www.ejop.org Emotional Intelligence and Psychological Distress: Testing the Mediatory Role of Affectivity

The study tested the extent to which positive and negative affect at work mediate emotional intelligence effects on psychological distress. Participants were 200 middle-level managers who completed the Wong and Law Emotional Intelligence Scale, 20-item affectivity scale, and a measure of psychological distress. Results using covariance based structural equation modeling indicated that only negative affect fully mediated the relationship between emotional intelligence and psychological distress. Furthermore, the direct effect of emotional intelligence was stronger for positive affect as compared to its influence on negative affect. Finally, negative affect had a significant direct influence on psychological distress, whereas the relationship between positive affect and psychological distress was insignificant.

Emotional Intelligence Test: Schutte et al., 1998;Trait Emotional Intelligence Questionnaire: Petrides, Pérez-Gonzalez, & Furnham, 2007).The plethora and diversity of EI models and measures in the field give rise to the need for a way to classify them.Currently there are mainly two approaches to conceptualizing and measuring EI: 'Ability EI' and 'Trait EI' (Petrides & Furnham, 2001).Ability models have been identified as those that define EI as 'intelligence' in the traditional sense (e.g., Mayer & Salovey, 1997;Salovey & Mayer, 1990).Proponents of ability EI conceive EI as an ability to process the information contained in emotions to determine the meaning of emotions and their connections to one another; and to use emotional information as the basis for thought and decision making (e.g., Salovey & Mayer, 1990).Ability EI is measured through performance tests with correct and incorrect answers, for example, Mayer-Salovey-Caruso emotional Intelligence Test (MSCEIT) (Mayer et al., 2003).On the other hand, trait EI is defined as "constellation of behavioral dispositions and self perceptions concerning one's ability to recognize, process, and utilize emotion laden information" (Petrides & Furnham, 2001, p. 426).Basically, the trait EI construct encompasses two kinds of variance: variance covered by personality dimensions (Big Five & Giant Three) and variance that lies outside these dimensions (Petrides, Pita, & Kokkinaki, 2007).In contrast to ability EI, the trait EI is measured via self-reports.In short, the distinction between ability EI (cognitiveemotional ability) and trait EI (trait emotional self-efficacy) is mainly based on the method of measurement (self-report versus performance based) and should not be confused with Mayer, Salovey, and Caruso's (2000) distinction between 'ability EI models' and 'Mixed EI models' .In contrast to Petrides, and Furnham's (2001) distinction, Mayer, Salovey, & Caruso (2000) conceive mixed models of EI (e.g., Bar-On, 1997;Goleman, 1995) as those which "mixes" cognitive abilities with other characteristics.In line with Petrides and Furnham's (2001) approach (trait EI vs. ability EI), the present paper seeks to find the relationships between trait EI, psychological distress, and affectivity.

EI and Affectivity
Affectivity is a general tendency to experience a particular mood or to react to objects in a particular way (Lazarus, 1993).Affectivity is split up into two distinct dimensions: Positive affect (PA) and negative affect (NA).PA refers to tendency of experiencing good feelings such as enthusiastic, active, and alert, whereas, NA refers to having bad feelings such as anxiety and disgust (Watson, Clark, & Tellegen, 1988).Research has documented that PA and NA are two independent separate constructs (Morris & Feldman, 1996;Watson & Clark, 1984;Watson, Clark, & Tellegen, 1988).

23
According to Lopes et al. (2006), people are usually motivated to seek pleasant feelings and avoid unpleasant ones.The EI abilities can help people to identify and interpret cues that inform self-regulatory action to nurture positive affect and avoid negative affect (Lopes et al., 2006;Mayer & Salovey, 1997).Various researches have documented a positive relationship between EI and PA and negative relationship between EI and NA (e.g., Kafetsios & Zampetakis, 2008;Lopes et al., 2006;Sevdalis, Petrides, & Harvey, 2007).Thus, it is expected that trait EI will be positively associated with PA and inversely to NA at work.Hypothesis 1a: There is a positive relationship between EI and PA.
Hypothesis 1b: There is a negative relationship between EI and NA.

Psychological Distress
Most of theorists look at a broader definition of psychological health containing two factors: Psychological well-being (positive mental health states like life satisfaction) and psychological distress (negative mental health states like anxiety and depression) (Massée et al.,1998;Veit & Ware, 1983;Wilkinson & Walford, 1998).
The theoretical structure of the construct of psychological distress and its operationalization has been subjected to extensive research (e.g., Goldberg, 1978;Massée et al., 1998;Ridner, 2004).Ridner (2004) after extensive review of literature pointed out that, psychological distress is often embedded in the context of stress, strain, and distress and is seldom treated as a distinct concept.She differentiated between strain, stress, distress, and psychological distress and defined the construct of psychological distress as, "the unique discomforting, emotional state experienced by an individual to response to a specific stressor or demand that results in harm, either temporary or permanent, to the person."(p.539).Psychological distress is a complex and multidimensional construct (Massée et al., 1998;Ridner, 2004;Veit & Ware, 1983, Wilkinson & Walford, 1998).In its simplest form psychological distress is viewed as a construct that represents aspects of negative functioning.For instance, according to Massée et al., (1998), psychological distress is usually operationalized by measures of self-depreciation, irritability, anxiety, depression, and social disengagement.

Affectivity and Psychological Distress
It has been argued that affectivity may affect perceived level of psychological health by influencing perceptions of self or environment (Oliver & Brough, 2002).
Individuals with high levels of NA are more likely to appraise events as threatening (Gallagher, 1990), tend to view their environment more negatively (Spector et al., 2000), report more negative interpretations of trivial problems (Watson & Pennebaker, 1989), and encode more negative information (Larsen, 1992).This highly negative view of their objective environment lead them to rate stressors as highly aversive experiences, which in turn leads to poorer psychological well-being outcomes (e.g., anxiety & depression) (Oliver & Brough, 2002;Spector et al., 2000).Thus, it is expected that PA will be negatively associated with psychological distress and NA will be positively associated with psychological distress.Hypothesis 2a: There is a negative relationship between PA and psychological distress.
Hypothesis 2b: There is a positive relationship between NA and psychological distress.

EI and Psychological Distress
In their review of psychological well-being research, Diener and colleagues (Diener et al., 1999) assert that, "personality is one of the strongest and most consistent predictors of subjective well-being" (p.279).In this nexus, there are several reasons why trait EI may influence psychological distress.Research suggests that EI abilities and traits contribute to good physical and psychological health (Salovey et al., 1999;Salovey et al., 2000;Tsaousis, & Nikolaou, 2005).Emotionally intelligent individuals have good physical and psychological health because they are better able to cope with life's challenges and can control their emotions more effectively (Taylor, 2001).In literature, various empirical studies have well documented the significant negative relationship between trait EI and psychological distress (e.g., Besharat, 2007;Dulewicz, Higgs, & Slaski, 2003;Tsaousis & Nikolaou, 2005) and between trait EI and sub dimensions of psychological distress, such as, depression and anxiety (Bauld & Brown, 2009;Extremera & Fernàndez-Berrocal, 2006;Fernandez-Berrocal et al., 2006).
Recently, researchers have argued that affective states at work, might serve to be the linking mechanism through which EI affects a variety of employee behavior and organizational outcomes (e.g., Kafetsios, 2007;Kafetsios and Zampetakis, 2008).The study of affective states at work is an important area of organizational behavior research (Ashkanasy, Hartel, and Zerbe, 2000) and help in understanding the Emotional Intelligence and Psychological Distress 25 processes underlying the effects of various working conditions, work events and personality dispositions on various work attitudes, behaviors and individual's psychological states (Kafetsios, 2007;Kafetsios & Zampetakis, 2008;Weiss & Cropanzano, 1996).Affective Events Theory (AET) (Weiss & Cropanzano, 1996) proposes that antecedent events at work with dispositions lead to affective states that in turn, lead to attitudinal and behavioral outcomes.Relevant to current study, AET suggests that dispositions (personality) may directly influence the employees' affective reactions.Thus, it is important to understand specific information about different dispositions in addition to work events in order to make predictions about individual's behavior.Individual differences influence reactions to the same work event, and these differences in reactions lead to different types of behaviors and attitudes (Weiss & Cropanzano, 1996).Various studies have supported the idea that affective states mediate the effect of work events and/or dispositions on outcomes (e.g., Fisher, 2000;Kafetsios, 2007;Kafetsios & Zampetakis, 2008;Weiss, Nicholas, & Daus, 1999).Thus, Hypothesis 3: PA and NA mediate EI effects on psychological distress.

Method Participants
The sample for this study consisted of 206 middle level managers from three public sector organizations situated in Pakistan.95 participants of the total sample (46 percent) were males and 111 (54 percent) were females.The mean age for this sample was 31.48 years (SD = 8.10).The sample was collected using non-probability purposive sampling method in order to obtain the appropriate number of participants for the study.Purposive sampling involves collecting any cases that contain the most representative attributes of the population (Cohen, Manion, & Morrison, 2000, p.99).Before distribution of questionnaires, permission was obtained from each of the organization.Attached to the survey instrument was a letter that explained the objective of the survey in general terms, assured respondents of the confidentiality of their responses, and notified them that participating in the survey was voluntary.All participants were treated in accordance with the "Ethical principles of Psychologists Code of Conduct" (American Psychological Association, 2002).Administration of the questionnaires was carried out by post graduate students who acted as research assistants and no monetary incentive was provided.

Instruments
Wong and Law Emotional Intelligence Scale (WLEIS: Wong & Law. 2002).WLEIS consists of 16 items and taps individuals' knowledge about their own emotional abilities rather than their actual capacities.Specifically, the WLEIS is a measure of beliefs concerning self-emotional appraisal (SEA) (e.g., "I have a good sense of why I have certain feelings most of the time"), others' emotional appraisal (OEA)(e.g., "I always know my friends' emotions from their behavior"), regulation of emotion (ROE) (e.g., "I always set goals for myself and then try my best to achieve them"), and use of emotion (UOE) (e.g., "I am able to control my temper and handle difficulties rationally").The response scale has been seven point Likert-type scale ranging from one (strongly disagree) to seven (strongly agree).Coefficients alphas for the four dimensions were: SEA: .80;OEA: .82;ROE: .81;UOE: .82.

Affectivity. Affectivity was measured by 20 items Positive and Negative Affect
Schedule (PANAS) (Watson, Clark, & Tellegen, 1988).PANAS is composed of two tenitem mood scales one to measure positive affectivity and the other to measure negativity affectivity.The higher scores on both PA and NA items indicate the tendency to experience a positive and negative mood.The ten positive affective states were: motivated, excited, feel strong, enthusiastic, proud, alert, inspired, determined, attentive, and active.The ten negative affective states were: distressed, upset, guilty, scared, hostile, irritable, ashamed, nervous, jittery, and afraid.Research has demonstrated the sound reliability of PANAS (i.e., Morris & Feldman, 1996).Respondents were requested to rate the statement on a 5-point scale (not at all to extremely) by comparing themselves during the past 2 weeks with their 'usual selves'.In this study, the positive and negative affect parts of PANAS had good internal consistency (Alphas .71and .85respectively).Psychological distress.Psychological distress was measured by Chan's (2005) twenty item scale.This scale measures psychological distress in terms of current nonpsychotic symptoms in the five symptom areas represented by scales of health concerns, sleep problems, anxiety, dysphoria, and suicidal ideas.Respondents were requested to rate each symptom statement on a 5-point scale (not at all to extremely) by comparing themselves during the past 2 weeks with their 'usual selves'.

Statistical Analysis
Prior to hypothesis testing data were tested for missing values and deviation from normality.Model fitting was performed with the help of AMOS (Arbuckle, 2006).
Covariance based structural model (CBSEM) was analyzed and interpreted in two stages: the measurement model and the structural model.The measurement model relates to the relations between manifest variables (observed items) and latent variables.The measurement model is tested by assessing the discriminability of the constructs in the model.This ensures that only valid constructs' measures are used before assessing the nature of relationships in the overall model.For testing the measurement model (discriminant validity of four constructs), chi-square difference test was employed (Anderson & Gerbing, 1988;Bagozzi & Phillips, 1982).The chisquare values for unconstrained model and the chi-square value for fully constrained model (all correlations constrained to 1) were determined.According to Anderson and Gerbing (1988, p.416) and Bagozzi and Phillips (1982, p.476), if the χ 2 for the first model is significantly smaller than second model, discriminant validity is achieved.This is because the better fitting model is the one where the two dimensions/constructs are viewed as distinctly different or not perfectly correlated.
Structural model specifies relations between latent constructs.The structural model is tested by estimating the paths between the constructs.Structural model was tested by computing path coefficients (ßs).A bootstrapping procedure using 1000 subsamples was performed to evaluate the statistical significance of each path coefficient.AMOS (Arbuckle, 2006) performs parametric bootstraps to find an approximate confidence interval for any model parameter under normal distribution theory.Steiger's Power Anlaysis (StatSoft, 2001) was used to estimate SEM modellevel power.Various fit indices were used to assess the fit of the model to the data, i.e., χ2/df, CFI, GFI, AGFI, RMSEA, and AIC.According to Hair et al. (2006), multiple indices should be used to assess a model's goodness of fit.The fit indices should include: the Chi-square value and associated degree of freedom; one absolute fit index (like the GFI, RMSEA, or SRMR); one incremental fit index (like CFI or TLI); one goodness of fit index (like GFI, CFI, or TLI); and one badness of fit index (like RMSEA, SRMR).A χ 2 /df value close to 2 indicates an acceptable fit (Bentler & Bonett, 1980) and the value close to 0.90 for CFI and GFI indicates a good fit (Schumacker & Lomax, 1996).The value below 0.08 for root mean square error of approximation (RMSEA), a measure that takes into account the error of approximation in the population, indicates good fit (Browne & Cudeck, 1993).Regarding AIC, the model with smallest AIC value is considered to be the best model (Schumacker & Lomax, 1996).

Results
Missing Value Analysis, Data Cleaning, and Item Parceling Prior to hypothesis testing, the validity of participants' responses was examined.In order to make data entry errors unlikely, the data was entered twice and comparison was made between two entries for data entry mistakes (Barchard & Christensen, 2007).Little's (1988) missing completely at random (MCAR) test revealed that the missing data were missing completely at random (χ 2 = 810.72.df = 935, p > .05).When the missing data is MCAR, any imputation method can be used (Hair et al., 2006).Expectation-maximization (EM) method was employed to impute missing values.
Values of skewness and kurtosis below the absolute value of 1 can be considered as acceptable (Miles and Shevlin, 2004).With the exception of 5 items, all items showed skewness and kurtosis smaller than 1 and these exceptions were close to the criterion.As well as, a visual check of QQ-plots and histograms also revealed unimodel distribution for all items.Three cases with extremely low z scores were found to be univariate outliers and three cases were identified through Mahalanobis' distance as multivariate outliers with p < .001.These six cases were removed from subsequent analysis.
Table 1 presents means, standard deviations and intercorrelations among variables.
* P < 0.05 ** P < 0.01 The practice of item parceling (combining items into small groups of items within scales or subscales) is much popular for reducing data analysis problems e.g., nonnormality and small sample sizes (e.g., Bandalos, 2002).Thus, in order to allow robust statistics, the size of the model was reduced by creating item parcels for 20 items PANAS (Watson, Clark, & Tellegen, 1988).Based on factor loadings (exploratory factor analysis) for each PANAS i.e., ten item's PA scale and ten item's NA scale, two 5-item parcels were created.Coefficient alphas for two PA scales were: PA1: .64;and PA2: .71.And coefficient alphas for two NA scales were: NA1: .70;and NA2: .80.   3).

Discussion
This study investigated the associations between EI and variables theoretically linked to it: PA, NA, and psychological distress.
In line with previous studies (e.g., Kafetsios & Zampetakis, 2008 ;Lopes et al., 2006 ;Sevdalis, Petrides, & Harvey, 2007)  are key resources that should be recognized for their worth…..They appear essential for effective and optimal personal and social functioning".In sum, emotionally intelligent employees are better able to apply these broad and build strategies in work settings (Kafetsios & Zampetakis, 2008) in order to successfully regulate their negative emotional experiences, which in turn produces beneficial consequences to their psychological and physiological well-being (Tugade & Fredrickson, 2002).
The results of this study confirmed the assertion that people high on negative affectivity are more vulnerable to poorer psychological well-being outcomes (e.g., anxiety and depression) (Oliver & Brough, 2002;Spector et al., 2000).This is because, they rate stressors as highly aversive experiences, and ruminate about their negative feelings often, which in turn amplify and increase in number of depressive episodes.
The results of this study support key assumptions of Affective Events Theory (AET) (Weiss & Cropanzano, 1996) which proposes that antecedent events at work with dispositions lead to affective states that in turn, lead to attitudinal and behavioral outcomes.Partial support was found for the mediating role of PA and NA in the relationship between EI and psychological distress.Only NA fully mediated the relationship between EI and psychological distress.Results suggest that EI helps in the formation of affective states at work, which in turn influence employee's psychological health.

Implications
The results of current study suggest following strategies to manage the issue of psychological distress in work setting.First, special attention should be given to EI during the selection process.Second, intervention strategies should be introduced to enhance EI among current employees.Literature provides us evidence that EI can be improved via various strategies (Goleman, 1995;Slaski, 2003).Third, based on Fredrickson's (2001) broaden-and-build model of positive emotions, it is suggested that intervention strategies should be introduced within work setting that cultivate positive emotions among employees.According to Fredrickson's (2000, p. 1), positive emotions (such as joy and contentment) broaden an individual's momentary thought -action repertoire, and help in eradicating the hold of negative emotions on an individual's mind and body.Enhancement of positive emotions help in preventing and treating problems such a psychological distress, deeply rooted in negative emotions.Fredrickson's (2000), suggested many intervention strategies that may help in preventing and treating psychological health related problems, as well as, help in building personal strengths, resilience and wellness of people.These

Limitations and Recommendations
The findings of this study are subject to several limitations which are common in this type of research.First, the results are specific to organizations in one geographical area and may or may not be generalizable to other areas.Second, the crosssectional data precludes any inference of causality.The direction of causality (in cross-sectional studies) cannot be established and will have to be examined using longitudinal data.Third, all our respondents were full-time employees and these findings may not be applicable to part-time employees.Four, since all measures were self report based measures; we cannot avoid the social desirability bias.Thus, the utility of self-report EI measure may be supplemented by employing performance based measures of EI (e.g.MSCEIT).Finally, the relationships among these variables might differ depending on the type of job.Future research can examine these relationships for jobs that differ in terms of intellectual and interpersonal demands (e.g., sales vs. Engineering).
intervention strategies include, (a) relaxation therapies (e.g., imagery exercises, muscle exercises, mediation exercises), (b) decreasing the intensity of unpleasant events and increasing the rates of engagement in pleasant activities, (c) cognitive therapies, (d) training employees in finding positive meaning in daily life, (e) building empathy between people and groups.Finally, giving feedback to individuals about their own EI levels might give them greater awareness of their own resources, which might help in lowering psychological distress.

Table 1
Descriptive statistics and inter-scale correlations Note : N = 200.Internal reliabilities in parenthesis.

Table 2
presents the fit statistics for measurement model.Results indicate that, the hypothesized measurement model fit the data well than a single factor model, both in terms of various fit indices and chi-square difference test.

Table 3
Standardized direct and indirect effects