as powerful motivators of observable behavior (Bandura, 1997; Dweck, 1999; Elliot & Dweck, 2005). Believers in the fixed nature of intelligence see challenge and failure as signs of their lacking ability; because they believe their ability cannot be changed, they choose easier tasks, exert less effort, and react to setbacks with defensiveness and withdrawal, in attempts to avoid demonstrating their perceived incompetence (Dweck, 1999; Molden & Dweck, 2006). In contrast, believers in the incremental nature of intelligence see challenge and failure as opportunities to grow; consequently, they seek out more difficult tasks, put in more effort, and persevere in the face of setbacks, in attempts to improve their skills and abilities. For these reasons, students who hold high academic self-efficacy beliefs—or expectancies about their capability to succeed on academic tasks—have been consistently found to outperform their less academically self-efficacious peers (Bandura 1997; Elliot & Dweck, 2005). Of course, this is not to say that actual abilities do not matter. Aptitude tests and past academic performance are still the best predictors of future academic grades, with competence beliefs adding incrementally to the prediction (Marsh, 1990; Valentine, DuBois, & Cooper, 2004). In other words, given equal levels of academic ability, someone who is more confident in their abilities will be more likely to succeed than someone who doubts their abilities.
The picture is no different for emotional competence beliefs. Comparative multi-method studies (including those published in this Special Issue) continue to find only weak-to-moderate associations between self-reported emotional competencies and performance on external tests of emotional abilities, indicating that people are generally poor judges of their emotional skills (Brannick, Wahi, Arce, & Johnson, 2009; Lumley, Gustavson, Partridge, & Labouvie-Vief, 2005; MacCann, Wang, Matthews, & Roberts, 2010). Yet, just like with academic intelligence beliefs, emotional competence (EI) beliefs can have consequential effects on observable behavior. Individuals who believe that emotions are outside of their control tend to fear the experience of negative emotions, which they see as a nuisance or personal weakness; consequently, they either get overwhelmed by their emotions or attempt to avoid or suppress their feelings when under stress (Kappes & Schikowski, 2013; Tamir, John, Srivastava, & Gross, 2007). In contrast, individuals who believe that emotions are amenable to personal control tend to be more accepting of their negative emotions, which they view as a valuable source of information, and which motivates them to try to understand and work through their feelings. For these reasons, individuals high in emotional regulatory self-efficacy—or expectancies about one’s ability to succeed at regulating emotions—tend to be better adjusted socially and emotionally than their less emotionally self-efficacious peers (see Alessandri, Vecchione, & Caprara, 2015). Importantly, there is growing evidence that the predictive effects of self-reported emotional competencies are independent of, and additive to, the effects of externally measured emotional abilities (Davis & Humphrey, 2012c; Di Fabio & Saklofske, 2014a; Nightingale et al., 2013; Qualter, Gardner, Pope, Hutchinson, & Whiteley, 2012).
In short, two things are clear about competence self-perceptions from the extant research. First, self-report competence ratings do not necessarily reflect the same “objective” abilities that are captured via external methods of assessment, and so they should not be used in lieu of such. Second, self-report ratings provide unique information that can meaningfully add to our understanding and promotion of resilience and adaptive functioning, and so they ought to be included in both research and practice (ideally, in complement with other assessments). Accordingly, one of the ongoing tasks for emotional competence researchers has been to articulate an empirically supported and practically informative theoretical framework that integrates emotional competence self-perceptions with other relevant constructs, while validating them as discrete variables in their own right (Mikolajczak, 2009; Cherniss, 2010; Matthews, Zeidner, & Roberts, 2012; Petrides, 2010). The aim of this Special Issue is to further advance this important goal.
Scope of the Special Issue
This Special Issue brings together a group of leading world experts to weigh in on the following key questions: (a) What is it that self-report scales of emotional competencies do and do not measure? (b) What are their unique contributions relative to other constructs and methods of assessment? (c) And, most critically, what are their boundaries and limitations? Apart from tackling these overarching methodological and validity issues, the contributing articles also showcase the diversity of theories and concepts characterizing this field. The first three articles (Alessandri et al., 2015; Qualter et al., 2015; Salguero, Extremera, Cabello, & Fernández-Berrocal, 2015) approach the topic from the perspective of Bandura’s (1986, 1997) social-cognitive theory, drawing on the concept of self-efficacy in particular to understand the nature and functions of emotional competence self-perceptions. The other three articles (Anguiano-Carrasco, MacCann, Geiger, Seybert, & Roberts, 2015; Matthews et al., 2015; Siegling, Furnham, & Petrides, 2015) fall back on McCrae and Costa’s (2008) Five-Factor Model of personality as the basis for evaluating the validity of various emotion-related self-reports. Matthews et al. additionally bring in the concept of alexithymia—a clinically significant personality trait characterized by deficits (rather than competencies) in identifying, processing, and regulating emotions (Taylor, Bagby, & Parker, 1999). Importantly, all but one of the articles offer an explicit illustration of similarities and differences between subjective self-reports and objective tests of emotional abilities, which are often combined in the literature under a common umbrella term of EI.
The readers of this Special Issue are also introduced to a variety of existing and new measures for assessing self-reported emotional competencies. Alessandri et al. (2015) review research conducted with the Regulatory Emotional Self-Efficacy (RESE) scale (Caprara et al., 2008). Qualter et al. (2015) develop and validate a youth adaptation of the Emotional Self-Efficacy Scale (ESES; Kirk, Schutte, & Hine, 2008). Salguero et al. (2015) use the Schutte Emotional Intelligence Scale (SEIS; Schutte et al., 1998), Siegling et al. (2015) use the short and long forms of the Trait Emotional Intelligence Questionnaire (TEIQue; Petrides, 2009), and Matthews et al. (2015) use a variety of measures, including the TEIQue, the Wong and Law Emotional Intelligence Scale (WLEIS; Wong & Law, 2002), the Trait Meta-Mood Scale (TMMS; Salovey, Mayer, Goldman, Turvey, & Palfai, 1995), and the Twenty-Item Toronto Alexithymia Scale (TAS-20; Bagby, Taylor, & Parker, 1994). Last, Anguiano-Carrasco et al. (2015) introduce three different forms of a new Three-branch Emotional Intelligence Assessment. Other scales not represented in this issue but commonly used in the field include the Emotional Quotient Inventory (EQ-i) and its derivatives (Bar-On, 1997; Parker, Keefer, & Wood, 2011) and the Swinburne University Emotional Intelligence Test (SUEIT; Palmer & Stough, 2001).
It is also noteworthy that the collective work presented in this issue is truly international in scope, with contributors from Italy, Spain, the Netherlands, Germany, the United Kingdom, Israel, Australia, Canada, and the United States. Despite this rich diversity of perspectives, approaches, and geographic locales, there are several methodological and validity issues that cut across the contributing articles and other studies in this area. These themes are highlighted next.
Methodological Considerations
Response Biases
A long-standing reservation about using self-report rating scales is their susceptibility to systematic response biases unrelated to the target construct (Cronbach, 1946). One of the most common sources of response bias is socially desirable responding, characterized by a tendency to portray oneself in an overly positive light, exaggerating the positive and downplaying the negative self-attributes (Edwards, 1957). Whether intentional (faking) or not (dispositional), such responding results in elevated scale scores and inflated scale reliabilities (Paulhus, 1984). This issue is especially pronounced in high-stakes assessments (e.g., selection or placement decisions), when respondents are motivated to present themselves in the best light possible. Given the desirable nature of emotional competencies and the transparent content of the items, most existing self-report EI scales are susceptible to response faking under high-stakes conditions (Day & Carroll, 2008; Grubb & McDaniel, 2007).
Another common response style that can bias self-report assessments is acquiescent responding, characterized by a tendency to consistently agree with questionnaire items regardless of their content or wording (Jackson & Messick, 1965). Such indiscriminate “yea”-saying produces discrepant patterns of results for true-keyed and reverse-keyed items, which can significantly distort the psychometric properties of a scale (Soto, John, Gosling, & Potter, 2008). Factors that contribute to greater acquiescence include difficulties with item comprehension, limited attentional or cognitive resources, and lack of motivation, among others (Knowles & Condon, 1999; Krosnick & Presser, 2010). When respondents are unwilling or incapable of carefully completing the survey, they tend to pick answers that appear satisfactory but are, more or less, arbitrary.
Although there are numerous methods of controlling for response biases (Ellingson, Sackett, & Hough, 1999; Ten Berge, 1999), one way to circumvent this issue altogether is to administer the items in a forced-choice format instead of a rating scale (Brown & Maydeu-Olivares, 2013). The logic behind this alternative procedure is that respondents will be unable to fake “good” or acquiesce to everything when they are forced to choose between two or more equally desirable response options. In this issue, Anguiano-Carrasco et al. (2015) provide an informative review of the advantages and limitations of forced-choice assessments, and present a new innovative measure of perceived emotional competencies that capitalizes on this technique.
Age-Bound Effects
The issue of psychometric response quality becomes particularly salient in self-report assessments of pre-adult respondents (Borgers, de Leeuw, & Hox, 2000; Marsh, Debus, & Bornholt, 2005). Compared with self-reports of older adolescents and adults, self-reports of younger children and pre-adolescents are notorious for having higher incidence of incomplete, inconsistent, exaggerated, and acquiescent responding (Borgers et al., 2000; Keefer, Holden, & Parker, 2013; Soto et al., 2008). Much of this effect can be attributed to the limited language skills and still-developing cognitive functions at younger ages (Borgers et al., 2000; Davis-Kean & Sandler, 2001). As a result, items that contain unfamiliar words, complex syntax, double negations (i.e., having to disagree with a statement to endorse the trait), or rating scales with too many response categories all pose a challenge for younger respondents, increasing the likelihood of careless, arbitrary, or otherwise invalid responding (Borgers, Hox, & Siikkel, 2004). Simplifying item wording and improving the overall readability of the scale is a recommended strategy for mitigating this problem (Parker, Eastabrook, Keefer, & Wood, 2010; Scott, 1997).
Apart from age-related changes in children’s capacity to self-report accurately and reliably, assessment of emotional competencies in youth can be further confounded by developmental changes in the structure of the target construct itself (Denham, Wyatt, Bassett, Echeverria, & Knox, 2009; Obradović, van Dulmen, Yates, Carlson, & Egeland, 2006). As young people mature cognitively and socially, their emotional competencies become increasingly more complex, nuanced, and contextualized (Lane & Garfield, 2005; Saarni, 1999). Youth’s competence self-perceptions similarly become more refined with age, diverging from a global sense of being “smart” in general into distinct self-concepts in specific performance domains (Harter, 2012). As a result of these changes, the magnitude of inter-correlations among the various self-concept domains tends to be considerably lower for adult respondents than for children or adolescents (Marsh, Craven, & Debus, 1998). A similar age differentiation effect has been recently documented in a longitudinal study of EI self-perceptions (Keefer et al., 2013), where the amount of shared variance in self-reports of different EI abilities decreased from 26% at age 10 to 11 years to 5% by age 16 to 17. In this issue, Qualter et al. (2015) replicate the age differentiation effect for the sub-domains of emotional self-efficacy, and further illustrate a successful adaptation of the adult Emotional Self-Efficacy Scale for use with young adolescents.
The methodological caveats associated with self-report assessments impose certain restrictions on the uses of emotional competence scales. First, scales that do not have a forced-choice format or alternative means of guarding against faking are not appropriate for high-stakes assessment purposes (e.g., diagnosis, selection, placement, or promotion decisions). However, they can and have been used successfully as formative assessments (e.g., in screening, personal coaching, program evaluation, and basic and applied research). Second, scales that have not been explicitly validated for use with pre-adult populations should not be used with respondents younger than 16 years old. To be suitable for use with children and adolescents, the wording and format of the items need to reflect the reading level and simplicity appropriate for the respondents’ age. Similar concerns would apply to respondents with cognitive or reading difficulties (e.g., due to inattention, learning disorder, intellectual disability, low literacy, or advanced age). As a rule, personality inventories should have a fifth- to sixth-grade reading level to be suitable for use with general adult population (Schinka & Borum, 1994). Last, researchers are encouraged to screen for the presence of response biases in their questionnaire data (e.g., social desirability, acquiescence, random responding), and to implement appropriate statistical controls prior to testing the substantive hypotheses.
Construct Validity Considerations
What’s in a Name?
One of the major obstacles to the advancement of a unified theory around emotional competence self-reports has been the proliferation of multiple definitions for the underlying constructs (Cherniss, 2010; Zeidner, Roberts, & Matthews, 2008). In this Special Issue alone, there are half a dozen terms used in reference to self-report scales of emotional competencies, including “emotional self-efficacy,” “perceived EI,” “typical-performance EI,” “trait EI,” and “mixed models of EI.” More often than not, these terms are used interchangeably, but if a distinction is made, it usually resides in the content coverage of a particular scale. If the scale items are written in the form of “I can” self-statements about concrete emotional competencies and abilities (e.g., “I can easily describe my feelings”), the scale can be said to assess emotional self-efficacy—or beliefs about one’s ability to successfully perform specific emotional tasks. Such self-efficacy scales may also be referred to as measures of perceived EI or typical-performance EI, if the abilities they measure map directly onto the four branches of Mayer et al.’s (2008) seminal ability EI model (Emotion Perception, Facilitation, Understanding, and Management). Some scales may include items that, in addition to ability-based self-efficacy beliefs, also tap self-evaluations of other emotion-related characteristics, such as values (e.g., “Emotions are a major source of meaning in my life”), global dispositions (e.g., “I am a happy person”), and personality traits (e.g., “I do not get upset easily”). Such scales are commonly referred to as measures of trait EI or mixed models of EI, due to their overlap with other emotion-related aspects of personality.
Needless to say, the scale labels matter, as they invariably guide our interpretation of what the scale scores purportedly mean. In this regard, many have criticized the use of the label “intelligence” in reference to EI self-reports, arguing that it would be inappropriate to infer individuals’ actual EI from their self-reports, much like it would be inappropriate to accept self-estimates of cognitive abilities as valid proxies of intelligence (Zeidner et al., 2008). Particular terminology may also direct the research paradigms we adopt. Not coincidentally, studies using scales labeled as “emotional self-efficacy” have been conducted primarily under the tenets of the social-cognitive theory, whereas studies using scales labeled as “trait EI” have been connected almost exclusively to the Five-Factor Model of personality, with little mutual cross-referencing despite their shared item content (but see Caprara, Vecchione, Barbaranelli, & Alessandri, 2013). Conversely, the one common paradigm that is not explicitly captured by the scales’ labels, yet that is implicit in all of them—the self-concept theory—is conspicuously missing from the bulk of empirical research. I see this as a major oversight, for the long-standing tradition of scholarship on the self (James, 1890/1963) can greatly facilitate the integration of different strands of research on emotional competence self-reports, as well as generate novel hypotheses and insights about their nature and functions.
Self-Concept as an Integrative Framework
Broadly defined, self-concept represents an individual’s perception of his or her own abilities, attributes, attitudes, and evaluative judgments about the self in a particular domain, be it academic, social, physical, or any other arena of functioning (Marsh, 2007; Shavelson, Hubner, & Stanton, 1976). As a social-cognitive construction, self-concept is inherently subjective, in that it captures
how [individuals] construe themselves—what attributes they think they possess, what roles they presume they are expected to play, what they believe they are capable of, how they view they fare in comparison with others, and how they judge they are viewed by others. (Bong & Skaalvik, 2003, pp. 1-2)
Because the same emphasis on subjective self-construal is implied in all self-report scales of emotional competencies, they can be said to measure a person’s self-concept in emotion-related domains, or more simply an individual’s “EI self-concept” (or “emotional self-concept,” if one wishes to refrain from using the term “intelligence”). Under this common label, the subtler distinctions among the scales can be made in the same way that self-concept has been differentiated from the narrower construct of self-efficacy or from broader personality traits.
Self-concept versus self-efficacy
The traditional distinction between self-concept and self-efficacy beliefs has been based on their degree of specificity (Bong & Skaalvik, 2003). For example, a self-efficacy scale would assess beliefs about one’s capabilities in relation to specific tasks or activities, such as distinguishing between different facial expressions and matching them to correct emotion labels, whereas a self-concept scale would assess a more generalized self-evaluation of one’s ability to understand others’ emotions. However, this distinction has been blurred by the introduction of the General Self-Efficacy Scale (Schwarzer & Born, 1997), which measures a global sense of confidence in one’s ability to deal effectively with a variety of challenging situations. Another commonly cited distinction is that, unlike self-efficacy beliefs, self-concepts encompass the so-called “hot” cognitions (attitudes, motives, value judgments) in addition to pure competence-related beliefs (Pajares & Schunk, 2005). For example, a self-concept scale might assess the degree of personal importance placed on being able to understand others’ emotions, the strength of motivation to engage in this behavior, or the amount of pride derived from one’s ability to do so well. In this regard, trait EI and mixed-model EI measures, such as the TEIQue and EQ-i, are highly congruent with this broader self-concept definition.
Location in personality space
Contemporary integrative theories of personality distinguish between at least two qualitatively discrete yet inter-connected levels, or layers, of personality (McAdams & Pals, 2006; McCrae & Costa, 2008; Nilsson, 2014). At the higher-order level, or the core layer, are the most basic and generalized dispositions, such as temperaments and the Big Five traits (Neuroticism, Extraversion, Conscientiousness, Openness, and Agreeableness). At the lower-order level, or the surface layer, are the dynamic social-cognitive constructs, such as motives, goals, values, competence beliefs, and attitudes. Core dispositions are regarded to be more strongly grounded in the genetic and neurobiological differences that are already evident in the first few years of life, whereas surface characteristics are thought of as more contextualized and motivated adaptations that begin to take shape in the elementary school years and become integral features of personality by middle childhood (McAdams & Olson, 2010; McCrae & Costa 2008). The two levels/layers of personality are systematically and reciprocally inter-related, in that certain core traits can engender specific goals and attitudes, whereas the pursuit of certain goals can bring about particular changes in dispositional tendencies (Bleidorn et al., 2010; Caprara, Vecchione et al., 2013; K. M. Sheldon, Cheng, & Hilpert, 2011). Although the two tiers of personality overlap in meaningful and logical ways, they are thought to represent distinct and irreducible sources of information about an individual, and their effects on behavior tend to be both shared/mediated and independent/unique (Albuquerque, de Lima, Matos, & Figueiredo, 2013; B. W. Roberts, O’Donnell, & Robins, 2004; K. M. Sheldon & Hoon, 2007).
Within this multi-level personality space, both EI self-concepts (De Raad, 2005; Petrides, Pita, & Kokkinaki, 2007) and other domains of self-concept (Marsh, Trautwein, Lüdtke, Köller, & Baumert, 2006) have been empirically situated at the lower levels of the taxonomy, establishing their status as surface-level characteristics. Because the two personality levels/layers are inter-connected, there is an expected moderate degree of domain-congruent overlap between EI self-concepts and the core Big Five dimensions (Brackett & Mayer, 2003; Parker et al., 2011; Pérez-González & Sanchez-Ruiz, 2014). For example, in this issue Siegling et al. (2015) report correlations for the TEIQue in the magnitude of .60s with Neuroticism (negative), .50s with Extraversion, .40s with Conscientiousness, and .20s with Openness and Agreeableness, aggregated across several different samples and measures of the Big Five. Behavioral genetic evidence suggests that these phenotypic associations are attributable primarily to correlated genetic factors, with no significant contributions from the shared environment (Vernon, Villani, Schermer, & Petrides, 2008).
Although theoretically meaningful, such substantive construct overlap underscores the importance of establishing incremental validity of EI self-concepts vis-à-vis basic personality. More and more recent studies have been testing for incremental validity, finding that self-report EI scales indeed add significant variance over and above the Big Five in predicting academic performance (Di Fabio & Palazzeschi, 2009; Ferrando et al., 2011; Sanchez-Ruiz, Mavroveli, & Poullis, 2013), job performance (O’Boyle, Humphrey, Pollack, Hawver, & Story, 2011), internalizing and externalizing problems (Andrei, Mancini, Trombini, Baldaro, & Russo, 2014; Davis & Humphrey, 2012b; Gardner & Qualter, 2010; Parker et al., 2011; Russo et al., 2012), and resilience and well-being (Di Fabio & Saklofske, 2014b; Petrides, Pérez-González, & Furnham, 2007). In this issue, Anguiano-Carrasco et al. (2015) and Matthews et al. (2015) also perform incremental validity tests, but arrive at mixed results. The extent of incremental prediction may very well depend on the particular scale used, so consumers of research in this area are advised to pay attention to the issue of incremental validity (Zeidner et al., 2008).
If emotional competence self-reports share the construct characteristics of self-concepts (as I propose here), then the same properties that characterize self-concepts in other domains should also apply to EI self-concepts, including their regulatory function, multi-dimensional structure, and gendered nature.
Regulatory Function
Social-cognitive perspective
The central premise of the self-concept theory is that self-concept is not merely a snapshot of one’s abilities and attributes, but it is an active structure that organizes and gives meaning to past and current experiences, provides goals and standards for behavior, and motivates future choices and actions (Harter, 2012; Markus & Wurf, 1987). As mentioned earlier, competence-related cognitions can act as powerful regulators of behavior, by promoting a stronger sense of agency over performance outcomes and a more proactive approach to dealing with challenges (Bandura, 1997; Dweck, 1999; Elliot & Dweck, 2005). The causal role of positive self-concepts in influencing subsequent performance, over and above prior performance, has been demonstrated across multiple performance domains (Marsh & Craven, 2006; Marsh & Martin, 2011; Valentine et al., 2004).
In this issue, Alessandri et al. (2015) provide an excellent review of how the social-cognitive paradigm has been applied to the specific domain of emotion regulation in predicting various well-being and prosocial outcomes. Drawing on the same principles, Salguero et al. (2015) put forth and test an intriguing proposition that having a positive EI self-concept may in fact be a necessary condition for adaptive functioning, in addition to possessing high emotional abilities. They found that given equally high levels of objectively measured EI, women who had lower EI self-concepts experienced significantly higher levels of depression than women who had higher EI self-concepts. In explaining their results, Salguero et al. conclude that possessing sufficient emotional skills may not be enough to motivate adaptive behavior; one must also feel sufficiently competent to put those skills into action. The hypothesized causal role of EI self-perceptions in influencing emotionally intelligent behavior is supported by recent attempts to manipulate individuals’ perceived EI experimentally (Schutte & Malouff, 2012). In these experiments, priming of self-concepts related to successful emotional competence lead to better performance on an objective test of EI abilities than did priming of general emotion knowledge, task motivation, or self-directed attention. Perhaps not surprisingly, EI self-reports have emerged as more robust correlates of socioemotional functioning than objective EI abilities (Davis & Humphrey, 2012c; Gohm, Corser, & Dalsky, 2005; Malouff, Schutte, & Thorsteinsson, 2014; Martins, Ramalho, & Morin, 2010; Zeidner, Matthews, & Roberts, 2012)—a finding that is not easily explained by criterion contamination or common method variance (Williams, Daley, Burnside, & Hammond-Rowley, 2010).
Transactional coping perspective
What are the specific mechanisms that connect EI self-concepts to adaptive outcomes? Much of the existing research has approached this question from the transactional coping paradigm (Lazarus & Folkman, 1984), which holds that, when faced with a stressful situation, individuals purposefully engage in various coping behaviors that can either offset or aggravate the adverse effects of stress on well-being. Proactive coping strategies aimed at resolving the problem or regulating the emotional distress are generally considered to be more adaptive compared with disengaging strategies, such as avoidance, suppression, or self-blame (Austenfeld & Stanton, 2004; Endler & Parker, 1994). Whether individuals choose to tackle, tolerate, or escape perceived stressors depends on their appraisals of coping resources as either sufficient or lacking, and of the situation as either manageable or outside of personal control (Lazarus & Folkman, 1984).
Emotional competencies can be construed as an important personal resource in the context of coping (Matthews & Zeidner, 2000). Accordingly, individuals who perceive their emotional competencies to be high tend to appraise stressful situations as more manageable (Fernández-Berrocal & Extremera, 2006; Mikolajczak & Luminet, 2008), exhibit healthier physiological stress response (Laborde, Brüll, Weber, & Anders, 2011; Mikolajczak, Roy, Luminet, Fillée, & de Timary, 2007), and engage in more proactive and fewer maladaptive coping practices, which in turn modulate the deleterious effects of stress on performance and well-being (Downey, Johnston, Hansen, Birney, & Stough, 2010; Mikolajczak, Petrides, & Hurry, 2009; Saklofske, Austin, Galloway, & Davidson, 2007). Thus, subjective beliefs in one’s emotional capacities appear to enhance the effectiveness of coping efforts, supporting the notion that “adaptive coping might be conceptualized as emotional intelligence in action” (Matthews & Zeidner, 2000, p. 460).
In this issue, Matthews et al. (2015) explore the coping function of EI self-concepts in the context of performance on two emotion-related laboratory tasks. As is typical of this literature, they found that EI self-reports failed to predict actual performance on emotion-ability tasks. However, higher perceived EI was associated with reduced subjective distress experienced during the experiment, and with less maladaptive coping via avoidance and emotional disengagement. In general, emotional competence self-reports tend to have much stronger and more consistent relationships with outcomes related specifically to emotional well-being (Martins et al., 2010) and resilience under stress (Davis, 2013) than with more compounded outcomes such as academic achievement (Perera & DiGiacomo, 2013) or healthy lifestyles (Keefer, Parker, & Saklofske, 2009). With respect to these broader outcomes, the effects of EI self-concepts tend to be most pronounced under high-stress conditions, such as during exam times (Goldman, Kraemer, & Salovey, 1996) or major educational transitions (Parker, Summerfeldt, Hogan, & Majeski, 2004), or when the behavior in question is directed toward alleviating distress (e.g., smoking or drinking to improve mood; Keefer et al., 2009). This moderated effect is consistent with the hypothesized coping function of EI self-perceptions; when the level of stress is tolerable, or when the behavior is motivated by reasons other than stress management, emotional competence beliefs do not seem to play as much of a role.
Multi-Dimensional Structure
The differential predictive scope of emotional competence self-reports across different areas of functioning is to be expected given the highly differentiated multi-dimensional structure of self-concepts. Self-concepts in specific domains (e.g., academic, physical, social, emotional) are only loosely inter-correlated (except in childhood), and different self-concepts tend to relate to various external criteria in highly domain-specific ways, with stronger relationships observed for outcomes most logically related to beliefs in that particular domain (Marsh & Craven, 2006). Thus, beliefs about one’s competence in math are better predictors of math performance than beliefs about one’s emotional competencies, with the latter predicting math performance only to the extent that they facilitate better coping with math-related anxiety (Tariq, Qualter, Roberts, Appleby, & Barnes, 2013). Not surprisingly, consideration of multiple specific self-concepts tends to offer greater explanatory utility for compounded life outcomes than consideration of a single global self-concept such as self-esteem (Marsh & O’Mara, 2008). Thus, contemporary self-concept theorists strongly advocate a differentiated multi-dimensional perspective on the internal structure of self-views (Marsh, 2007; Marsh & Craven, 2006).
The structure of emotional competence self-perceptions is also multi-dimensional, with most models differentiating between perceiving, understanding, and regulating emotions as the core competence domains. Some models (e.g., Brasseur, Grégoire, Bourdu, & Mikolajczak, 2013) additionally incorporate the intrapersonal (own emotions) versus interpersonal (others’ emotions) dimension, whereas others (e.g., Caprara, Di Giunta, Pastorelli, & Eisenberg, 2013) subdivide the competencies by valence (positive vs. negative emotions) and discrete emotions (anger vs. sadness). Mirroring the broader self-concept literature, self-concepts in different EI domains are only moderately inter-correlated, and although they do share some variance in common, the prediction of various criteria tends to be significantly enhanced by including multiple individual dimensions rather than a global composite score (Gardner & Qualter, 2010; Parker et al., 2011). Moreover, domain-specific EI self-concepts have been associated with distinct coping behaviors: emotion management—with less emotional disengagement; emotional adaptability—with more task-focused coping; and interpersonal competencies—with greater social support seeking (Austin, Saklofske, & Mastoras, 2010; Saklofske, Austin, Mastoras, Beaton, & Osborne, 2012). Even more telling is the finding that, despite having comparable global EI self-concepts, individuals can have very different profiles of specific strengths and weaknesses (e.g., high emotion management self-concept but low interpersonal self-concept, or vice versa), and these different self-concept profiles have differential implications for long-term behavioral outcomes (Keefer, Parker, & Wood, 2012).
Taken together, the existing evidence underscores the importance of adopting a multi-dimensional approach to the assessment and study of emotional competence self-reports, as it affords greater overall predictive utility, more nuanced exploration of substantive relationships, and better differentiation between individuals with qualitatively different self-concept profiles. Conversely, failure to account for the multi-dimensionality of complex constructs can result in ambiguous or conflicting validity results and inaccurate decisions for individual test-takers (McGrath, 2005). Accordingly, four of the articles published in this Special Issue (Alessandri et al., 2015; Anguiano-Carrasco et al., 2015; Matthews et al., 2015; Qualter et al., 2015) take strategic advantage of multi-dimensionality with the explicit purpose of advancing either the assessment or the validity base of emotional competence self-reports.
Gendered Nature
The value of a multi-dimensional paradigm is further illustrated by the observation that null or inconsistent gender differences in global EI self-concepts often mask reliable and counter-balancing gender differences in specific EI self-concept domains (Siegling, Vesely, & Saklofske, 2012). Women consistently report higher EI self-concepts in the areas of emotional attentiveness, expressivity, and empathy, whereas men hold higher EI self-concepts in the areas of emotion regulation, stress tolerance, and self-control (Fernández-Berrocal & Extremera, 2008; McIntyre, 2010; Parker et al., 2011; Petrides & Furnham, 2000). These gender differences are evident in self-reports of children as young as 10 years old (Keefer et al., 2013). The same trend applies to the broader self-concept literature, where small gender differences in global self-esteem conceal larger and contrasting gender differences in specific self-concepts, favoring girls in the verbal and arts domains, and boys in the math and sports domains (Jacobs, Lanza, Osgood, Eccles, & Wigfield, 2002; Marsh & Craven, 2006).
In explaining these robust gender differences, many developmental psychologists highlight the socializing role of gendered stereotypes and expectations (by parents, peers, teachers, and society at large) in shaping and reinforcing children’s dispositions and self-perceptions (Bussey & Bandura, 1999; Jacobs, Bleeker, & Constantino, 2003; Sanchez-Nunez, Fernandez-Berrocal, Montanes, & Latorre, 2008). As children get older, these external stereotypes become internalized into their own sense of self and identity (Harter, 2012). A recent study (Siegling et al., 2012) has mapped specific EI self-concepts onto aspects of gendered identity, showing that female-typed EI self-concepts were uniquely associated with stereotypically female traits (e.g., compassion, nurturance, communion), whereas male-typed EI self-concepts were uniquely associated with stereotypically male traits (e.g., assertiveness, competitiveness, agency). These gendered associations remained significant after controlling for differences due to biological sex, supporting the social-cognitive nature of EI self-concepts. Furthermore, gender differences in perceived emotional competencies do not always correspond to gender differences in actual emotional abilities (Sanchez-Nunez et al., 2008; Siegling et al., 2012), yet they consistently translate into gender-typed coping behaviors. Congruent with their perceived EI strengths, women tend to rely more on interpersonally oriented coping (e.g., emotional disclosure, support seeking), whereas men tend to engage more in externally oriented coping (e.g., problem-solving, distraction; Brody & Hall, 2000).
If emotional competencies carry different social weights for men and women, could they also be associated with other variables in gender-specific ways? Although gender differences in emotional competence self-reports are routinely assessed in the literature, research on their substantive implications is only starting to appear. In this issue, the question of functional invariance by gender is explored by Siegling et al. (2015) and Salguero et al. (2015), albeit only for global EI self-concept rather than for its specific domains. Siegling et al. found that the associations between perceived EI and the Big Five traits were invariant across gender, suggesting that the two personality structures are linked in gender-congruent ways. Salguero et al. discovered that high perceived EI was doubly consequential for women: Not only was it directly associated with less depressive affect (a finding that also applies to men), but it also activated similar protective effects of EI abilities (which are otherwise non-significant for women). The next step in this line of inquiry is to test whether the domain-specific EI self-concepts are more consequential for one gender or the other, and whether those effects are gender-congruent or counter-stereotypical. An intriguing proposition in this respect is that an androgynous EI self-concept, that is, one characterized by endorsement of both stereotypically masculine and stereotypically feminine EI competencies, may be maximally adaptive for both men and women (Guastello & Guastello, 2003).
So far, I have used the self-concept framework as a way of synthesizing past and current validity research on emotional competence self-reports. However, the added value of conceptualizing emotional self-reports as emotional (or EI) self-concepts lies in its potential to shape future theoretical thought and research in this area. By drawing on the well-established self-concept literature, one can answer outstanding questions and derive novel hypotheses about the nature and practical implications of emotional self-perceptions.
Practical Implications, Outstanding Questions, Future Directions
Discrepancies With Emotional Abilities
One of the most obvious practical implications of emotional competence self-reports being distinct from emotional abilities is that both need to be considered when devising socioemotional education programs, for enhancing emotional abilities alone may not necessarily result in concomitant increases in EI self-concepts, and vice versa. The challenge associated with such a multi-method strategy is what to make of the discrepancies between individuals’ self-reported and externally measured abilities. In this issue, Matthews et al. (2015) offer several compelling arguments as to why accurate self-assessment of emotional abilities may be difficult to achieve, including lack of specific benchmarks for what constitutes emotionally intelligent behavior, few opportunities to receive explicit feedback on one’s socioemotional skills in everyday encounters, and a self-enhancement bias that may motivate people to selectively seek positive and disregard negative feedback about themselves. Ironically, it is the people with the lowest level of emotional ability who tend to overestimate their emotional competence the most (O. J. Sheldon, Dunning, & Ames, 2014). This is thought to occur because individuals who lack skills in a particular domain also lack the capacity to accurately judge their strengths and weaknesses in that domain, suggesting that accurate self-perception may be contingent on a certain requisite level of ability (Ehrlinger, Johnson, Banner, Dunning, & Kruger, 2008).
In the absence of clear standards, direct feedback, or sufficient self-insight, what information do people use to construct their emotional competence self-views? Although there are no studies on this topic in the emotional competence literature, research on the processes involved in self-concept formation can offer some specific insights in this regard. As social-cognitive constructions, self-concepts are influenced by multiple frames of reference, external and internal (Marsh, 2007). Externally, individuals may derive important information about themselves from their impressions of how they are perceived and judged by others, known as reflected self-appraisals (e.g., I must be competent because my parents think so; Harter, 2012; Pfeifer et al., 2009). Social comparison is another powerful source of self-evaluative information, whereby individuals compare themselves with others in their environment to estimate how competent they are (e.g., I must be incompetent because my classmates are better than I am; Festinger, 1954; Marsh & Hau, 2003). Internally, individuals may engage in temporal comparisons, evaluating their current level of competence in relation to their past performance (e.g., I must be competent because I am much better now than I was before) or their ideal self-standards (e.g., I must be incompetent because I am not as good as I should be; Harter, 2012; Higgins, 1987). Individuals may also engage in dimensional comparisons, whereby competence in one domain is evaluated on the basis of competence in other domains that are perceived to be incompatible (e.g., I am very good at math therefore I cannot be as good at arts; Möller, Pohlmann, Köller, & Marsh, 2009). The notion of dimensional comparisons further stresses the need to use multi-dimensional self-concept assessments, as a global composite score cannot account for these complex cross-domain dynamics (Marsh & Ayotte, 2003).
A fruitful avenue for future research would be to test the extent to which emotional competence self-reports are subject to the various internal and external frame-of-reference effects, and to delineate the conditions under which certain frames of reference are more salient than others. It would also be important to know the implications of self-perception biases that arise from different comparison processes. To date, only a handful of studies have examined this question, finding that grossly over- or underestimating one’s emotional competence may undermine adaptive functioning (Salguero et al., 2015; O. J. Sheldon et al., 2014). However, no studies have linked these profiles to the specific sources of discrepancies; it may be that some sources of emotional over-confidence are problematic, whereas others promote resilience and adaptation. Answers to these questions would inform the most appropriate choice of intervention strategies for individuals with maladaptive self-perception profiles. The primary intervention goal for these individuals would be to identify, de-construct, and re-structure the specific meta-cognitions that underlie their destructive emotional self-views. Until the various frame-of-reference effects are better understood, however, promoting high EI self-concepts indiscriminately may be premature.
Stability and Change Over Time
Another set of questions that has direct implications for socioemotional programs concerns the dynamics of stability and change in emotional competence self-perceptions over time. Knowing when EI self-concepts are more or less malleable can suggest windows of opportunity during which interventions might have greater impact. Likewise, understanding the normative increases and decreases in EI self-concepts can inform more appropriate expectations of what constitutes typical level of functioning at different life periods. These questions require multi-wave longitudinal research designs, which remain few and far between. One notable exception is a 6-year population study of Canadian youth that tracked changes in EI self-concepts from late childhood (age 10-11) to adolescence (age 16-17; Keefer et al., 2013). This study found that individual differences in EI self-concepts were most malleable during the immediate transition from late childhood to adolescence (ages 10-13), but became increasingly more stable with age, paralleling the general stabilization of personality structures with maturation and life experience (B. W. Roberts & DelVecchio, 2000). However, changes in mean levels were not continuously positive during this developmental period, as might be expected based on children’s increasing EI abilities. Rather, EI self-concepts declined, on average, between childhood and early adolescence.
To understand this seemingly contradictory finding, it is once again helpful to turn to the self-concept literature, as similar declines in self-confidence at the transition to adolescence have been observed in other domains as well (Jacobs et al., 2002; Marsh, 1989). Developmental self-concept research suggests that this downward trend is in fact normative, reflecting a re-calibration of frames of reference prompted by new roles, expectations, and experiences of adolescent life (Harter, 2012; Jacobs et al., 2003). From this perspective, decreases in self-concepts at the transition to adolescence represent a cognitive shift from simplistic and overly positive self-representations of childhood to contextually sensitive and more realistic self-construals of emerging adulthood. In this issue, Alessandri et al. (2015) review additional longitudinal evidence on the temporal dynamics of regulatory emotional self-efficacy beliefs, revealing differential trajectories of change for the domains of positive and negative emotions. Continuing future research in this vein is a clear priority for the field, as knowing when, how, and why individuals’ EI self-concepts are likely to change, and the outcomes those changes are likely to bring, can ultimately inform more effective educational and intervention strategies.
Summary
Educational programs aimed at enhancing individuals’ emotional competencies have become a fast-growing enterprise across all levels of the education hierarchy. As the evidence base for emotional competencies and their applications grows, so is the need to critically re-appraise the assessment tools on which this evidence is based. This Special Issue focuses specifically on self-report scales as one of the most commonly used yet also contentious methods for assessing emotional competencies. In the six articles comprising this issue, leading scholars in the field present new evidence, state-of-the-art measures, and thought-provoking commentary on the validity, applied utility, and limitations of emotional competence self-reports. Here, I have integrated their contributions within the broader self-concept framework, as a way of contextualizing the existing knowledge and generating novel directions for future research. Collectively, the articles reinforce the idea that although self-reports of emotional competencies do not assess actual emotional abilities, they do provide unique information about a person’s self-identity that can add incrementally to the prediction of relevant outcomes. In addition to illustrating what these relevant outcomes might be, the articles also highlight a number of methodological caveats associated with the use of self-report assessments under various conditions. It is my hope that this Special Issue will promote more informed interpretations and applications of emotional competence self-reports, as well as stimulate further discussion and research into their methods and meaning. I thank the authors for their high-quality and innovative submissions, and Dr. Donald Saklofske for his direction and support in bringing this corpus of work to the JPA readership.
Footnotes
Acknowledgements
The author thanks Drs. James Parker and Ronald Holden for their feedback on earlier versions of this manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Work on this article was supported by a research fellowship 756-2012-0552 from the Social Sciences and Humanities Research Council of Canada.