Measuring stress reaction style: A construct validity investigation

Measuring stress reaction style: A construct validity investigation

Personality and Individual Differences 44 (2008) 250–262 www.elsevier.com/locate/paid Measuring stress reaction style: A construct validity investigat...

224KB Sizes 0 Downloads 28 Views

Personality and Individual Differences 44 (2008) 250–262 www.elsevier.com/locate/paid

Measuring stress reaction style: A construct validity investigation q Nigel Guenole a, Sasha Chernyshenko a,*, Stephen Stark b, Keith McGregor c, Siva Ganesh d a

Department of Psychology, University of Canterbury, PB 4800, Christchurch, New Zealand b University of South Florida, United States c Selector Ltd., New Zealand d Massey University, New Zealand Received 7 March 2007; received in revised form 5 August 2007; accepted 8 August 2007 Available online 27 September 2007

Abstract A stress reaction style describes the stress symptoms that will be experienced in the presence of environmental stressors. We present validity evidence for this construct from a theoretical review of related constructs, exploratory and confirmatory factor analyses, and a series of correlation and regression analyses. Results showed (1) stress reaction style is distinct from neuroticism, type-A personality, and state-based stress; (2) stress reaction style mediates the relationship between current stress and work withdrawal, and (3) stress reaction style has incremental validity over other predictors of work withdrawal including neuroticism, job satisfaction and current stress. These results provide construct validity evidence for the stress reaction style construct, and suggest potential utility in selection settings where measures of current stressors and strains are less applicable. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Stress; Personality; Individual differences; Stress reaction style; Job satisfaction; Work withdrawal

q

This research was partially funded by Selector Ltd. (New Zealand), The Centre for High Performance Development (London) and a grant from the New Zealand Department of Labour. * Corresponding author. Tel.: +64 03 364 2987; fax: +64 03 364 2181. E-mail address: [email protected] (S. Chernyshenko). 0191-8869/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2007.08.004

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

251

1. Introduction Organizations are increasingly aware of the detrimental impact of stress on organizational performance. While estimates vary widely, assessments of the cost of occupational stress to economies are typically in the billions of dollars. It is now accepted that effective management of psychological stress may have a number of benefits such as improved contextual performance (important employee behaviours not part of formal job descriptions) and reduced turnover intentions (e.g. Cropanzano, Rupp, & Byrne, 2003). It is unsurprising then that organizations use many measures that assess current stressors and strains, such as the Stress-in-General scale (SIG; Stanton, Balzer, Smith, Parra, & Ironson, 2001) and the Brief Symptom Indicator (BSI; Derogatis & Melisaratos, 1983). Such measures afford tremendous utility for the diagnosis of extant stressors and strains, and where necessary, prescription of interventions to address their effects. There has been comparatively little research, however, examining stress reaction style. Stress reaction style refers to individual differences in the expected experience of future strain. The domain covered by stress reaction style represents the range of symptoms expected by individuals when faced with environmental stressors. The focus of stress reaction style on the anticipated experience of future stress means it is conceptually distinct from both personality traits and measures of current psychological stress. Moreover, the future focus of the stress reaction style construct makes it appropriate for situations where more traditional measures assessing current stress cannot be used. For example, it would be useful to know which individuals are likely to experience severe stress reactions in high pressure occupations at the time of selection (e.g. police, military). Another application at the organizational level might include stress management programs, where the focus is on reducing aggregate levels of stress in organizations. Finally, individuals themselves may be interested in identifying and understanding their typical reactions to stress. This awareness could cause them to seek less stressful jobs, or to initiate more effective coping processes. For these reasons, a measure that assesses stress reaction style, as opposed to current stressors or strains, could complement existing stress measures in applied settings. We begin with an overview of individual differences research as it relates to stress. This serves to clarify the nature and range of stress reaction style, distinguish it from closely related constructs, highlight similar constructs, and demonstrate its potential utility. We then describe the development of a stress reaction style measure, including item pool development and exploratory and confirmatory factor analyses using large independent samples. Next, we present discriminant validity evidence for potentially confounded constructs, and examine predictive validity evidence for important work outcomes. We conclude with a discussion of limitations and suggestions for future research. 1.1. Individual differences and stress reactions Jex, Bliese, Buzzell, and Primeau (2001) noted that individual differences have not typically been a focus in stress research. However, there have been exceptions. Hardiness (Kobasa, 1979) is a personality trait comprised of commitment, control, and challenge. Individuals high on hardiness are thought to be less susceptible to certain forms of illness (Maddi, 1999). Funk and Houston (1987), however, encountered difficulties replicating the purported factor structure, and observed equivocal results when examining the links between hardiness and strain. Resilience

252

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

and Sense of Coherence are two constructs closely related to hardiness that have also received attention. Central to definitions of resilience is the idea of overcoming setback, or persistence in the face of adversity (Tugade & Fredrickson, 2004). Sense of Coherence (Anotonovsky, 1993) refers to a positive psycho-medical framework that empowers people to respond to stressors in an adaptive fashion. Hardiness, resilience, and sense of coherence assess people’s ability to deal with setbacks, and/or moderate the relationship between stressors and strains. They do not assess, however, respondents’ expectations of the nature and extent of future strain. An array of other personality constructs has been studied in relation to stress. The Big Five Neuroticism factor has been studied by stress researchers, and has been found to be related to many poor health outcomes (e.g. Williams & Wiebe, 2000). Neurotic individuals perceive stressors as more stressful, which may underlie some of the noted health risks (this is consistent with Neuroticism–health relations assessed under Eysenck’s three factor model, e.g. Walsh, Eysenck, Wilding, & Valentine, 1994). Similarly, type-A personality has been found to predispose individuals to poor health outcomes. For example, type-A has been linked to coronary heart disease and a variety of other health risk factors (e.g., Ewart, Jorgenson, Suchday, Chen, & Matthews, 2002). Neuroticism and type-A personality are personality traits. They are distal, individual difference attributes manifested in enduring tendencies to think, feel, and behave in certain ways across situations and settings. We expect to see their influence in the presence or absence of environmental stressors. The effect of stress reaction style, on the other hand, will be most strongly manifested in the presence of environmental stressors. It will be most evident when an appraisal of the stressor has led to a feeling of being unable to cope. Also common to constructs discussed so far is that they have a main effect on, or moderate, the intensity of strain, though they do not directly assess one’s likely reaction to future stressors. A stress reaction style measure would describe the strain likely to be experienced when faced with environmental stressors. One other personality trait, trait anxiety, warrants special mention in relation to stress reaction style. This is most commonly assessed using the Endler Multidimesional Anxiety Scales (EMAS; Endler & Parker, 1990). The difference between trait anxiety and reaction style is that whereas trait anxiety assesses anxiety by degree to which individual finds each of several scenarios threatening, a reaction style describes the symptoms, or reactions, they will experience when they find a situation threatening. Reinforcement Sensitivity Theory (RST: Gray, 1990) has given rise to models of personality that have underlying biological explanations. Gray proposed that individual differences in the behavioural inhibition system (BIS) and behavioural activation system (BAS) give rise to personality dimensions of anxiety and impulsivity respectively. While specific measures have been developed to assess BIS/BAS (e.g. BIS/BAS: Carver & White, 1994), research suggests that well developed personality scales measuring anxiety are satisfactory markers of BIS (Gomez & Gomez, 2005). Existing assessments of BIS/BAS, therefore, are closely related to Neuroticism, and inspection of scale content of these measures reveals they do not assess the symptoms of strain one will experience in the face of future stressors. Closer to measuring stress reaction style are measures of current strain (e.g., BSI). They assess whether symptoms of stress are, or were, present over a specified period of time, but do not assess how one is likely to manifest strain when faced with environmental stressors in the future. This is the gap in the research literature that our paper hopes to address.

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

253

2. Study 1: Developing a measure of stress reaction style A challenge we faced was demonstrating that our measure is not simply assessing state-based stress. To operationalize this distinction, we manipulated instructions to participants. We directly assessed the respondent’s expected reaction to stressors by instructing the respondent to rate how often they experience symptoms when faced with a stressful situation. Participants were instructed: ‘‘Please indicate which response option best describes how often each symptom occurs when you are under stress’’. The response scale for each item was (a) ‘‘this happens less when I am stressed’’; (b) ‘‘this stays the same whether or not I am stressed’’; (c) ‘‘this happens more when I am stressed’’; or (d) ‘‘this happens much more when I am stressed’’. 2.1. Item pool development The content domain for the stress reaction style measure could be any broad-spectrum model of current psychological distress, because we expected the symptoms experienced to be similar to the general manifestation of stress, albeit they are anticipated future symptoms in the case of stress reaction style. We therefore wrote new items to measure the domain assessed by the BSI, which is widely recognized as a measure of general psychological distress (Piersma, Boes, Janna, & Reamue, 1994) covering a range of physical, affective and behavioural stress reactions experienced by individuals. We wrote a total of 50 items tapping four stress reaction domains rationally derived from the BSI by researcher discussion. The first stress reaction domain was anticipated anxiety, which reflects an expectation of restlessness, nervousness, tension, and panic when faced with environmental stressors. Items typical of those in this scale described manifestations of the experience of anxiety, such as, ‘‘Becoming restless and unable to relax’’. The next domain was termed anticipated withdrawal. It corresponds to an expected strain experience representing a blend of the social alienation and interpersonal sensitivity dimensions of the BSI. An example item for this domain was, ‘‘Just wanting to be left alone’’. The third domain identified was anticipated somatization. In our conception, this scale assesses expectations of bodily dysfunction when faced with stressors. A typical item was ‘‘Getting upset stomach or bowels’’. Finally, we developed items that reflected expectations of cognitive dysfunction of the obsessive-compulsive dimension of the BSI, which we called anticipated distraction. A typical item was ‘‘Finding your mind going blank’’. 2.2. Participants For exploratory analyses, data were collected from a convenience sample of 1898 working adults, comprised of 944 Males (average age = 38.60 years, standard deviation = 8.67 years) and 954 females (average age = 34.06 years, standard deviation = 8.12 years). For confirmatory analyses, data were collected from a new sample of 1575 workers, comprised of 887 Males (average age = 39.09 years, standard deviation = 8.69 years) and 688 females (average age = 34.90 years, standard deviation = 9.39 years). Participants were from a broad range of industries, occupational groups and levels. Data were collected through a computer-based environment that allowed subjects to complete a questionnaire in their own time. Participation was voluntary and

254

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

participants received a stress reaction style report as compensation (contact first author for a sample report and the initial set of 50 items). 2.3. Procedure SPSS 15 was used for exploratory principal axis factor analyses with oblimin rotation on responses to items. A four-factor solution was extracted based on our hypothesized solution, parallel analysis, and scree plot criteria, and the best 6 items (i.e., with high primary loadings on the target factor and low secondary loadings) for each dimension were retained. Results were then cross-validated using confirmatory factor analysis on a new sample with LISREL 8.54 using WLS. 2.4. Results We named the four factors anticipated anxiety, anticipated somatization, anticipated distraction and anticipated withdrawal. We prefix scale names with the verb ‘anticipated’ to underscore we are assessing the respondent’s expected response when faced with environmental stressors. Resulting scales were unidimensional with high primary loadings on target factors and low secondary loadings (see Fig. 1 and Table 1). They showed moderately high internal consistency (anticipated anxiety = .79; anticipated somatization = .87; anticipated distraction = .83; anticipated withdrawal = .80). Evidence in favour of using four subscales rather than a single strain factor model came from patterns of inter-item correlations, with inter-item correlations in the subscales considerably higher than the inter-item correlations in the general scale. Scale average inter-item correlations for subscales were.40 for anticipated somatization,.53 for anticipated anxiety,.45 for anticipated distraction, and.39 for anticipated withdrawal. The four factors showed moderate inter-correlations, ranging in absolute value from.37 to.64 suggesting a higher order factor. Fit statistics for the second-order model showed a reasonably well fitting model by commonly accepted standards (e.g. Byrne, 1998) (Fig. 2; x2 = 1092.86, 248 df; CFI = 0.91; GFI = 0.98; AGFI = .97; RMSEA = 0.047; and NNFI = 0.90). This model showed similar fit to a correlated four-factor model without the second-order factor (Fig. 3; x2 = 1076.49, 248 df; CFI = 0.91; GFI = 0.98; AGFI = .97; RMSEA = 0.046; and NNFI = 0.89). The second-order model showed better fit than a competing model where a single general factor was used to model all item responses. Because these models are non-nested, a lower AIC value determines the best fitting model (Byrne, 1998). AIC for the second-order model was 1196.86 compared with 1786.08 for the single factor model indicating that the structure of our measure is hierarchical with 4 facet dimensions and a single broad dimension (Table 2).

3. Study 2: Convergent and discriminant validity of stress reaction style Our second study presents a series of correlation and regression analyses to examine the relationship between stress reaction style (operationalized as the sum of 4 facet scores) and (a) potentially confounded constructs including emotional stability, type-A behaviour, and current stress;

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

255

Fig. 1. Scree plot for factor analysis.

Table 1 Factor loading matrix for oblimin rotated solution Item

Mean

SD

Anxiety1 Anxiety2 Anxiety3 Anxiety4 Anxiety5 Anxiety6 Somatization1 Somatization2 Somatization3 Somatization4 Somatization5 Somatization6 Distraction1 Distraction2 Distraction3 Distraction4 Distraction5 Distraction6 Withdrawal1 Withdrawal2 Withdrawal3 Withdrawal4 Withdrawal5 Withdrawal6

1.61 1.91 1.64 1.59 2.03 1.96 3.17 3.26 3.16 2.99 3.05 2.65 2.30 2.47 2.26 2.33 2.28 2.40 2.52 2.40 2.27 2.16 2.62 2.17

.89 .96 .80 .78 .99 1.07 .99 .98 .99 1.05 1.03 1.11 .85 .81 .87 .97 .88 .88 .98 .94 .98 .93 .97 .92

1

2 .87 .56 .55 .47 .40 .31 .04 .05 .06 .05 .03 .26 .09 .05 .09 .03 .06 .04 .03 .10 .11 .01 .08 .13

3 .00 .03 .07 .02 .16 .13 .83 .83 .80 .64 .59 .50 .00 .01 .04 .10 .07 .02 .05 .08 .03 .02 .18 .03

4 .02 .11 .05 .01 .16 .13 .03 .02 .03 .07 .00 .03 .81 .66 .64 .58 .57 .54 .02 .05 .03 .20 .07 .14

.06 .05 .18 .09 .02 .09 .05 .02 .05 .07 .21 .02 .13 .02 .08 .03 .00 .12 .67 .64 .56 .48 .38 .36

256

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

.75

Somatization 1

.44

Somatization 2

.45

Somatization 3

.47

Somatization 4

.64

Somatization 5

.52

Somatization 6

.60

Anxiety 1

.29

Anxiety 2

.28

Anxiety 3

.70

Anxiety 4

.26

Anxiety 5

.51

Anxiety 6

.48

Distraction 1

.39

Distraction 2

.40

Distraction 3

.50

Distraction 4

.50

Distraction 5

.46

Distraction 6

.52

Withdrawal 1

.47

Withdrawal 2

.45

Withdrawal 3

.50

Withdrawal 4

.70

Withdrawal 5

.49

Withdrawal 6

.66

.15 .74 .73 Anticipated Somatization .60 .69 63 0 .92 .84 .39 .85 .86 Anticipated Anxiety .70 .77 .78 .72 Reaction Style .78 .17 .75 .70

.91 Anticipated Distraction

.71 .73

.69

.73 .96

.09 .74 .71 Anticipated Withdrawal .71 .71

.58

Fig. 2. Second-order CFA model.

and (b) important organizational criteria such as job satisfaction and work withdrawal. We hypothesized that: (1) Correlations of stress reaction style scores with neuroticism, type-A behaviour, perceived work stressors, and current stress symptoms should be only moderate, providing evidence of discriminant validity. (2) Stress reaction style and current strain should be highly correlated when individuals are under stress at work, but this correlation will diminish when people are not under stress.

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

.75 .74 .73 Anticipated Somatization .60 .69

.63 .71

.84 .85 .86 Anticipated Anxiety .70

Somatization 1

.44

Somatization 2

.45

Somatization 3

.47

Somatization 4

.64

Somatization 5

.52

Somatization 6

.60

Anxiety 1

.29

Anxiety 2

.28

Anxiety 3

.26

Anxiety 4

.50

Anxiety 5

.41

Anxiety 6

.48

Distraction 1

.38

Distraction 2

.44

Distraction 3

.50

Distraction 4

.50

Distraction 5

.46

Distraction 6

.52

Withdrawal 1

.47

Withdrawal 2

.45

Withdrawal 3

.50

Withdrawal 4

.50

Withdrawal 5

.49

Withdrawal 6

.66

257

.85 .77

.72 .68 .79 .75 .86

.71 Anticipated Distraction .76

.71 .73

.69

.73 .87 .74 .71 Anticipated Withdrawal .71 .71

.58

Fig. 3. Correlated four-factor model. Table 2 CFA model fit statistics Model

x2

df

CFI

NNFI

GFI

AGFI

RMSEA

AIC

Second-order model Correlated four-factor model Single factor model

1092.86 1076.49 1690.08

248 248 252

.91 .91 .85

.90 .89 .83

.98 .98 .96

.97 .97 .96

.05 .05 .06

1196.86 1184.49 1786.08

In other words, stress reaction style should moderate the relationship between stressors and current strain.

258

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

(3) Because stress reaction style anticipates the experience of strain, which in turn predicts job satisfaction, current stress symptoms should mediate the relationship between stress reaction style and job satisfaction. (4) Stress reaction style should show incremental validity over current stress, emotional stability and job satisfaction in predicting work withdrawal.

3.1. Participants and procedure Two hundred and twenty-nine employees at two call centres completed an internet-based questionnaire. There were 135 females (mean age = 34.87 years, SD = 9.48) and 94 males (mean age = 40.50 years, = 9.77). Participation was voluntary and participants received a stress reaction style report as compensation. Discriminant and predictive validities of the stress reaction style were examined using zero-order correlations (hypothesis one), moderated and mediated multiple regressions (hypotheses two and three), and hierarchical regression (hypothesis four). 3.2. Measures Stress Reaction Style was assessed with the 24-item measure developed in study 1. Scores were calculated as the sum of the 24-items from its 4 facet subscales. Internal consistency reliability estimate for this sample was .92. Job satisfaction was assessed with a 10-item version of the Satisfaction with Work subscale of the Illinois Job Satisfaction Index (IJSI; Chernyshenko, Stark, Crede, Wadlington, & Lee, 2003). For each item, respondents were asked to indicate the degree to which an item accurately described their work situation. A four-point scale (‘‘Strongly Disagree’’; ‘‘Disagree’’; ‘‘Agree’’; ‘‘Strongly Agree’’) was used. The internal consistency reliability estimate for this scale was .86. Current stress symptoms were measured with the somatization, depression, and anxiety subscales from the BSI. Items asked participants, ‘‘over the past week, have you been distressed by any of the following?’’ The response scale was 0 (not at all), 1 (a little bit), 2 (moderately), 3 (quite a bit), and 4 (extremely). Items were aggregated to form a single measure, because research suggests this instrument measures a unidimensional construct (e.g. Derogatis & Melisaratos, 1983; Piersma et al., 1994). The internal consistency reliability estimate for this scale was .90. Emotional Stability. The 10-item scale from the International Personality Item Pool (Goldberg, 1999) was used. Responses were collected using a 5-point strongly disagree to strongly agree format. The internal consistency reliability estimate for this scale was .87. Work stress. General work stress was assessed with the 15-item Stress-in-General scale (SIG; Stanton et al., 2001) that uses a three-point format (‘‘Yes’’, ‘‘?’’, ‘‘No’’). Internal consistency reliability was estimated at.86. Work Withdrawal. A 10-item Work Withdrawal scale developed by Hanisch and Hulin (1991) assessed absenteeism, tardiness, and other behaviours reflecting employee desires to avoid work tasks and the work environment. A 5-point response format was used ranging from 0 (Never) to 4 (Many times). The internal consistency reliability for this scale was .77.

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

259

Type A Behaviour was measured using the simple sum of Goldberg’s (1999) Big Five markers for Conscientiousness and Extraversion, an operationalization we adapted from Hough and Ones (2001). Internal consistency reliability was estimated for this scale at .90. 3.3. Results Zero-order correlations between stress reaction style scores and potentially confounded constructs are presented in Table 3. As expected, there were only moderate relationships between stress reaction style and the potential confounds of type A ( .23), Emotional Stability ( .38), and current stress (.46). Our new measure also showed potentially useful criterion-related validities with job satisfaction ( .18), and work withdrawal (.34). Moderated regression results presented in Table 4 indicated that current work stress moderated the relationship between stress reaction style and strain. The significant interaction term for stress reaction style and work stress means that the relationship between stress reaction style and current stress symptoms (strain) depends on the level of work stress (SIG). This is an important result given the difficulty of detecting moderating effects in psychological research (Aguinis, Beaty, Boik, & Pierce, 2005). We explored the nature of the interaction using simple slopes analysis (O’Connor, 1998). Consistent with our hypothesis two, results of this analysis showed that when work stress (SIG) was high, the standardized b between reaction style and BSI was higher (SIG = 84th percentile, b = .56, 95% CI.39–.73) than when work stress was moderate (50th percentile, b = .41, 95% CI.29–.53) or low (SIG = 16th percentile, b = .256, CI.11–.41).

Table 3 Correlations between stress reaction style and potential confounds Scale

Items

Mean

SD

1

Stress Reaction Style Type-A Emotional Stability BSI Job Satisfaction Work Withdrawal Stress-in-General

24 20 10 19 10 10 15

65.17 72.70 34.98 28.62 32.06 17.61 27.00

11.34 9.38 7.32 9.06 4.84 5.35 7.22

2 .92 .23 .38 .46 .18 .34 .29

3 .79 .25 .20 .31 .18 .04

4

.86 .49 .26 .26 .27

5

.90 .30 .48 .37

.86 .38 .32

6

7

.77 .23

.82

Note: N = 229. Reliabilities on diagonal.

Table 4 Regression results for moderator hypothesis Predictor

B

Standard error

b

Constant Stress Reaction Style SIG Interaction

12.02 .13 1.39 .02

11.17 .17 .40 .01

.16 1.11 .88

t

p 1.08 .77 3.47 2.70

.28 .44 .00 .01

Note: Dependent variable is current strain (BSI), R2 = .30. Interaction p-value is in fact .0074 rounded to .01, so the interaction is significant.

260

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

Table 5 Regression results for current stress mediating current stress and job satisfaction Model

Predictor

1

Intercept Stress Reaction Style

8.31 .37

2.63 .05

.46

3.16 7.89

.00 .00

Intercept Stress Reaction Style

36.19 .08

1.56 .03

.18

23.18 2.70

.00 .01

Intercept Stress Reaction Style BSI

37.42 .02 .15

1.55 .03 .04

.05 .28

24.17 .67 3.87

.00 .50 .00

2 3

B

Standard error

t

b

p

Note: Model 1 dependent variable is BSI, R2 = .22; Model 2 dependent variable is Job Satisfaction, R2 = .03; Model 3 dependent variable is Job Satisfaction, R2 = .09.

Our third hypothesis, that current stress symptoms mediated the relationship between reaction style and job satisfaction, was also supported (see Table 5). In the first regression equation, stress reaction style significantly predicted current stress symptoms (BSI). In the second regression equation, stress reaction style scores significantly predicted job satisfaction. In the third regression equation, where job satisfaction was regressed on both stress reaction style and the BSI, the BSI significantly predicted job satisfaction. These three regression results satisfy the requirements for mediation (Baron & Kenny, 1986). Finally, we found support for our hypothesis that reaction style has incremental validity over neuroticism, job satisfaction, and current stress symptoms in the prediction of work withdrawal. The standardized regression coefficient of.14 for the reaction style disposition was significant (p < .05) when all four predictors of work withdrawal were in the regression equation (see Table 6).

4. Discussion Estimates of the costs of stress to individuals, organizations and economies are typically in the billions of dollars. Fortunately, psychologists have identified useful strategies for identifying Table 6 Hierarchical regression results for incremental validity hypothesis Model

Predictor

B

Standard error

Step 1

(Constant) BSI Job Satisfaction Emotional Stability

19.62 .24 .29 .01

3.19 .04 .07 .05

.41 .26 .01

6.16 6.24 4.44 .23

.00 .00 .00 .82

(Constant) BSI Job Satisfaction Emotional Stability Stress Reaction Style

16.00 .21 .29 .03 .07

3.54 .04 .06 .05 .03

.36 .26 .04 .14

4.52 5.18 4.44 .64 2.25

.00 .00 .00 .52 .03

Step 2

Note: Dependent variable is work withdrawal, R2 = .31.

t

b

p

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

261

individuals who have experienced (or are currently experiencing) stress, and how stressors and their consequences (or strains) can be ameliorated. Research to date has focused, however, on assessing stress after it has occurred. This limits the application of many stress measures in selection contexts. Our research examined the construct validity of stress reaction style, which describes individual differences in the tendency to experience strains, as opposed to whether stress has occurred in the past, or if it is occurring presently. We found theoretical and empirical support for the stress reaction style construct. The structure of stress reaction style was mapped with exploratory analyses, and the derived scales showed high internal consistency reliabilities. Support was subsequently observed for the proposed structure of the stress reaction style measure from confirmatory factor analyses. This allowed us to proceed with investigating the relationships between stress reaction style and other constructs to establish discriminant validity and examine the utility of stress reaction style in work settings. Results of these further analyses suggested that the stress reaction style has the ability to complement current measures of stressors and strain, and importantly, incrementally predict withdrawal (over neuroticism) at the point of selection. Perhaps the biggest limitation of this research is that we have proposed longitudinal stability for the reaction style in the absence of a longitudinal design. The reaction style construct would therefore benefit from longitudinal research to show that anticipated symptoms are the symptoms that are subsequently experienced. This research would also benefit from replication using measures other than self reports. Finally, while our investigation has shown preliminary evidence of the likely relationship with BIS, because Emotional Stability can be viewed as an operationalization of BIS, more work is needed using measures specifically designed to assess this construct. We hope that our paper stimulates research into the stress reaction style construct, as these findings suggest it is a valid psychological construct with potential utility in applied settings.

References Aguinis, H., Beaty, J. C., Boik, R. J., & Pierce, C. A. (2005). Effect size and power in assessing moderating effects of categorical variables using multiple regression: A 30-year review. Journal of Applied Psychology, 90, 94–107. Anotonovsky, A. (1993). The structure and properties of the sense of coherence scale. Social Science and Medicine, 36, 725–733. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182. Byrne, B. (1998). Structural Equation Modelling with Lisrel, Prelis, and Simplis: Basic concepts, applications, and programming (Multivariate Applications Book Series). Lawrence Erlbaum Associates. Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment: The BIS/BAS scales. Journal of Personality and Social Psychology, 67, 319–333. Chernyshenko, O. S., Stark, S., Crede, M., Wadlington, P., & Lee, W. (2003). Improving the measurement of job attitudes: The development of the IJSI. Paper presented at the 18th annual conference for the Society of Industrial and Organizational Psychologists, Orlando, Fl, April. Cropanzano, R., Rupp, D., & Byrne, Z. (2003). The relationship of emotional exhaustion to work attitudes, job performance, and organizational citizenship behaviours. Journal of Applied Psychology, 88(1), 160–169. Derogatis, L. R., & Melisaratos, N. (1983). The Brief Symptom Inventory: An introductory report. Psychological Medicine, 13, 595–605.

262

N. Guenole et al. / Personality and Individual Differences 44 (2008) 250–262

Endler, N. S., & Parker, J. D. A. (1990). Multidimensional assessment of coping: A critical evaluation. Journal of Personality and Social Psychology, 58, 844–854. Ewart, C. K., Jorgenson, R. S., Suchday, S., Chen, E., & Matthews, K. A. (2002). Measuring stress, resilience and coping in vulnerable youth: The social competence interview. Psychological Assessment, 14(3), 339–352. Funk, S. C., & Houston, B. K. (1987). A critical analysis of the hardiness scale’s validity and utility. Journal of Personality and Social Psychology, 53(3), 572–578. Goldberg, L. R. (1999). A broad-bandwidth, public-domain, personality inventory measuring the lower-level facets of several five-factor models. In I. Mervielde, I. Deary, F. De Fruyt, & F. Ostendorf (Eds.). Personality psychology in Europe (Vol. 7, pp. 7–28). Tilburg, The Netherlands: Tilburg University Press. Gomez, R., & Gomez, A. (2005). Convergent, discriminant and concurrent validities of measures of the behavioural approach and behavioural inhibition systems: Exploratory and confirmatory factor analytic approaches. Personality and Individual Differences, 38, 87–102. Gray, J. A. (1990). Brain systems that mediate both emotion and cognition. Cognition and Emotion, 4, 269–288. Hanisch, K. A., & Hulin, C. L. (1991). General attitudes and organizational withdrawal: An evaluation of a causal model. Journal of Vocational Behavior, 39, 110–128. Hough, L., & Ones, D. (2001). The structure, measurement, validity, and use of personality variables in industrial, work, and organizational psychology. In D. S. O. Anderson, K. Sinangil, & C. Viswesvaran (Eds.), International handbook of work and organizational psychology. Sage Publications. Jex, S. M., Bliese, P. D., Buzzell, S., & Primeau, J. (2001). The impact of coping and self-efficacy on stressor-strain relations. Journal of Applied Psychology, 86(3), 401–409. Kobasa, S. C. (1979). Stressful life events, personality, and health: An enquiry into hardiness. Journal of Personality & Social Psychology, 37, 1–11. Maddi, S. R. (1999). The personality construct of hardiness I. Effects on experiencing, coping and strain. Consulting Psychology Journal: Practice and Research, 51(2), 83–94. O’Connor, B. P. (1998). SIMPLE: All-in-one programs for exploring interactions in moderated multiple regression. Educational and Psychological Measurement, 58, 836–840. Piersma, H. L., Boes, J. L., Janna, L., & Reamue, W. M. (1994). Unidimensionality of the Brief Symptom Inventory (BSI) in adult and adolescent inpatients. Journal of Personality Assessment, 63(2), 338–344. Stanton, J. M., Balzer, W. K., Smith, P. C., Parra, L. F., & Ironson, G. (2001). A general measure of work stress: The Stress-in-General Scale. Educational and Psychological Measurement, 61(5), 866–887. Tugade, M. M., & Fredrickson, B. L. (2004). Resilient individuals use positive emotions to bounce back from negative emotional experiences. Journal of Personality and Social Psychology, 86(2), 320–333. Walsh, J., Eysenck, M. W., Wilding, J., & Valentine, J. (1994). Type A, neuroticism, and physiological functioning (actual and reported). Personality and Individual Differences, 16(6), 959–965. Williams, P. G., & Wiebe, D. J. (2000). Individual differences in self-assessed health: Gender, neuroticism and physical symptom reports. Personality & Individual Differences, 28(5), 823–835.