PERGAMON Prevonality and Individual Differences 36 (2004) 447-456, ——$__ ‘ow elvis comlocefpaid Fffects of neuroticism and workload history on performance ‘Luz-Eugenia Cox-Fuenzalida®*, Rhonda Swickert>, James B. Hittner> "Department of Psychology, University of Oklahoma, 455 W. Lindsey Street, DHT-705, Norman, OK 73019, USA “Department of Psychology, Collegeof Charleston, 65 George Street, Charleston, SC 29424, USA Received 25 July 2002; received in revised form 23 December 2002; accepted 7 February 2003 Abstract This study manipulated workload levels to test Eysenck’s theory of neuroticism by examining indivi- duals’ differential responses to the stress of sudden shifts in workload. A sample of 194 participants who had completed an inventory to assess neuroticism were randomly assigned to either a High-to-Low or Low-to-High testing condition, Participants performed an auditory vigilance task while reaction time and correct responses were assessed. Multiple regression analyses were conducted and results indicated that higher levels of neuroticism were associated with decreased reaction time in both conditions. In addition, individuals higher in neuroticism evidenced fewer correct responses in the Low-to-High workload condi- tion. Results of this study are generally consistent with Eysenck’s theory of neuroticism, © 2003 Blsevier Ltd. All rights reserved. Keywords: Neuroticism; Workload history; Task performance; Stress The concept of neuroticism has received considerable attention in the stress and coping litera- ture as it is believed to be an important predictor of how people deal with stressful events, Although many theories have been proposed to explain this dimension, one of the most widely cited is Hans Eysenck’s theory. Fysenck (1967) defined neuroticism as 2 dimension that ranges from emotional stability to instability and he suggested that individual differences in neuroticism are a result of arousal differences that are mediated by the limbic system. The putative function of the limbic system is to regulate autonomic and emotional reactions, in particular, reactions that are cued by physically or psychologically stressful experiences (Eysenck, 1967). Consequently, individuals who are high in neuroticism, compared with those who are low, are believed to evi- dence greater activation of the limbic system when responding to environmental stressors. This © Corzesponding author, Tel: + 1-405-325-4511; fax: + 1-405-325.4737, Bxmail addvess: cox-fuenzalida@ou.edu (LB. Cox-Fuenzalia), 0191-8869/09/8 «see front matter © 2003 Bleewee Lid. All rights reserved o10.1016)50191.8889(03)00108.843 LB. Cox-Puencalida etal sonality and Individual Diferences 36 (2004) 447-455 differential autonomic arousal level is believed to form the basis for the behavioral differences observed between those scoring high and low in neuroticism. Although behavioral differences across the neuroticism continuum have been examined using a variety of approaches, one important strategy used to test Eysenck’s arousal-based theory involves the examination of performance differences between high and low neurotics using com- puter-based human performance tasks. Indeed, several studies have demonstrated that indivi- duals scoring high in neuroticism (and trait anxiety) typically perform more poorly on a variety of tasks compared with those low in neuroticism (Eysenck, 1983; Eysenck & Eysenck, 1985; Spence & Spence, 1966). Typically, the detrimental effects of neuroticism are particularly evident in studies using vigilance and monitoring tasks—tasks that require high levels of attention (for reviews, see Eysenck, 1983; Matthews é& Deary, 1998). For example, Newton, Slade, Butler, and Marphy (1992) conducted a study with 123 participants that employed a visual search task Tequiring participants to scan a random display of letters to determine the presence or absence of a target letter. This study found neuroticism to be significantly related to slower reaction time in the predicted direction, that is, high neurotics evidenced slower reaction times. Although some studies have found an association between newroticism and improved performance (Eysenck & Calvo, 1992), generally, findings are consistent with Newton et al. (1992) suggesting that neuro- ticism effectively predicis decrements in performance on vigilance tasks (Darke, 1988; Eysenck, 1992, 1997). In addition to predicting simple vigilance and monitoring task performance, studies have also demonstrated that neuroticism influences performance on other, more complex, types of tasks. For instance, Mayer (197) conducted a study manipulating task difficulty while examining the effect of trait anxiety on easy versus difficult tasks. He concluded that although high trait anxiety facilitated performance on easy tasks (requiring simple, rote problems—e.g. searching for the letter “a’), it had a detrimental effect on dificult tasks (c.g. anagram solving). ‘Although the studies cited earlier have established a link between neuroticism and task perfor- mance, to date little is known about the relationship between neuroticism and basic performance dynamics such as variability in workload, or changes in task demand. In other words, while pre- vious research seems to be generally consistent with Eysenck’s arousal-based theory of neuroti- cism, most of these performance-based studies lack generalizability because they do not examine parameters associated with real-world work environments, such as changing workload levels. The study of workload history may help to address this problem. ‘The study of workload history typically involves the manipulation of workload levels over time with individuals moving from either high-to-low, ot low-to-high, workload levels. Because per- formance is examined in this manner, the study of workload history might be considered more ecologically valid than less dynamic approaches (e.g. fixed workload level studies). Currently, little is known about the effects of neuroticism and workload history on performance, however, there is a body of work that has examined the manner in which workload shifts influence general performance. One consistent finding reported in the workload history literature suggests that there is a significant decrement in performance following a sudden decrease in workload level (Cumming & Croft, 1973; Goldberg & Stewart, 1980; Matthews, 1986), In 2 recent study, how- ever, Cox-Fuenzalida (2000) found that either a sudden decrease or increase in workload level results in a significant performance decrement. One explanation for the petformance decrements following both types of workload shift is that any change in workload conditions might serve as aLB, Cox-Puencalida et al. | Personality and Individual Differences 36 (2004) 447-456 449 stimulus that induces stress. In fact, it has been widely discussed in stress literature that one essential component of stress is change in environmental conditions (Holmes & Rahe, 1967; Selye, 1956), Therefore, one might argue that the study of workload history creates the necessary conditions for directly studying stress responses and, in so doing, allows for a direct test of new- roticism theory. Moreover, by manipulating stress in this manner (i.e. by changing workload levels), this approach may represent a more ecologically valid procedure for testing Eysenck’s theory of neuroticism than has been utilized by previous studies, Therefore, the purpose of this study was to examine the association between neuroticism and task performance across two dif- ferent types of changing workload conditions (Low-to-High versus High-to-Low). Given that higher levels of neuroticism are believed to be associated with greater stress-proneness, and that a sudden change in workload constitutes a stressful event, it was hypothesized that neuroticism would be significantly positively associated with performance deccements following either a sud- den increase or decrease in workload demand. 1. Method 1.1. Participants Participants were 194 undergraduate students (103 females, 91 males) from the University of Oklahoma. Participants were randomly selected from a larger pool of 450 lower division psy- chology students who had participated in a pretesting session in which neuroticism was assessed, Following the random selection procedure, participants were contacted by telephone and invited to participate in the study. 1.2. Materials A computer-based version of the Bakan Vigilance Task (1959) was utilized. The Bakan is an auditory vigilance task comprised of a sequence of digits presented to participants by way of earphones. Each training and test trial was 3 min in duration. During each trial, participants were instructed to detect odd-even-odd sequences of digits (the signal, e.g. 7-8-3). Participants were directed to press a specified key on the computer keyboard when they detected a signal. A total of 10 signals were presented in each 3-min period among a string of random digits for the high and Jow workload conditions (225 and 90 digits total, respectively). Workload level was manipulated by changing the speed of digit presentation (ie. high workload consisted of 1 digit every 0.8 s and Jow workload consisted of I digit every 2 5). In other words, the number of signals was the same for both high and low workload conditions but the total number of digits, and the difficulty level, varied due to digit rate presentation during the trial. Following signal presentation, the time out period for participants’ responses was 4.8 s. An error was recorded if the participant did not respond within this time frame. The Eysenck Personality Inventory (EPI, Form A; Eysenck & Eysenck, 1968) was used to assess the dimension of neuroticism. Items on the Neuroticisn Scale of the EPI measure an individual's level of emotional stability versus emotional reactivity. Higher scores indicate greater levels of neuroticism. Reliability estimates for the EPI neuroticism scale range from 0.84 to 0.92,450 LB. Cox-Puencalida et al.j Personality and Individual Differences 36 (2004) 447-456 while internal consistency coefficients range from 0.89 to 0.95, Construct validity for this test is reported in the EPI Manual (Bysenck & Eysenck, 1968). In the current study the mean of neu- roticism was 11.08 with a standard deviation of 4.51 which is generally consistent with normative data reported in the EPI. 13, Procedure When participants arrived at the laboratory they were seated at individual workstations and asked to complete informed consent forms. Workstation partitions minimized distraction from other participants yet permitted individuals to view the experimenter during instruction admin- istration. JA. Training and baseline data collection Participants performed the Bakan Vigilance Task during three phases: training, baseline, and the experimental testing session. Table 1 presents the counterbalanced orderings for all training, baseline, and test trials. Bach of the 194 participants was block randomized to one of the four training/baseline/testing sequences in Table 1. Each participant was first familiarized with the location and operation of the computer response keys relevant for performing the monitoring task. Participants then completed an initial feedback trial followed by an 18-min training session (three, 3-min Uials at high difficulty and three, 3-min trials at low difficulty, counterbalanced in their order of presentation) to ensure understanding of task instructions and to lessen the likelihood of the test trials being con- taminated by learning effects. The training regimen was patterned after that used by Schlegel and Gilliland (1990), who found that thorough instructions followed by five to six, 3-min trials were sufficient to produce reasonably asympiotic performance on a variety of human performance tasks. Training session trials were followed by an 18-min baseline session (three, 3-min trials at high difficulty and three, 3-min trials at low difficulty, in counterbalanced order) to establish baseline data for later comparisons. To minimize fatigue, S-min rest breaks were given between each series of three training and three baseline trials. During these breaks, participants were required to engage in a low-demand distracter task (.e. completion of participant demographic survey). The three trials of either training or baseline were each 3 min long and were presented with no discernable break between trials (i.e. the task appeared to be nine continuous minutes). Following the second series of three baseline trials and prior to the testing session, participants were given a 15-min break. 1.5. Testing session A participant's assignment to the High-to-Low or Low-to-High testing condition was based on the training/baseline/testing protocol sequence to which the participant was originally block randomly assigned (see Table 1). During the test session, participants in the High-to-Low condi- tion engaged in three, 3-min trials at high task difficulty followed immediately by a 3-min trial at ow task difficulty. In contrast, participants in the Low-to-High test condition engaged in three, 3- ‘min trials at low task difficulty followed immediately by a 3-min trial at high task difficulty. These451 LE, Cax-Fuenzalida etal. | Personality and Individual Differences 36 (2004) 447-456 uBprasan | Movasay “heron wes | morauyaseg [wen mus | yByroumeseg fore evs | morte, [es wwe | ybie-SuUesL, WErerMoT =a we “er Oe a Sao worpuen Suneo 1 on vas | morroureseg [moawes | uSrauoseg [ron mus | ybpi Suu, fears] movSue peo se | yOpsouiaseg [ren ws pron wes | mor-Gucess fevers | yoni Bae. Sao we awe cane oe Jee west | uonioureeeg [wan avs | moy-auroseg foeen mus | yBiunses [een were | mor Suu FaoRpUeS BoNesee pus BONE, suonipues MOTO a pu yBeH-00 oy sen a1 pur auaseg ‘Buysten soy sSzeps0 paoweyEas3aNOD rare.452 LE, Cox-Fuenzallda eral. | Personally and Individual Diferences 36 1) HI-456 two test conditions created a situation where participants developed a workload history at one workload level and then moved immediately to a dramatically different workload level. Thus, transitions between workload levels during the High-to-Low and Low-to-High test sessions were uninterrupted by rest periods and consequently were perceived by participants as 12 min of con- tinuous work. Participants were tested between the hours of 9.00 a.m. and 4.00 p.m, to control for time-of-day effects (Revelle, Humphreys, Simon, & Gilliland, 1980). 2, Results ‘The two dependent variables, reaction time and number of correct responses, were analyzed separately. In addition, for each dependent variable, the Low-to-High and High-to-Low work- load training conditions were first examined independently before being compared. Finally, for each dependent variable, only the last 3-min trial of the 12-min testing trial was considered in the analyses. Analyses were limited to this trial because only the last trial reflects the effects of chan- ging workload level. Means and standard deviations for the reaction time and correct response variables, for both the baseline and testing conditions, are presented in Table 2. Prior to addressing the primary hypotheses, the univariate distributions for the two dependent variables, across both workload training conditions, and the independent variable of neuroticism, were examined for departures from normality. Due to the significant negative skewness of the correct response variables in both conditions, and the significant positive skewness of the reaction time data in both conditions, these variables were transformed toward normality using Manly’s (1976) single parameter exponential transformation, These transformed variables, in turn, were used in all inferential data analyses, Given that all of the variables in the study were continuously scaled, a multiple regression approach was used to analyze the data. In particular, prior to examining the influence of neuro- ticism on testing performance, the baseline level of performance was controlled for by regressing the testing variable on the baseline variable, and then saving the standardized residuals, or residualized change scores, as a new variable. The residualized change scores—free from the influence of baseline training performance—were then analyzed as the dependent variable of interest. To screen for bivariate outliers, scatterplots were examined (x axis neuroticism; y axis=residualized scores) and all extreme outliers that deviated from the swarm of data poinis were identified. Next, a regression analysis was conducted, excluding the bivariate outliers, in which the residualized scores were regressed onto neuroticism. Results for the reaction time con- ditions are as follows. For the Low-to-High condition (= 89 after removing 2 bivariate outliers), Table 2 ‘Means and standard deviations by testing condition Lowso-High High-to- Low Low-to-High High-to-Low reaction time reaction time correct response correct sesponse Baseline 081 (&D.~0.37) 082 SD.=019 6.80 S.D.=197) 9.23 (SD.=1.54) Testing 094 (.D.=045) 0.87 (SD.=0.28) 6.15 (SD.=2.08) 8.78 (SD.=1.70) ‘Note. Reaction time is measured in seconds and correct response scores can range from 0 to 10.LB, Cox-Puenzolida et al.| Personality and Individual Differences 36 (2004) 447-456 453 neuroticism was @ significant predictor of testing performance accounting for 5.7% of the var- iance [t (87) = 2.29, P=0.024, 6=0.24}. In the High-to-Low condition (n= 103, no bivariate out- liers), neuroticism also significantly predicted testing performance accounting for 4.2% of the variance {t (101)=2.10, P=0.038, 6=0.2I). To determine whether the strength of association between neuroticism and reaction time performance differed across the two workload conditions (Low-to-High versus High-to-Low), a Fisher’s z-test for independent regression coefficients was conducted, The result of the 2-test was statistically nonsignificant (=0.15, P>-0.05). In analyzing the correct response variables, the same approach as described above for the reaction time data was utilized. Regarding the Low-to-High condition (n=84 after removing 7 bivariate outliers), neuroticism was a significant predictor of testing performance accounting for 6.6% of the variance [¢ (82) = ~2.42, P=0.018, 6= -0.26]. However, in the High-to-Low condi- tion (n=96 after removing 7 bivariate outliers), neuroticism was not a significant predictor of testing performance {¢ (94)=—1.14, P=0.259, 6=—0.12}. Despite this difference across the two workload conditions, the result of a Fisher's z-test comparing the two cegression coefficients was not statistically significant (z= 0.80, P>0.05), 3. Discussion This study is the first to suggest that the examination of workload history might provide systematic method for testing Eysenck’s theory of neuroticism. It was hypothesized that higher levels of neuroticism would be associated with significant decrements in performance following changes in workload history. Results revealed significant findings confirming this prediction for reaction time in both conditions (High-to-Low and Low-to-High). Indeed, it appears that at least in terms of reaction time, either a sudden increase or decrease in workload can result in a sig- nificant performance decrement for those scoring higher in neuroticism, Regarding the correct response data, individuals scoring higher in neuroticism showed a significant performance decre- ment only in the Low-to-High workload condition. There was no significant association between neuroticism and performance in the High-to-Low condition, although the data did show a trend in the expected direction. ‘These findings add to a growing body of research suggesting that individuals high in neuroti- cism adapt poorly to changing environmental conditions. For instance, researchers have noted that individuals higher in neuroticism report greater levels of psychological distress when coping with job relocation (Moyle & Parkes, 1999), and they adapt more poorly to shift work changes (Akerstedt and Theorell, 1976). Additionally, it has also been found that individuals higher in neuroticism are less likely to change work environments compared to those lower in neuroticism, perhaps because of the aversiveness of changing work environments (for a teview, sec Tokar, Fischer, & Subich, 1998). The findings of the present investigation complement the aforemen- tioned studies in that unlike previous research that has focused on psychological and physi- ological outcome measures, the present inguiry documented the interactive effects of neuroticism and changing workload conditions on behavioral performance indices. While the findings of this study complement previous research and appear supportive of Eysenck’s theory of neuroticism, it should be noted that other factors might also account for these effects. For example, one alternative interpretation of these results is that a vigilance434 L-R, Cox-Fuenzalida etal. | Personality and Individual Differences 36 (2004} 447-456 decrement due to either fatigue or time-on-task, might explain the reduction in performance. However, it is well-documented in the human performance literature that the vigilance decrement is especially sensitive to event rate (Davies & Parasuraman, 1982; Parasuraman & Davies, 1977; See, Howe, Warm & Dember, 1995). That is, according to previous research, fan increase in event presentation (e.g. the Low-to High shift) should reduce perceptual sen- sitivity, resulting in degraded performance (Matthews, Davies, Westerman, & Stammers, 2000). ‘Therefore, if time-on-task or fatigue were significant factors in this experiment, a stronger decrement in performance would be expected moving from a Low-to-High than a High-to-Low workload level due to the difference in event rate between the two conditions. Contrary to this prediction, in the present study the performance decrement moving from base- line to testing was generally consistent across the two workload conditions (High-to-Low and Low-to-High) for both reaction time and correct responses (see Table 2). Consequently, this alternative interpretation based on a vigilance dectement due to lime-on-task or fatigue does not appear applicable. ‘Another interpretation of the present findings can be derived from research on cognitive appraisal mechanisms, ‘This literature suggests that an individual's appraisal of a situation can strongly influence the subjective experience of stress (Lazarus, 1980; Lazarus & Folkinan, 1984). In addition, studies in personality research suggest that neuroticism may be associated with maladaptive appraisal styles. While some individuals are able to handle significant disturbances in stride, others may be anxious when faced with relatively minor challenges (for a review of appraisal and emotion, see Wells & Matthews, 1994), Therefore, individuals who score high in neuroticism, compared with those who score low, might interpret the shift (or change) in work- load more negatively. Such a negative appraisal might adversely affect mood, attention and con- centration, which could then negatively impact performance. Future studies might include a subjective stress questionnaire to evaluate how stressful the sudden change in workload is per- ceived by participants, A final alternative explanation for these results draws upon the relationship between neuroti ism and coping. To illustrate, a change or shift in workload requires @ coping response. The lit erature suggests that in response to stressful events, individuals scoring higher in neuroticism cope less adaptively (e.g, are less task-focused and more emotion-focused) than those scoring lower in neuroticism (Dorm & Matthews, 1992; Endler & Parker, 1990; Matthews et al., 2000; ‘McCrae & Costa, 1986). In fact, participants high in neuroticism often lack confidence in their abilities, and tend to employ coping strategies (e.g. worry) that are likely to impair performance (Wells & Matthews, 1994). In other words, in response to the sudden workload shift, individuals, high in neuroticism might use less adaptive coping strategies thereby resulting in poorer perfor- mance. Future studies could employ measure to investigate differential coping responses to the workload shifts, In conclusion, results of this study suggest that individuals across the neuroticism continuum respond differently to sudden changes in workload level, though more studies are necessary before any firm conclusions can be drawn, While these findings are consistent with Eysenck’s theoretical model of neuroticism, they might also be accounted for by cognitive-behavioral pro- cosses (e.g. cognitive appraisal, coping strategies). Additional work in this area might begin by utilizing a workload history approach to examine these alternative explanations for the effects of neuroticism on performance,LB, Cox Puenzalida et al.| Personality and Individual Differences 36 (2004) 447-456 455 Acknowledgements ‘The authors express their gratitude to Dr. Gerald Matthews and Dr. Kirby Gilliland for their many helpful suggestions. References. Akersiedt, ., & Theorell, T. (1976) Exposuse to night work: serum gastrin reactions, psychosomatic complaints and personality variables. Journal of Psychasomatic Research, 20, 479484 Bakao, P. (1959), Extraversion-introversion and improvement in an auditory vigilance task. British Journal of Psychology, $0, 325-332. Cox-Fuenzalida, L- E. 2000). Eifect of workload history on vigilance performance. Unpublished doctoral dissertation, University of Oklahoma, Norman. Cumming, R. W., & Croft, P. G. (1973). Human information processing under varying task demand. Eygonomics, 76, 581-586, Darke, (1988). Anxiety and working memory capacity. Cognition and Emetion, 2, 15-154 Davies, D, R., & Paraswraman, R. (1982). The psychology of vigilance, Loudon: Acadcmie Press Dom, L, & Matthews, G. (1992). Two further studies of personality correlates of driver stress. Personality and Individual Differences, 13, 949-952. Endler, N., & Parker, J. (1990). Mult-dimensional assessment of coping: a citical review. Jownal of Personality and Social Poychology, 58, 844-854. Eysenck, H. J. (1967). The biological bass of personality. Springfield, IL: Charles C. Thomas. Eyseack, H. I, & Bysenck, M. (1985) Personality and individual diferences. New York: Plenum Press Bysenck, H.J., & Bysenck, $. B, G. (1968). The Eysenck Personality Inventory manual. San Diego: Educational and Industral Testing Services. Eysenck, M. W. (1983). Anxiety, In G. R. J. Hockey (Ed), Stress and fatigue in human performance (pp. 273-295). ‘Chiebester, England: Wiley Bysenck, M. W. (1992). Anziesy the cognitive perspective, Hove, UK: Lawrence Erlbaum Astociates Ltd Bysenck, MW. (1999). Ansiety and cognition: a ified theory. Hove, UK: Psychology Press Bysenck, M. W., & Calvo, M. G. (1992). Anxiety end performance: the processing efficiency theory. Cognition and Brotion, 6, 409-434, Goldberg, D. R., & Stewart, M. R. (1980). Memory overload or expectancy affect? Hysteresis’ revisited. Ergonomics, 23, WYB=1178. Holmes, T. HL, & Rahe, R. H, (1967). The social readjustment rating scale. Jounal of Paychosomatic Research, 1, 213-218. Lazarus, RS, (19§0). The stress and coping paradigm. In C. Bisdorfer, D. Cohen, & A. Kleinman (Es.), Conceptual ‘models for psychopathology (pp. 77-214), New York: Spectrum. Lazarus, R. S., & Fotkman, S, (1984). Siress, appraisal, and coping. New York: Springer. Manly, B. F. J. (1976). Exponential data transformations. The Statistician, 25, 37-2. Matthews, G., Davies, D. R., Westerman, S.J, & Stammers, R. B. (2000). Human performance: cognition, stress and individual diferences. Philadelphia, PA: Payctology Press (part of the Taylor & Francis Group). Matthews, G., & Deary, I. J. (1998). Personality traits, Cambridge: Cambridge University Press, Matthews, M. L. (1986). The infivence of visual workload history on visual performance. Hunan Factors, 28, 623-632. Mayer, R. E, (197). Problem-solving performance with task overload: effects ofself-pacing and trait anxiety. Bullen of the Psychonomic Society, 9, 383-286. MeCrac, R. R., & Costa, P. T. (1986), Personality, coping, and coping effectiveness in an adult sample, Journal of Personality, 4, 385-05. Mork, P., & Parkes, K. (1995), The effects of transitional stress a relocation study, Journal of Organizational Behavior, 20, 625-646.456 L-B. Cox-Puenzalida et al. Personality and Individual Differences 36 (2004) 447-456 Newton, T, Slade, P., Bulle, N. M., & Murphy, P, (1992) Personality and performance on a simple visual search task. Personality and Individual Diferences, 13, 381-382, Parasuraman, R., & Davies, D. R. (1977), A taxonomic analysis of vigilance pevformance. In R. R. Mackie (Ed) Vigilance: theory, operational performance, and physiclogical correlates (pp. 559-574). New York: Plenum Pres. Revell, W., Humphreys, M., Simon, 1, & Gilliland, K. (1980). The interactive effect of personality, time of day, and caffeine: a test of the arousal model, Journal of Experimental Psychology General, 109, 1-26. Schlegel, R. E., & Gilliland, K. (1950). Evaluation ofthe Criterion Task Set-Part ICTS Performance and SWAT Data~ ‘Baselite Conditions (UZ) (Tech. Report AAMRL-TR-90-007 ). Ohio: Wright Paterson. See, J. E,, Howe, 5. R., Warm, J.$., & Dember, W.N. (1995). Meta-analysis ofthe sensitivity decrement in vigilance. ‘Peyehological Bulletin, 117, 230-249, Selye, H. (1956). The stress of ife. New York: McGraw-Hill, Spence, I.T., & Spence, K. W. (1966). The motivational components of manifest anxiety: drive and drive stimuli. Ia C.D. Spielberger (Ea), Anxleny and behavior (pp. 361~398). New York: Academic Press. ‘Tokar, D. M., Fischer, A. R., & Subieh, L. M. (1998). Personality and vocational belavior: a selective review of the Literature, 1993-1997. Joumal of Vocational Behavior, $3, 115-153 ‘Wells, A. & Matthews, G. (1994). Attention and emorion. Hove, United Kingdom: Lawrence Erlbaum Associates