Stressors and Stress Reactions Among University Personnel

International Journal of Stress Management, Vol. 9, No. 4, October 2002 ( 2002)

Stressors and Stress Reactions Among University Personnel Judy M. Hogan,1 John G. Carlson,1,3 and Jagdish Dua2

A large, ethnically diverse, and representative sample of university workers in administrative, instructional, and blue-collar/clerical support categories were administered a battery of questionnaires designed to assess job and nonwork stress, biopsychosocial reactions to stress, emotionality, medical symptoms and utilization, and perceived social support, among other variables. A total of 831 participants returned questionnaires. The principal results showed that job and nonwork stress correlated positively with behavioral, cognitive, and physiological reactions to stress as well as with negative emotionality. Job and nonwork stress correlated meaningfully with medical symptoms; nonwork stress also correlated at a useful level with reported medical utilization. Social support did not generally modulate reports of stress or reactions to stress. It was also found that support staff reported higher levels of nonwork stress and lower levels of work stress, but that 2 measures of job stress did not differentiate administrative and instructional personnel. Younger staff reported higher levels of job and nonwork stress, and females reported higher levels of nonwork stress, irrespective of job category. The results were discussed in terms of their importance for understanding the nature of stress among personnel on the university campus and in relation to prior research. KEY WORDS: job stress; university stress; stress and emotions; stress assessment.

1

University of Hawaii, Honolulu, HI USA. Sydney Stress Management Centre, Australia. 3 Correspondence should be directed to John G. Carlson, Department of Psychology, University of Hawaii, 2430 Campus Road, Honolulu, HI 96822; e-mail: [email protected]. 2

289 1072-5245/02/1000-0289/0  2002 Human Sciences Press, Inc.

290

Hogan, Carlson, and Dua

INTRODUCTION Stress in University Settings Although research on stress in occupational settings has increased exponentially over the past several decades, studies of university faculty and other staff members, such as administrators and support personnel, are relatively fewer in number. Previous research in this area has investigated a number of sources and/or consequences of stress experienced by university personnel (Abouserie, 1996; Gmelch, Lovrich, & Wilke, 1984; Hind & Doyle, 1996, cited in Winefield, 2000; Winefield & Jarrett, 2001). Illustrative is a study by Dua (1994) who investigated the relationship of job stressors, job satisfaction, and physical and emotional health among university personnel. Dua surveyed approximately 2250 employees, of which 551 males and 441 females responded. The instrument utilized was comprised of questionnaires in the areas of demographics, job stressors, general stressors (hassles), and emotional and physical health. The Job Stressors Questionnaire was factor analyzed and the six factors that resulted were used to make comparisons among groups based on gender, job tenure, status, enrollment, disability, age, campus, qualifications, job type, and ethnic background. Although overall levels of job stress were not found to be significantly different between subgroups, specific stress factors were. Males reported more workload stress than females, and females reported more stress due to work politics than males. In general, younger staff reported more job stress than older staff, and staff at lower levels reported higher levels of stress overall. One general conclusion resulting from Dua’s analyses was that supervisory status and job type were the two most common and significant predictors of stress. Other results from the study supported research by others showing that high job and nonwork stress were associated with more job dissatisfaction, psychological distress, negative affect, manifest anxiety, absence from work due to illness, doctor visits, and physical ill health. In a recent review of the literature in this subfield of occupational stress (Winefield, 2000), it was concluded in part that: (a) academic stress levels have increased in the past 15 years; (b) academic stress is high compared with other occupations (but see Spielberger & Reheiser, 1994, who reported similar levels of stress among several occupations surveyed); (c) perceived sources of stress in academe are similar across disciplines, although untenured and female academics may report higher levels; and (d) there are differences in levels of stress in academic institutions in different countries (specifically, the United Kingdom, the United States, and Australia). While most previous studies have focused on faculty only, among exceptions is a study by Blackburn, Horowitz, Edington, and Klos (1986) who investigated the relationship between job strain and quality of life indicators among

Stress Among University Personnel

291

nonacademics, high level administrators, and faculty from humanities and natural science departments at the University of Michigan. Administrators showed a significant relationship between job strain and supervisor satisfaction, whereas faculty appeared to manifest job strain in the form of poorer life satisfaction, slightly higher stress-related symptomatology, and greater number of days ill (see also Winefield and Jarrett, 2001).

The Problem The present study was designed as a partial replication of that by Dua (1994) in Australia on a large campus in the USA and with modifications based on theoretical considerations mainly in the form of additional questionnaires. The approach was based on a biopsychosocial model (Carlson, Harrigan, & Seeley, 1997) and involved investigating relationships that may exist between reported levels of job stress, nonwork stress, social support, physical health, behavioral, cognitive, and physical symptoms of stress, emotional reactions, and coping styles. As Dua previously summarized, correlates of stress in the workplace include: poorer emotional health (psychological distress), depression and anxiety; poorer physical health (e.g., insomnia, headaches); and organizational symptoms (including lower job satisfaction, lower productivity, and poor work quality). For the present study and in accordance with the model (Carlson et al., 1997), it was posited that participants would: (a) identify various events in their environments (both on and off the job) as stressful based on demographic variables (including age, gender, and others beyond the scope of this paper), employment status, and cognitive appraisal of events, among other variables; (b) in turn, they would report related behavioral and cognitive reactions, adverse physical changes, and/or negative emotional effects, possibly moderated by such factors as social support on the job; and (c) report that consequences of reactions due to stress were affected by attempts to cope. (While the theoretical model provided a general heuristic for this research in terms of overall conceptualization and measures that were examined, no specific tests of the model were intended or performed.) Besides providing a replication of some important features of Dua’s (1994) previous study of a university in another country, the current design allowed for comparisons among faculty, executive- and managerial-level administrators, and university support staff (that is, blue-collar and clerical workers). By contrast with some earlier studies in this area, a potentially very large N was available, as indicated below. A battery of questionnaires also enabled us to examine reactions to stress in ways that have not often been considered, including the biopsychosocial dimensions indicated above, and emotionality as measured more precisely by way of a standardized personality inventory. Additionally, it was

292


possible to examine the factor structures of the job stress questionnaires as well as the frequencies of stressors that were identified in order to gain further knowledge and make comparisons among stress-relevant variables in the university setting. Most importantly, the following hypotheses consistent with the conceptual model (Carlson et al., 1997) and previous research, especially the Dua (1994) study, were considered: 1. Reports of stressors within and outside of the campus work environment would correlate positively with reported stress-related behaviors, cognitions, and physiological reactions. 2. Reports of stress would correlate positively with personality measures of emotionality; in particular, measures of the “negative” emotions, anger, anxiety, and depression. (Assessment of a positive emotion, curiosity, also provided a indicator of discriminative validity within the emotion assessment instrument.) 3. Reported stress and stress reactions would relate positively with reports of physical symptoms and medical utilization. 4. Reported levels of social support, in terms of both frequency and satisfaction, would correlate negatively with reports of stress-related behaviors, cognitions, and physiological reactions. Additional hypotheses that were tested with these data based on previous findings (especially, Dua, 1994) are as follows: 5. University job type would be a significant predictor of stress. 6. Younger university staff would report higher levels of job stress and nonwork stress than older staff. 7. Females would report more job stress in the university setting than males.

METHODS Participants Faculty and staff at the University of Hawaii at Manoa in 1998 served as participants. According to statistics provided by the University administration, the population of faculty and staff at the time of the study consisted of 3,579 in four major occupational categories: Executive/Managerial (N = 130), Administrative/Professional/Technical, APT’s (N = 1002), Tenured and Tenured Track


293

Faculty (N = 1326), and Civil Service Employees (N = 1121). The first two categories were combined into “administrative,” for consistency and N size. The population was comprised of 53% males, 47% females; 37% White, 36% Japanese, 10% Korean/Chinese, and 17% Other/Mixed. A total of 831 questionnaires were returned; it is not possible to calculate the return rate because of the method of distribution, which left uncontrolled the actual number of personnel who were given questionnaires of the number sent to campus departments (see below).

Procedures The initial announcement for this study was made through the University Wellness Newsletter, which was distributed via intercampus mail to each department on campus. The announcement stated that the Wellness Office was conducting a study on stressors, stress reactions, and general health within the university, and requested the assistance of the faculty and staff through their participation. The questionnaires, along with a cover letter requesting anonymous participation in the project and a coupon for a free soft beverage, were distributed in bulk packages via intercampus mail to all campus units and departments. A letter from the Chairperson of the Wellness Council on campus to the Dean, Chair, or Director of each unit was included in each bulk packet requesting distribution to all faculty or staff in the unit. Actual distribution of the questionnaires was at the discretion of the department. A deadline was indicated for return of the completed questionnaires via intercampus mail to the principal investigator. Questionnaire data for the sample were coded and entered into an SPSS computer analysis program.

Instruments: General A model for the format of the questionnaire was provided by that previously used on the University of New England, Australia, campuses by Dua (1994). The present questionnaire was designed to include several of the essential components of the Dua instrument for comparison purposes. The final questionnaire consisted of nine separate sections, as described in the following subsections, along with scoring procedures. The questionnaire was contained in a single booklet to encourage continuation to each succeeding section.

294


Instruments Demographics Demographic information requested included gender, age, marital status, number of children, years at the university, employment status and title, primary ethnicity, height, and weight. Also included were six general lifestyle questions assessing self-perception of weight, diet, alcohol consumption, smoking, exercise, and sleep (based primarily on Belloc & Breslow, 1972), that will only be considered in the demographics section of this article, as they were not found otherwise to be useful.

Stressors and Support Job Stressors Questionnaire (JSQ). The Job Stressor Questionnaire (JSQ) was constructed by Dua (1994) in his study based on organizational stressors identified in previous research (Matteson & Ivancevich, 1987; Sutherland & Cooper, 1988). The JSQ as used by Dua contained 21 items concerning such issues as job satisfaction and significance, job clarity, job feedback, working conditions, workload, and job security. (One item, having to do with amalgamation of universities in Australia, was omitted from the current study due to its irrelevance.) Items were scored on a 5-point scale (1, Never; 2, Rarely; 3, Sometimes; 4, Often; 5, Always) indicating the extent to which each item applied to the job over the last 3 months. An average score for the 20 items was derived by reverse scoring of appropriate items then calculating a mean score for each participant. A higher score indicated greater reported job stress. Other characteristics of the questionnaire were reported by Dua. Job Stress Survey (JSS). Job stress was also measured using the Job Stress Survey (JSS, Spielberger, 1991). The original survey was designed to assess both the perceived severity and frequency of 30 working conditions considered likely to adversely affect employee psychological well-being. For purposes of this study, only the frequency scale was utilized, in consideration of time demands and comparisons with Dua’s (1994) JSQ. Spielberger and Reheiser (1994) made use of the original JSS to examine the perceived work stressors of 1,781 working adults (922 females, 859 males) employed in university, military, and corporate settings. In the current study, a mean score on the 30 frequency items was calculated for each participant. A higher score indicated greater frequency of reported job stressors. Reliability and validity characteristics of the JSS were reported by Spielberger. Nonwork Stress (Hassles). Nonwork stressors, defined as “hassles” (health, work, financial, family, social, environmental) were assessed using a six item


295

questionnaire devised by Nowack (1990) to indicate the level of frequency that each occurred. (1, Never; 2, Rarely; 3, Sometimes; 4, Often; 5, Always). Although Dua (1994) utilized all six items as an indicator of general stress, for this study we chose to analyze the five items that did not include “work hassles” in order to provide an indicator of nonwork stress that otherwise overlapped the scale used by Dua. A mean score was calculated using these items. Characteristics of the questionnaire are reported by Dua. Social Support. Social support was determined by the reported frequency and satisfaction levels of support perceived from bosses, coworkers, family, significant others, and friends/neighbors (1, Never or not at all; 2, Rarely or slightly; 3, Sometimes or moderately; 4, Often or very; 5, Always or extremely). The items were adopted from Dua (1994) without modification. Mean scores were calculated for each of the frequency and satisfaction scales. Higher scores indicated greater reported frequency and satisfaction levels with respect to social support. Characteristics of the questionnaire are reported by Dua.

Reactions to Stress Physical Health. Participants reported on three aspects of their physical health: absence from work, frequency of visits to a medical doctor, and overall rating of their general health (which was not a useful measure and will not be reported below). In addition, participants were asked to report on the frequency of illness or symptoms in seven categories (Injuries/Accidents, Infections, Respiratory, Gastrointestinal, Vascular/Neurological, Cardiovascular, and Miscellaneous). Frequency was rated on a 5-point scale (1, No Times; 2, One Time; 3, Two Times; 4, Three Times; 5, Four or More times). A mean medical symptom score was derived (see Dua, 1994). Medical utilization, as indicated by the two questions regarding absence from work and frequency of visits to a medical doctor, was also expressed in a mean score. On these scales, a higher score indicated more reported medical symptoms and greater use of the medical system, respectively. Characteristics of the physical health questionnaire were reported by Dua. Distress Reactions. A scale designed to measure psychological distress (Nowack, 1990) was modified from the original 57-item scale to a 55-item scale in Dua (1994). The present study further refined that scale to 40 items: 10 items indicating behavioral reactions (e.g., making mistakes, avoiding contact with others); 10 items indicating cognitive reactions (e.g., feeling confused, trouble concentrating); and 20 items assessing physiological reactions (e.g., headaches, nausea), associated with psychological distress. These items were rated on a 5point scale indicating frequency of occurrence (1, Never; 2, Rarely; 3, Sometimes; 4, Often; 5, Always). Three subscale means for the two 10 item and

296


single 20-item scales were calculated. Higher scores on each subscale indicated greater reported frequency of stress reactions in that domain. Characteristics of the questionnaire are reported by Dua. State-Trait Personality Inventory (STPI). The STPI scale (Spielberger, 1995) was incorporated. The original scale consisted of 80 items that represented three subscales; anger, anxiety, and curiosity items, the first two categories of which were derived from the State-Trait Anxiety Inventory (STAI; Spielberger, Gorsuch, & Lushene, 1970) and the State-Trait Anger Expression Inventory (STAXI; Spielberger, 1988). A fourth subscale, depression, was added by Spielberger (1995). Four subscale totals were determined for anxiety, anger, curiosity, and depression, respectively. Higher scores on each of the subscales indicated a greater frequency of identification with that personality factor. A recent study conducted with maintenance workers at the University of Hawaii (Peters & Carlson, 1999) revealed that treatment subjects showed an increase on the curiosity subscales as they reduced their health risk behaviors. At the time of the current research, the STPI was undergoing reliability and validity assessment by Spielberger and colleagues. Coping with Stress. A coping questionnaire was devised by the current authors to evaluate various methods of coping. Based on a four factor theoretical model of coping styles, the questionnaire was intended to explore active and passive styles of both cognitive and behavioral coping strategies. Owing to space limitations, this portion of the questionnaire was not further analyzed for purposes of the present study.

RESULTS Descriptive Information Descriptive information requested of the participants included gender, age, marital status, number of children, years at the uniiversity, employment status and title, primary ethnicity, and six general lifestyle questions concerning weight, diet, smoking, exercise, sleep habits, and alcohol consumption. The main demographic and lifestyle results are summarized in Table 1. (Differences between the number of questionnaires returned, N = 831, and entries in this and other tables reflect missing data.) As seen in the table, a majority of the participants were female (58%), married (64%), and permanent (62%), full-time (94%) employees at the University for over 10 years (43%). A majority reported having no children (58%). Participants’ ages ranged from 20 to 75, with a mean age of 45. For comparison, participants were separated into age groups of 20 to 40, 41 to 50, and over 50, which yielded similar N sizes, as shown. In addition, participants were catego-


297

Table 1. Demographics of the Participants (N = 831)1

Gender Age Marital Status Children Religion Years at University

Ethnicity

Employment

Job Category

Health Orientation

Male Female 20 to 40 years 41 to 50 years 51 and older Married Not married Has children No children Practicing Nonpracticing Less than 1 year 1 to 5 years 6 to 10 years More than 10 years Caucasian Japanese/Okinawan Chinese/Korean Other/Mixed Full-time Part-time Permanent Temporary Administrative Instructional Support Staff Consider yourself overweight? Try to eat a lowfat diet? Do you smoke? Exercise at least 3 times per week? Get good quality sleep? Consume Alcohol?

N

%

345 482 220 241 209 533 290 333 485 476 293 34 205 196 356 330 260 93 148 777 51 512 74 229 309 266 Yes % 51.0 69.8 10.8 54.9 54.5 65.5 (Seldom or moderate)

41.5 58.2 32.8 36.0 31.2 64.1 34.9 40.1 58.4 57.3 35.3 4.1 24.7 23.6 42.8 39.7 31.3 11.2 17.9 93.5 6.1 61.6 8.9 27.6 37.2 32.0 No % 47.9 29.6 88.4 44.3 43.9 (None) 34.2

1

Differences between numbers of returned questionnaires and tabled Ns reflect missing data.

rized by job title into the following groups: (a) Administrative—consisting of dean, department chair, director, and administrative support (APT); (b) Instructors—consisting of full, associate, or assistant professors, specialists, or researchers; and (c) Support Staff—consisting of both supervisors and nonsupervisors in clerical and nonclerical domains. As seen in Table 1, these categories again yielded similar (and substantial) N sizes. The ethnic distribution of the population at the University was mirrored by the sample distribution indicating a high degree of representativeness in terms of percentages: Caucasian, 36.7% (University) vs. 39.7% (sample), respectively;

298


Japanese/Okinawan, 36.3% vs. 31.3%, respectively; Chinese/Korean, 10.1% vs. 11.2%, respectively (with “other” making up the small remainder). The lifestyle items yielded responses that were fairly consistent with practice of healthy behaviors by a majority of respondents, with the exception of the self-perception of being “overweight,” which suggested that the majority of participants had a negative view of themselves on this dimension.

Factor Analyses and Frequencies: The Job Stress Measures Our use of two job stress measures validated their use even to the extent of their underlying factor structures and enabled us also to assess validity of the Dua (1994) Job Stressor Questionnaire through correlation with the welldocumented Job Stress Survey (Spielberger & Reheiser, 1994). Factor analyses (varimax and promax) were performed. In brief, the results of these analyses were consistent with previous factor analyses of the instruments by their authors. Analysis of the JSQ revealed the same six factors reported by Dua (1994). Most importantly, Factor 1, Job Significance, and Factor 2, Workload, together accounted for 36% of the variance (26% and 10%, respectively), versus 31% in Dua’s analysis (20% and 11%, respectively), with considerable item overlap. The remaining four factors accounted for just 24% of the total variance in both our and Dua’s analyses, with highs of 7% of the total variance for any factor in both studies. Analysis of the JSS yielded also the two principal factors repeatedly obtained by Spielberger and his colleagues: Factor 1, Job Pressure, and Factor 2, Organizational Support. In our analysis, Factor 1 accounted for 32% of the variance; Factor 2 accounted for 8%. Again, there was considerable item overlap with those indicated by the instrument’s authors (Spielberger & Reheiser, 1994). The 10 highest frequency stressors reported on the JSS by employees in each job category are shown in Table 2. There were substantial commonalties as well as differences among the job categories. For all job categories, Item 23 (Frequent interruptions) ranked first or second; Item 26 (Meeting deadlines) ranked second or third; Item 25 (Excessive paperwork) ranked third or fourth; Item 19 (Inadequate salary) ranked second (for clerical/support staff), fourth, or fifth. It is also apparent among the 10 most frequent job stressors that there were a larger number in common for the administrative and instructional categories (8 stressors) than the clerical/support category (6 stressors). In the latter job category by comparison with the other two categories, stressors that appeared uniquely among the first 8 of the 10 most frequent were Item 11 (Assignment of increased responsibility) and Item 3 (Lack of opportunity for advancement). While there was some commonalty with the academic stressors reported by others (USA, Gmelch et al., 1984; UK, Hind & Doyle, 1996; NZ, Boyd &


299

Table 2. Mean Frequencies of Job Stressors (JSS) Job category Item

Administrative

23. Frequent interruptions 26. Meeting deadlines 25. Excessive paperwork 19. Inadequate salary 2. Working overtime 16. Making critical on-the-spot decisions 15. Insufficient personnel to adequately handle assignments 7. Dealing with crisis situations 9. Performing tasks not in job description 28. Covering work for other employees 14. Experiencing negative attitudes toward organization 10. Inadequate or poor quality equipment 11. Assignment of increased responsibility 3. Lack of opportunity for advancement 24. Frequent change from boring to demanding activities 29. Poorly motivated co-workers

Instructional

Support

6.55 6.40 5.99 5.68 5.48 5.01

(1) (2) (3) (4) (5) (6)

6.06 5.99 5.50 4.94 6.53 4.55

(2) (3) (4) (5) (1) (8)

5.58 4.81 4.62 4.99 2.89 3.15

(1) (3) (4) (2) (*) (*)

4.84 4.72 4.69 4.35

(7) (8) (9) (10)

4.58 3.99 3.70 4.43

(7) (*) (*) (10)

3.49 3.02 3.62 2.81

(7) (*) (5) (*)

4.20 3.06 4.14 4.18

(*) (*) (*) (*)

4.63 4.50 4.02 3.07

(6) (9) (*) (*)

3.25 2.89 3.62 3.45

(*) (*) (6) (8)

3.97 (*) 3.75 (*)

2.67 (*) 3.08 (*)

3.44 (9) 3.34 (10)

*Not among top 10 stressors for this category.

Wylie, 1994), there also were a number of differences that may owe to any number of procedural differences among these studies, not the least of which may have been our choice of job stress assessment instrument (the JSS) for this analysis.

Correlations A .05 level of significance was chosen for all statistical tests. With regard to the hypotheses concerning variables that would be meaningfully correlated, these were: 1. Reports of stress within and outside of the campus work environment would correlate positively with reported stress-related behaviors, cognitions, and physiological reactions. 2. Reports of stress would correlate positively with measures of negative emotions. 3. Reported stress and stress reactions would relate positively with reports of physical symptoms and medical utilization.

300


4. Reported levels of social support would correlate negatively with reports of stress-related reactions. Correlations among the major variables are shown in Table 3. Owing mainly to the large sample, most of these correlations were statistically significant. However, those correlations that were moderate or better (above .20), and therefore potentially useful/meaningful, were relatively fewer in number. These included the two measures of job stress with one another (r = .61) and the correlations between the measures of job stress and self-reported behavioral, cognitive, and physiological symptoms of stress, reported medical symptoms (especially, gastrointestinal, neurological, and miscellaneous illnesses), and all measures of negative emotionality. Reports of nonwork (“hassles”) stress were also moderately (or more highly) correlated with reactions to stress, medical symptoms and utilization, and negative emotionality, as well as with the two measures of job stress. Therefore, with regard to Hypothesis 1, reported stress (both job stress and nonwork stress) was meaningfully correlated with the various measured dimensions of stress, supporting this hypothesis. Hypothesis 2, that reported stress would correlate with negative emotionality, was also supported for both job stress and nonwork hassles. Interestingly, a measure of a positive emotion, “curiosity,” generally showed the reverse effect (that is, correlating negatively with most indicators of stress or its consequences), consistent with apparent discriminative validity of the emotion scales on the STPI. With regard to Hypothesis 3, reported stress levels on the job correlated meaningfully with medical symptoms (r = .31, .33, respectively) but not so highly with reported medical utilization (r = .19, .16, respectively). On the other hand, reported nonwork stress correlated meaningfully with both medical symptoms (r = .37) and medical utilization (r = .30). Thus, Hypothesis 3 was partially supported. With regard to Hypothesis 4, it is notable that, generally, measures of social support (satisfaction and frequency) did not correlate at useful levels with the three dimensions of stress reactions, nor with reported job stress or nonwork stress (most correlations were below .20, although statistically significant). One exception of note was a moderate negative correlation between Dua’s measure of stress on the campus and social support satisfaction (r = −.30). Thus, Hypothesis 4, postulating relationships between measures of social support and measures of stress, was not generally supported by these data.

Analyses of Means The additional hypotheses that were tested with these data were: 5. University job type will be a significant predictor of stress; 6. Younger staff will report

—

1 Job Stress—JSQ 2 Job Stress—JSS 3 Nonwork Stress 4 Behavioral Rxs 5 Cognitive Rxs 6 Physiological Rxs 7 Medical Symptoms 8 Medical Utilization 9 Soc. Support Satisfaction 10 Soc. Support Frequency 11 STPI–Anger 12 STPI–Anxiety 13 STPI–Curiosity 14 STPI–Depression

.61 —

2 .35 .34 —

3 .43 .42 .62 —

4

All unmarked correlations are significant at the 0.05 level. *Nonsignificant (p > .05).

1

1

Variables .42 .41 .59 .82 —

5 .34 .28 .57 .67 .63 —

6 .31 .33 .37 .37 .36 .53 —

7 .19 .16 .30 .25 .22 .37 .38 —

8 −.30 −.17 −.13 −.21 −.17 −.10 −.03* −.06* —

9

Table 3. Correlations Among the Variables (N = 819)1

.02* .17 .17 .14 .14 .18 .14 .11 .46 —

10 .34 .24 .40 .47 .46 .39 .22 .15 −.18 .04* —

11

.40 .27 .51 .64 .72 .47 .30 .22 −.24 .03* .52 —

12

−.21 −.03* −.24 −.32 −.29 −.27 −.13 −.17 .24 .08 −.18 −.42 —

13

.44 .31 .48 .63 .66 .44 .32 .16 −.30 −.02* .46 .85 −.45 —

14

Stress Among University Personnel 301

302


higher levels of job stress and nonwork stress than older staff; 7. Females will report more job stress than males. To test the significance of the differences between means, a multivariate analysis of variance (MANOVA) was performed (using SPSS, V 7.0) relevant to all the variables of major interest, as indicated below. (The total N of 831 was adjusted by the computer program as necessary owing to missing data, such as blanks on the questionnaires, for most of the variables.) Scheffe’s tests, when appropriate, were conducted on specific mean differences. Again, the rejection level was set at .05 for all tests.

Job Category The means by job category for reported job stress (on both the Dua and Spielberger questionnaires) and nonwork stress are shown in Table 4. The table also shows means by job category for the three dimensions of stress reactions: behavioral, cognitive, and physiological. No effect was obtained for Dua’s JSQ measure of job stress by job category, F (2, 624) = .82, despite a somewhat higher reported level of stress in the administrative category by comparison with the instructional and support categories (see Table 4). Notably, however, a significant effect for job stress was found on the Spielberger JSS measure, F (2, 624) = 6.41. A Scheffe’s analysis using Spielberger’s measure indicated that support staff reported significantly lower levels of job stress than both administrators and instructors, who did not differ statistically from one another (see Table 4). Similarly, a main effect was found for job category by nonwork stress, F (2, 624) = 2.95. Scheffe’s tests showed this effect to be due to higher levels of reported nonwork stress in the support category relative to both the administrative and instructional categories, which did not differ statistically from one another. In addition, in comparing stress reactions, a significant effect was found only in the category of reported physiological reactions, F (2, 624) = 9.79. Scheffe’s tests indicated that this effect owed to higher reported levels among the support staff by comparison with the administrative and instructional staff, which did not differ statistically from one another. (No effects were obtained in the measurement of behavioral or cognitive stress responses by job category.) In short, Hypothesis 5 was partially supported by these data, albeit not in expected directions: University support staff generally indicated higher levels of perceived stress off the job and higher levels of reported stress reactions, notably physiologic, by comparison with other job categories. By contrast, these support workers also indicated lower stress on the job, as measured by the Spielberger JSS. Interestingly, instructional staff (mainly faculty) did not differ significantly from administrative staff on any of our measures of stress or stress reactions.

M 2.57 2.51 2.52

Job category

Administrative Instructional Support

.50 .50 .47

SD

Job Stress—JSQ

3.84 3.53 3.08

M 1.81 1.79 1.88

SD

Job Stress—JSS

Stress

2.48 2.37 2.69

M .76 .73 .79

SD

Nonwork

23.81 22.78 24.92

M

6.41 6.79 6.70

SD

Behavioral

23.94 23.05 24.77

M

7.62 7.72 7.83

SD

Cognitive

Stress reactions

Table 4. Reported Levels of Job Stress, Nonwork Stress, and Stress Reactions by Job Category

36.22 33.89 40.72

M

10.67 9.62 11.77

SD

Physiological


304


Age and Stress The means by age category for reported job stress (on both the Dua and Spielberger questionnaires) and nonwork stress are shown in Table 5. The table also shows means by age category for the three dimensions of stress reactions. It is apparent in the table that the older university staff members (age 51 and up) reported less stress both on and off the job. Significant effects for age were found on both the Dua, JSQ, and Spielberger, JSS, measures, F (2, 624) = 12.55 and 11.93, respectively, and on the nonwork stress measure, F (2, 624) = 4.31. In follow-up Scheffe’s tests on the job stress and nonwork stress measures, the two younger age categories, 20–40 and 41–50, did not differ from one another, whereas each of these categories differed significantly from the age 51+ category on both reported job and nonwork stress. Similarly, in terms of stress responses, a significant effect for age was found on for both behavioral and cognitive reactions, F (2, 624) = 5.30 and 4.79, respectively; no effect was found on the assessment of physiological reactions. Scheffe’s tests showed that the significant overall effects were due to significantly higher reported behavioral and cognitive reactions in the two younger age categories, which did not differ significantly from one another. Therefore, Hypothesis 6, was generally supported—that younger workers (in this case, those aged 20–50) would report more stress than older workers.

Gender and Stress The means for gender by reported job stress (on both the Dua and Spielberger questionnaires) and nonwork stress are shown in Table 6. The table also shows means for gender by the three dimensions of stress reactions. While there were no significant gender effects in the measures of job stress, females reported significantly higher nonwork stress than males overall, F (1, 624) = 7.36. This same pattern was repeated on all the dimensions of reported stress reactions: behavioral, F (1, 624) = 14.70; cognitive, F (1, 624) = 5.35; and physiological, F (1, 624) = 10.97. Therefore, there was no support for Hypothesis 7, that females in the university work setting would report higher levels of stress.

Job Category, Age, and Gender: Interactions with Stress With two minor exceptions, there were no significant two-way or threeway interactions in the MANOVA between job category, age, or gender with respect to any of the measures of job stress, nonwork stress, or stress reactions.

M 2.58 2.60 2.35

Age

20–40 41–50 51+

.47 .48 .47

SD

Job Stress—JSQ

3.52 3.75 3.00

M 1.83 1.87 1.74

SD

Job Stress—JSS

Stress

2.58 2.60 2.36

M .71 .76 .84

SD

Nonwork

24.69 24.30 22.46

M

6.36 6.87 6.13

SD

Behavioral

25.05 24.11 22.46

M

8.01 7.40 7.02

SD

Cognitive

Stress reactions

Table 5. Reported Levels of Job Stress, Nonwork Stress, and Stress Reactions by Age

37.40 37.35 35.56

M

11.32 11.01 10.95

SD

Physiological


M 2.53 2.52

Gender

Female Male

.47 .50

SD

Job Stress—JSQ

3.47 3.41

M 1.87 1.81

SD

Job Stress—JSS

Stress

2.61 2.36

M .78 .75

SD

Nonwork

24.69 22.41

M

6.53 6.73

SD

Behavioral

24.55 22.80

M

7.52 7.86

SD

Cognitive

Stress reactions

Table 6. Reported Levels of Job Stress, Nonwork Stress, and Stress Reactions by Gender

38.58 34.40

M

11.55 9.85

SD

Physiological

306 Hogan, Carlson, and Dua


307

This greatly simplified interpretations of the effects of these variables outlined above.

DISCUSSION The main results obtained in this study were as follows: Notably, most of the correlations among the variables in the study were statistically significant and a number were also useful/meaningful in terms of predictability (r ≥ .20), enabling us to draw some important generalizations regarding stress and its impact on this large university community. In terms of hypotheses regarding correlations: First, Hypothesis 1 was supported; reported stress on and off the job was meaningfully correlated with stress-related behaviors, cognitions, and physiological reactions. Second, Hypothesis 2 was supported: Job and nonwork stress (and stress reactions) correlated meaningfully and positively with measures of the emotions, anger, anxiety, and depression (but negatively with curiosity, considered a positive emotion). Third, Hypothesis 3 was partially supported: Job and nonwork stress also correlated with medical symptoms. Nonwork stress correlated at useful levels with both medical symptoms and reported medical utilization. Fourth, Hypothesis 4 was not supported: Job stress and nonwork stress did not generally correlate at useful levels with social support satisfaction or frequency (with one exception being Dua’s, 1994, measure of job stress, which correlated with social support satisfaction). In terms of attempts to replicate previous results obtained by Dua (1994) regarding stress variables in a university population: Fifth, Hypothesis 5 was partially supported; university job classification was a significant predictor of stress. However, by contrast with Dua’s results (see also, Blackburn et al., 1986; Winefield & Jarrett, 2001), there were no differences between administrative and instructional staff in terms of reported stress levels. On the other hand, the blue collar plus clerical workers reported higher nonwork stress levels relative to the more white-collar university job categories. Sixth, Hypothesis 6 was supported: “Younger” staff at the University (50 and under) reported higher levels of stress both on and off the job relative to their older counterparts (in all job categories). Last, Hypothesis 7 was not supported: Contrary to Dua’s (1994) findings, there were no gender differences in reported job stress (but see Winefield & Jarrett, 2001, who report similar effects). However, female staff members did report higher nonwork stress than male staff. Therefore, with some notable exceptions, these results in a large university in the United States generally replicated those obtained by Dua (1994) in an

308


Australian university. This outcome is all the more notable in that the university population in the current study was geographically distinct and culturally very diverse. By contrast with Dua’s findings, the failure to obtain significant differences between administrative (supervisory) and instructional personnel in reported stressors is not readily understandable but may reflect some inherent differences in the organizational structures of the two universities themselves. In our study, there was considerable overlap in the most commonly cited stressors in these two job categories, suggesting that these groups of employees were responding to the same stressful job variables (and apparently in much the same manner). Age was an important variable in both this and Dua’s (1994) study, and in the same direction with older staff reporting less stress (see also, Winefield & Jarrett, 2001). However, the age at which the effect emerged was somewhat higher in our study (above 50). Another important result, the absence of gender differences in reported job stress in the present study, was not consistent with Dua’s results, which indicated higher job stress among male participants. On the other hand, our gender results were consistent with previous studies in which no differences were found between reported stress in men and women on university staffs (Abouserie, 1996; Winefield & Jarrett, 2001). Moreover, it is notable that women reported more stress off the job in the present study, perhaps reflecting (a) the frequently reported conflicts experienced by working women attempting to perform in both professional and home settings (Emmons, Biernat, Tiedje, Lang, & Wortman, 1990), and/or (b) the observation that working women are more affected than men by stressors in the home (Barnett, Raudenbusch, Brennan, Pleck, & Marshall, 1995). Other results of the present study were both suggestive and gratifying along a number of dimensions. For one, Dua’s instrument for assessing job stress in a university setting proved valid both in terms of content, by correlation with Spielberger’s (1991) well-documented JSS instrument, and in terms of predictiveness with regard to many other variables in the study, such as stress reactions and medical symptoms. Moreover, the factor structure of the instrument proved reliable by comparison between our results and Dua’s analyses. For yet another, our somewhat unique biopsychosocial division of measures of stress reactions and addition of standardized measures of negative and positive emotionality proved to be useful ways for assessing the impact of stress in this population. By contrast, the general failure of the social support measures to correlate in anticipated ways with measures of stress or stress reactions could either reflect the lack of utility of this variable in this setting or, more suggestively, limitations on the instrument itself. The voluminous literature on the role of social support in stress argues not only for its buffering capacity but for its general utility (Cohen, Gottlieb, & Underwood, 2000), as assessed with more effective measures than perhaps used in the present study. Relatedly, the general


309

usefulness of standardized instruments in this study, notably the JSS and STPI, suggests that a standardized instrument for assessing social support also may have been more useful with this university population, a possibility that merits further research. Overall, these data are consistent with studies showing that a university community at all levels—from administrative and instructional to blue-collar support—experiences considerable stress in a variety of forms, and that such stressors impact on a host of reactions—behavioral, cognitive, and physiological—as well as on reported symptoms of physical illness. There were also consistencies between these data and others in showing the disproportionate impact of stress on younger participants, irrespective of gender or job category. Further, these data show the importance of assessing nonwork sources of stress as well, as these were not only moderately correlated with job stress but sometimes even more salient in effect (e.g., in terms of utilization of medical services and impact on female participants). In fact, in absolute correlational terms, nonwork stress was often a better predictor than job stress with respect to many of the reactions to stress reported in this study. In short, stress and its effects were apparent at all levels in this university community indicating the importance of continuing research in this area from the standpoint of both understanding and possibilities for stress management.

REFERENCES Abouserie, R. (1996). Stress, coping strategies and job satisfaction in university academic staff. Educational Psychology, 16, 49–56. Barnett, R. C., Raudenbusch, S. W., Brennan, R. T., Pleck, J. H., & Marshall, N. L. (1995). Change in job and marital experiences and change in psychological distress: A longitudinal study of dual earner couples. Journal of Personality and Social Psychology, 69, 839–850. Belloc, N. D., & Brelsow, L. (1972). Relationship of physical health status and family practices. Preventive Medicine, 1, 409–421. Blackburn, R. T., Horowitz, S. M., Edington, D. W., & Klos, D. M. (1986). University faculty and administrator responses to job strains. Research in Higher Education, 25, 31–41. Boyd, S., & Wylie, C. (1994). Workload and stress in New Zealand universities. New Zealand Council for Educational Research and the Association of University Staff of New Zealand. Auckland, New Zealand. Carlson, J. G., Harrigan, R. C., & Seeley, Z. (1997). The wellness behavior interaction model. International Journal of Stress Management, 4, 145–169. Cohen, S., Gottlieb, B. H., & Underwood, L. G. (2000). Social relationships and health. In S. Cohen, L. G. Underwood, & B. H. Gottlieb (Eds.), Social support measurement and intervention (pp. 3–25). Oxford: Oxford University Press. Dua, J. K. (1994). Job stressors and their effects on physical health, emotional health, and job satisfaction in a university. Journal of Educational Administration, 32, 59–78. Emmons, C., Biernat, M., Tiedje, L. B., Lang, E. L., & Wortman, C. B. (1990). Stress, support, and coping among women professionals with preschool children. In J. Eckenrode & S. Gore (Eds.), Stress between work and family (pp. 61–93). New York: Plenum. Gmelch, W. H., Lovrich, N. P., & Wilke, P. K. (1984). Sources of stress in academe: A national perspective. Research in Higher Education, 20, 477–490.

310


Hind, P., & Doyle, C. (1996). A cross cultural comparison of perceived occupational stress in academics in higher education. Paper given at XXVI International Congress of Psychology: Montreal, Canada. Matteson, M. T., & Ivancevich, J. M. (1987). Controlling work stress. San Francisco and London: Jossey Bass. Nowack, K. M. (1990). Initial development of an inventory to assess any health risk. American Journal of Health Promotion, 4, 173–180. Peters, K. K., & Carlson, J. G. (1999). Worksite health promotion with high risk maintenance workers: A controlled study. International Journal of Stress Management, 6, 21–44. Spielberger, C. D. (1988). Manual for the State-Trait Anger Expression Inventory (STAXI). Odessa, FL: Psychological Assessment Resources, Inc. Spielberger, C. D. (1991). Professional Manual for the Job Stress Survey (JSS). Odessa, FL: Psychological Assessment Resources, Inc. Spielberger, C. D. (1995). State-Trait Personality Inventory. Palo Alto, CA: Mind Garden. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. D. (1970). STAI: Manual for the State-Trait Anxiety Inventory. Palo Alto: Consulting Psychologists Press. Spielberger, C. D., & Reheiser, E. C. (1994). Job stress in university, corporate and military personnel. International Journal of Stress Management, 1, 19–31. Sutherland, V. J., & Cooper, C. L. (1988). Sources of work stress. In J. J. Hurrell Jr., L. R. Murphy, & S. L. Sauter (Eds.), Occupational stress: Issues and development in research (pp. 3–40). New York: Taylor and Francis. Winefield, A. H. (2000). Stress in academe: Some recent research findings. In D. T. Kenny, J. G. Carlson, F. J. McGuigan, & J. L. Sheppard (Eds.), Stress and health (pp. 437–446). Amsterdam: Harwood Academic Publishers. Winefield, A. H., & Jarrett, R. (2001). Occupational stress in university staff. International Journal of Stress Management, 8, 285–298.