You are on page 1of 19
JOURNAL OF PERSONALITY ASSESSMENT, 1992, 58(1}, 160-178 Copyright © 1992, Lawrence Erfbaum Associates, Inc Workaholism: Definition, Measurement, and Preliminary Results Janet T. Spence and Ann S. Robbins University of Texas at Austin Questionnaires were developed to assess the concept of workchatim, defined in rerms of high scores on measures of work involvement end driveness and iow scores on a measure of enjoyment of work, and to contrast this profie with work ‘enthusiasm, defined as high work involvement and enjoyment end low driveness. Adddicionat scales were devised to test several predictions about the correlates 0} workaholsm. A test hattery including these seas was given in a mal survey to 3 mpie of male (n = 134) and female {n= 157) social workers with boscions. The paychomesrc properties of the scles are described. : ast (among other groups) on measures of perfectionism, nondelegation of responsibility, and job stress. They were also. higher on 2 measure of health complaints. Investigations are being initiated determine ea association coe ss As far as can be determined, the cerm workakolion, patterned after the word alcoholism, first eppeared in 2 book by Oates (1971) in which he described his earlier compulsion tc work incessantly and suggested some of the characteristics exhibited by work addicts and the negative consequences brought about by this, addiction. Since that time, the word has become widely known; references to workaholism are becoming common in the press and in everyday speech Industrial-onganizational psychologists, among others, have also written exten- rely on the topic in books and articies aimed at z lay audience (e.g., Boronson, 76: Machio 1974; Minurth, Meier, Wichern, Brewer, & # references to worksholism are ing fo appear in psychology’s prime data journals. Yer, despite its common currency, there is little consensus about the meaning of the workaholism construct beyond its core feature of heavy investment in WORKAHOLISM 161 work. For example, Mosier (1983 defined workaholies simply as those who work, at least 50 hr a week, whereas Machiowitz (1977) stated thet what distinguishes workaholics from ochers is their attitude toward work and not the number of hours they work, Some (e.g-, Oates, 1971) implicitiy defined workaholism in a negative way or ave focused on its deleterious aspects, whereas others (Cansarow, 1979; Machiowitz, 1980) have proposed that workaholics also have positive features, such as pleasure in work and creativity Furthermore, few formal procedures for assessing w: developed. Although several sel ular sources (e.g, Bradley, cited in Machlowits, 1977; Farnsworth, Machlowitz, 1980}, psychometric analyses of their properties are lacking. lariv, oniy occasional investigations of the correlates and consequences of workaholism, however defined and assessed, have appeared in profession journals (e.g., Doerfler & Kammer, 1986}, The popularity of the general notion of workaholism reflects its incuitive appeal end suggests that the topic deserves systematic investigation. As just noted, however, the term workaholism has been used in various ways, and no consensus has emerged about ite precise meaning. Thus, a number of different definitions could legitimately be proposed. The first task of anv investigator is to selece among various possibilities and to offer an explicit definition. In this article, we present our definition of workeholism and describe the series of self-report questionnaires we have developed to identify individ! this and other work-relaced categories. We also present the results ef an exploratory study of a group of male and female social workers with academic appointments who compieted these questionnaires and other scales developed to test several predictions about the correlates of workaholism. As such, these tests constitute an investigation of the validity of this and other work-related constructs as we have defined the holism have been sport questionnaires have appeared in pop- 1987, DEFINITION OF TERMS Asalready noted, the common element in discussions of workaholism is that che affected individual is highly committed to work, devoting « good dea! of time t0 it.! Sometimes, however, itis implied or explicitly stated that, hike alcoholism, workahotim isan addiction: the workaholic fels driven or compeled ro work, not because of external demands or pieasure in work, buc hecause of inner pressures that make the person distressed or guilty about net working. This is the sense in which we have chosen to define the term. More specifically, we and acotudes, At rustive, goeidiected ‘cis impli i= these statements same time, the intent seems to Be £6 i5 behaviors thar may occur avocationally 162 SPENCE AND ROBBINS fine the workaholic as a person whe exhibits three properties: In comparison te others, the workaholic is highly work involved, feeis compelled or driven to work because of inner pressures, and is iow in enjoyment of work. We were of the view, however, that in e correlational sense, these three properties are substantially independent of each other: The person who is highly work involved is not necessarily also driven; nor does such @ person necessarily lack pleasure in work, We have, therefore, chosen to contrast the worksholic with what we have jabeled the work enthusiast. We define the work en a person who, like the workaholig, is highly work involved, but unlike the latter, ic high ix enjoyment and is not driven. We anticipated that other profiles would aiso be found on the three cherac- teristics that we will call, generically, the workaholic triad—degree of work involvement, driveness, and enjoyment of work. However, we decided to et these emerge from the data rather than speculating about them in advance. Although she essentiaily lumped together what we have defined as the workaholic and the work enthusiast into a single category, we agree with Machiowite’s (1977, 1980} contention that intensity of involvement defines these individuals more than the sheer amount of time they devote to work, particularly to job-related activities. We should, therefore, explicate our concep- tion of work involvement. The individual who is highly work involved charac- teristically devotes himself or herself wholeheartedly to productive projects and. prefers to make constructive uses of time. This trait is often expressed in academic activities by students or in occupational activities by those with paid employment. But even in these groups, work involvement may be expressed in, addition or instead in other ways (e.g., hobbies, sports, domestic projects, volunteer work, etc.). Those activities that constitute what we have iabeled work are thus essentially defined as such by the individual, As this statement implies, we distinguish beeween work involvement, « general concepr, and the more specific concept of job (or school) involvement. We anticipate, however, thet the two tend to be correlated and, in practice, are nor siways easy t© distinguish. Work involvement and job involvement are particularly likely co be lated in people in professional and managerial positions or other challenging occupations. CORRELATES OF WORKAHOLISM AND WORK ENTHUSIASM, Based on the observations of other writers as well as our own speculations, we posited thet a number of attributes would distinguish workaholies from work enthusiasts and other individuals. First, we predicted that workaholics would exhibit more perfectionism than others and would also be iess willing to delegate responsibilities. Work enthusiasts might be higher on these characteristics than WORKAHOLISM 163 those who are less work involved, but nonetheless lower than workaholies. We forther anticipated that workaholics ¢but not work enthusiasts} would experi- ence more job stress than others, a feeling of being overwhelmed by their responsibilities and all they expected themselves te do. Finallv, we predicted that worksholism and its associated characteristics would have two major consequences. First, workaholics should repost experi- encing more physical symptoms and illnesses than others, particularly those of a minor nature. Second, we speculated that the performance of worksholics might be of lesser quelity than chat of work enthusiasts (Cherrington, 1980, Taylor & Martin, 1987). In this article, all these predictions were tested except the fatter. Data have yet to be collected relevant to our hypothesis about performance. ASSESSMENT SCALES We developed three selfreport scales~Work Involvement, Driveness, and Work Enjoyment—to identify the workaholic, work enthusiast, and other work-related profiles. Additional self-report scales were developed co assess Job Sires, Job Involvement, Perfectionism, Nondelegation of Responsibility, and Time Commitment. The latter scale was aimed at determining the amount of time devoted to occupations! duties. As described next, preliminary versions of, the scales were first developed and given to college students, the final versions showing satisfactory psychometric properties following revision. Adult versions of these scales (in which references were made to jobs and the work place rather than to schoo! and schoolwork) were administered to the sample. To determine the psychometric properties of the version of the scales for employed adults and to test our predictions about the correlates of workaholism, we conducted a mail survey of a sample of social workers who had academic appointments in colleges or universities. Academic social workers were selected as an example of a relatively homogenous group of individuals employed in demanding profes- siona! positions whose responsibilities are open-ended. That is, their duties are not restricted co set times and places, and there are essentially no limits to the amount of job-related activities thar they could take on. except those imposed b themselves. The social work profession had the additional advantage of hav’ substantial aumbers of both sexes as members. METHOD Respondents A survey booklet was sent to 2 sample of 800 social workers, whe were randomly selected by computer from a National Ascociation of Social Workers data hase 164 SPENCE AND ROBBINS containing the names of all members whe described themselves holding aca- demic appointments. Of chese, 368 returned their questionnaires within 2 months of the posting date. Twenty-one additional surveys were returned after ‘this time bur were too late to be included in the deta analyses. The demographic characteristics of these late respondents were similar to those of the other respondents. The everalt return rate was thus 49%.7 Of the 368 completed questionnaires, 49 were eliminated from further consideration because the respondent had retired or had a predominantly nonacademic position. Re- sponses from 28 individusis who were empioyed only part-time in = 20) or who did not indicate their employment status (n = 8} were aiso eliminated. The final sample thus consisted of 291 individuals (134 men and 157 women}. Median age was 43 vears (interquartile range = 35~49) for the women and 40 {interquartile range 31-47) for the men. Other demographic characteristics of the sample are shown in Tabie L Scale Development Before describing in detail the scales given to the social worker sample, we briefly outline their development. Based on rational considerations, we first wrote sets of eight or more items to assess the workaholic triad (Work Involvement, Driveness, and Work Enjoyment) and the attributes with which we expected these characteristics to be correlated (Job Stress, Job Involvement, Perfection- ism, Nondelegation of Responsibility, and Time Commitment). These items ‘were administered to maie and female students {ns > 130} enrolled in introduc- tory psychology classes ar the University of Texas at Austin. The statements referred to schoolwork rather then to jobs in the sense of paid employment. After examining, separately for each sex, Cronbach alphas, interitem corel tions, and part-whole correlations for the items on each scale, a number of items that displayed nonsignificant interisem and item-totel correlations and whose inclusion lowered the internal consistency coefficient for that scale were dropped or, in some instances, rewritten. The resulting revised and shortened scales were administered to a new sample of male and female students (ns > ). The items on each scale proved to have satisfactory psychometric proper- ties, as decermined by the same set of criteria for retention already described. Except for the substitution of references to *job” for those to *school,” these revised scales were the ones administered to the adult sample of social workers. For financial reasons, the surveys were sent out o€ bulk mail rates, which meant chat nondelivershle surveys were not returned, It was, therefore, impossible ¢9 determine how many surveys were not received because of incorrect addeesse, and thus the actual response rate coul actually be sbove 50%, WORKAHOLISM 165 TABLE Percentages of Male! and Female Soci Workers in Several Demographic Categaries Marital Seu Men Wome: Divorced separated # Widowed [Never married ober af Chir Wane 2 e + 1 Higher: Degree Esme Mer Waren BATES 3 a MA 4G 55 7D 55 # a Wome ‘Associate’ Asstt Profesor x Professor 8 Dean 6 Advice 2 : Other 3 e Survey Booklet In addition to questions aimed at determining the demographic cherocter of the respondents, the survey booklet sent ro the social workers contained the self-report scales described here. Except when indicated otherwise, the items on the various scales were intermingled; each item was answered on 2 S-point scale ranging from strongly agree (1) to strongly disagree (5). Iter responses were scored con a scale ranging from 0 to 4 and summed to vield a tote! score for each scale. The alphas, means, and standard deviations of each sex on cach scale are reported in Table 2 Vork Involvement. The Work Involvement Scale consists of seven items. Sample items are “I like to use my time constructively on and off the job,” “Between my job and other activities 'm involved in, I don't have much free 166 © SPENCE AND ROBBINS: TABLE 2 Alpha Coefficients, Means, Standard Deviations, and Significance of t Ratios Assessing, Differences Between Men* and Women’ on the Work-Related Measures Alpha Means (SD3 Seale Men | Women Mer > ‘Work Involvement R032) oo oe 54 6.2 164 6.2) a Driver, R028) a a B36 763 16 Enjoyment of Work 0-40) 86 86 49 66 nA 038 8 243 65) 058 88 88 209 0.25 0001 ao 89 120 6.5 36 68) R032) 88 8 BS 64 2 Nondelegation: Be 0-28) 8 84 99 G8 96 6.0 pe ‘Nice. The range (A) of possible sores for each scale is nalcated ip parentheses below © Se = 14 tn = 157 time.” and “I get bored and restless on vacations when I haven't anything productive to do.” Driven. The Driven Scale consists of seven items. Sample items are “I feel obligated to work hard, even when it’s not enjoyable,” “I often feel there's something inside me that drives me to work hard,” and “I often find myself thinking about work even when E want to get away from it for awhile.” ale is composed of nine engaged on a project,” lovable,” and “Sometimes i enjoy my work nave 2 hard time stopping.” Enjoyment of Work. The Enjoyment of Work S items. Sample items are “I lose track of time when [ “Most of the time my work is very so much F Job involvement. The Job Involvement Scale contains seven ixems. Sample items are “Much of my satisfaction in life comes from my job,” “My job is & very important part of who ! am,” and “Iam very absorbed in my job.” Job Stress. ‘The Job Stress Scale consists of nine items. Sample items are “I worry about all the work I have to de and whether Pl get it ali done," *I am under ‘2 good deal of stress at work,” and "I often feel overworked.” WORKAHOLISM — 167 Time Commitment to Job. The Time Commitment Scale i composed of seven items. Sample items are “I usually take some job-related work along on vacations,” “I work most nights and weekends,” and“! frequently work unti! fm oe tired to do more.” Two additional questions asked how many hours they worked during the average work week (Monday through Fridav} and weekend. Perfectionism, ‘The Perfectionism Scale consists of eight items. Sample terns are “L am satisfied with nothing short of perfection in my work,” "My high standards sometimes make it difficult for me to get everything done.” and “I's important ro me not to make mistakes, even in small details.” Nondelegation. The Nondeiegation Scale consists of seven stems. Sample ikems are “I double check the work of others so thoroughly cha: delegating saves no time at ali," “When I work on team projects I do most of the work myself,” and “I feel that if vou want something done correctly you should do it yourself.” Heaith Complaints and Behaviors. The Health Complaints Scale, devised by Spence, Helmreich, and Pred (2987), has 22 items, each accompanied by a S-point rating scale, asking about the frequency with which the respondents had experienced various illnesses and symptoms during the past year. These in- cluded questions about sleep, headaches, problems with digestion and elim tion, and respiratory problems. Because of the degree of ternal consistency (Cronbach alphas of .82 for men and .83 for women), a tote! health complaints score was obtained for each individual. Objectively scored questions w asked about smoking and amount and frequency of alcchol consumption. In addition, respondents were asked to list anv medications they were taking on a regular basis, the number and nature of hospitalizations they bad under gone in the past year, and the number and nature of serious health probl they had experienced in the past. However, the sample as a whole was healthy, with the large majority reporting no surgical interventions or serious health problems and not taking prescription medications. The members of the sar also seemed very heaith conscious in regard to smoking and drinking, with the response distributions of these items also demonstrating severe negative skew. Due to the nonnermality of their distributions, these measures were dropped from further analysis, na RESULTS The results of t tests comparing the means of men and women on the vation objectively scored measures are shown on the right side of Table 2. A glance at this table reveals that women displayed significantly higher means that the men. 168 SPENCE AND ROBBINS on the Driven, Enjoyment, Job Stress, Job Involvement, and Time Commit. ment Scales; the sexes did not differ significantly on the Work Involvement, Perfectionism, and Nondelegation scales. Of greater importance chan the mean differences between the sexes are the relationships among the measures within each sex. The correlation matrix is displayed in Table 3. As may be observed, parallel results were found for both sexes. These findings simultaneously demonstrate the similarity of the sexes with respect to the relationships among the variables (as indeed we anticipated they would be) and the stability of the findings across two independent sampies. An examination of the relationships among the three seeles defining worksholism reveals a substantial positive correlation between the Work Involvement and Driven scales and a modest but still significas ive correlation between Work Involvement and Enjoyment. On the other hand, the correlation be- tween Enjoyment and Driven scores was close to 0. Looking at the association of the three workaholic scales with other measures on the personality battery, Work Involvement and Driven were both significantly correlated with all other scales in the expected direction, However, in the case of Work Involvement, the relationship was substantial only with Time Commitment, whereas, in the case of the Driven scale, substantia! relationships (rs > .40) were found with Job Stress, Perfectionist, and Nondelegation. The only strong correlation with Enjoyment, on the other hand, was with Job involvement. These data suggest that when considered independently, the three measures defining workeholism have different patterns of correlaces. ‘Also reported in Table 3 are the corretations of the personality scales with the Health Complaints measure. (High scores indicate more reported symptoms. TABLE 3 Correlations Between Measures for Men* and for Women WI Driven Joy = TCE Stes Perfect, «= NDC wi = 50 48 38 ne Driven 460 = 4 ss 4 8 Jor 2 28 nae -08 te a = 4s 23 aH 3 2 40 7 16 Sees 335 NLS “oH 3 Perfect 3365 foo peer oO naar 2! 46 ND 85 HB -m4 - HC O36 oF 3 (ta, The corelatons for women appear above the dagonsh For mem, a) = 150, andr 01 = 23, For women, df = 155, tq, = 16,79) = 21. WI = Work nvolvments Joy = Enjoyment of Work, TC = Time Commitment; It = Job Involvement; Srewr = job Stes, Perfect = Nondelegation, HC = Health Compaints. 15 WORKAHOLISM 169 Significant positive correlations were found with the Job Stress, Perfectionism, Nondelegation, and driven Scales and significant negative correlations with Enjoyment of Work. Identification of Profiles ‘The correlations among the measures were of less interest than identification: of profiles of scores on the trio of workaholism measures and comparison of the groups exhibiting the various profiles on other variables. The initial step was ro determine whether clusters of individuals would be found whose score proffies corresponded to our definition of the workaholic and the work enthusiast, as well as co determine whether other score profiles would emerge. Scores on each ofthe core characteristics were first transformed into z scores, separately for each sex, in order to standardize the variances of the scales. Cluster analyses were then performed on these standardized measures using Ward's (1963) method. nary analyses suggested that the distance within the clusters could be minimized with the use of a solution that involved six or fewer clusters ‘Therefore. cluster solutions resulting in three, four, five, and six profiles were exemined for each sex. The profiles that emerged in the six-cluster solution, which were basically the seme for men and women, were the most conceptually distinct and easily interpreted. The six-cluster sohition was therefore adopted. Because cluster-analvtic techniques assign all individuals to the profile thev ‘most nearly match, each contained some individuals whe did not acceptably fit \¢ profile on one or more scaies. To remedy this, we set appropriate cutoff : scores for each cluster on the three core scales and removed individuals whose scores on any of these scales fell outside these cutoffs. The scores of these individuals were then examined to see if they fit any of the other profiles; those that did were placed in the appropriate group. Fourteen men and 20 women were redirected in this manner. The 21 men and 34 women whe did not fit into any of the profiles were dropped from the sample. The percentages of men and women who fit each profile after the cufoff scores were applied are reported Table 4. As observed, it was possible to classify 84% of the men and 78% of the women. Also shown in this table are the mean ¢ scores on the three core characteristics for each profile. Inspection of Table 4 shows thar ewo of the clusters correspond to out a priori Prel definitions of the Workaholic profile (above average on Work Involvement and Driven, below: average on Enjoyment) and the Work Enthusiast profile (above A average on work Involvement and Enjoyment, below average on Driven). third cluster, containing individuals we labeled Enthusiastic Workaholics, con bines features of beth~this group is above average on all three measures. The fourth cluster, Unengaged Workers, represents the opposite extreme, its members being below average on all the measures. Members of the fifth cluster, Relaxed Workers, are below average on Work Involvement and Driven, but 176 SPENCE AND ROBBINS TABLE 4 Profile Means (in z scores) and Percentage of Sample Betonging to Each Group Seat ork Percentage ‘Group Name Driver Enjoyment on Semple ‘Work Enthusiasts Men #8 798 sa 8 ‘omen 556 -68 a? n Worksholis Men ig? hss Women S36 1.004 OH B Relaxed Workers Mer 795 692 a 6 = 808 896 % 924 6 » 1a? 3B 5 985 S95 xe 2 19 760 % 7 -785 58 2 16 3 2 above average on Enjoyment. The final cluster, Disenchanted Workers, includes those below average on Work Involvement and Enjoyment, but above average on Driven, ‘Demographic Characteristics of the Profiles The firse set of analyses performed on the clustered deta involved determining the demographic characteristics associated with each profile. These analyses were carried out via chi-square tests of independence and one-way analyses of variance (ANOVAs) for each sex separately. In both sexes, cluster membership *Due ts the smal! frequencies in the divorced, separated, and widowed categories, respondent were classified as currently heing either married or unmarried, For the analysis of educations! level, Individuals were placed in one group if they reported holding PRD and in another group if they reported holding a MA or iess. The academic rank variable wa ais spit into twe levels; individuals whe responded thar they held the rite professor were placed in one group, and associate professors were placed in 2 second group. Respondents wi instead of academic ranks (ce, dean, counselor} were not WORKAHOLISM 171 was found to be unrelated (ps > .05) to marital status, ievel of education, and academic work. ANOVAs carried out on the respondents’ age and number of minor children also indicated no significant differences among the profiles. Comparison of the Profiles on Work-Related Measures The next set of analyses compared the means of the profiles on each of the five work-elated measures by one-way ANOVAs, The mean scores ate reported in, Teble 5 for men and in Table 6 for women. The F ratio for each ANOVA appears beiow the tabled means. All Fe were either significant (p < .05) or approached significance (p < .08).* Due to the differing sizes of the groups we wished ro compare, we employed the Games-Howell Test (Games, Keselman, & Rogan, 1981; Jaccard, Becker, & Wood, 1984), « pairwise comparison recknique that has been demonstrated to be robust and powerful in such a situation, The resuits of the comparisons are also shown in Table 5 and 6. The numbers that appear below each mean indicate the identity of che groups whose means significantly differ from that of the given profile at the .05 level or bevond Inspection of the tables indicates that, for both men and women, Workaholics and Work Enthusiasts did not differ significantly on Job Involvement and Time Commitment, both groups displaying elevated means in comparisons to the remaining groups. Workobolics, however, were significantly higher than Work Enthusiasts on the Job Stress, Perfectonism, and Nondelegetion Sczles, as predicted. Comparisons of the entire set of profiles on the various measures are also of interest. In men, Enthusiastic Workaholics and Disenchanted Workers joined with the Worksholics with respect to the Perfectionism, Job Stress, and Nondelegation Scales, displaving significantly higher means than Work Ene sinsts, Relaxed Workers, and Unengaged Workers. The means of the Disenchanted Workers on these three measures were similar to chose of the Workaholics and Enthusiastic Workaholics; like the ewe latter groups, Disen- chanted Workers scored significantly higher than Work Enthusiasts and Unengaged Workers on Perfectionism and Nondelegation and higher than Relaxed Workers and Unengaged Workers on the measure of Job Stress. These group differences reflect the substantial zero-order correlations berween the Driven Scale and the measures of Job Stress, Perfectionism, and Nondelegation, the three profiies high on the Driven scate differing on these measures from the profiles low on this variable, The pattern of women’s scores on these measures was very similar. On the Time Commitment scale, the trio high in Work Involvement (Work Enthusiasts, Workaholics, and Enthusiastic Workehol “Analysis performed axing clastes membership without the application of cutoff scores vieldedt similar results, as did perforzsing chese analyses with marital status, number of c ighest degree earned, and acad ak included ae covariates 39 > 4 pweayniiis nayip dnos# uasi8 ay saoysojy potleuour, sesqiogg PHULPUDSICE SEO PREP Sooqepegy 2BSeISMUG Uap 20) SOINSEAIR PaeIBY HOM UO SuosTEdUIOL osImaIe PUE SVAONT Jo sHinsay Pue dnosy ye Jo SUED eanave v2 109 Ins op oxi sy od ngre any, a pg muomomony 4 UW M 10} SOANSEAR PAY BOM UO SU dio asimated PUE SVAON' Say pur dnowy qoug jo Sueay, ow 1dVL 173 174 SPENCE AND ROBBINS were found in both sexes ro have significantly higher scores than the three profiles iow in Work involvement. As for the Job Involvement measure, the three groups high in Work Involve- ment (Work Enthusiasts, Workaholics, and Enthusiastic Workaholics) were joined by the Relaxed Workers; these four groups scored higher than the ‘Unengaged and Disenchanged Workers. The similarity of the Relaxed Workers, to those high in Work Involvement is attributabie to their high scores on Enjoyment. (Recall chat Enjoyment was substantially correlated with Job ment.) The Unengaged and Disenchented Workers, of course, were low on qualities—Work Involvement and Enjoyment—related to Job Invelve- bot ment ‘The next pair of analyses compared the groups on the number participants reported working during an average week and an average weekend. The means for these measures are reported in Table 7. Among men, the Work Enthusiasts reported working the most hours during the week and over the weekend, foliowed by the Workaholics and Enthusiastic Workaholics. Although the means of the latter two groups are higher than those for the Relaxed, Disenchanted, and Unengoged Workers, only the Work Enthusiasts consistently differed significantly from them on both measures. The data from women were less clear-cut. As in men, Work Enthusiasts reported working most during the week, followed by che Workaholics and the Enthusi- astic Workaholics. Pairwise comparisons indicated that the Work Enthusiasts vg the week than did Disenchanted Workaholics was also significantly higher than that of the Disenchanted Workers. For the weekend measure, it was the Enthusiastic Workaholics and che Workaholics who re- ported working the greatest number of hours; the Werk Enthusiasts exhibited a lower mean than expected. The only significant differences, however, were betewen the Enthusiastic Workahotics, on the one hand, and the Work Enthusiasts, Disenchanted Workers, and Unengaged workers, on the other. f hours Comparisons on Health Measures The results of the ANOVAs comparing profiles on the Heal measure are reported in Table 8. ‘As expected, Workaholics of both sexes reported significantly more health complaints then Work Enthusiasts. Among women, Workeholics reported @ significantly greater number of heaith complaints than did all other groups except the Disenchanced Workers. Comparisons of the scores for men reve similer pattern of mean scores, with Disenchanted Workers reporting a signif cancly greater number of complaints than Work Enthusiasts, Relaxed Workers and Unengaged Workers, and Workaholics reporting more aches and pains Complaints la 175 a set wpa 4 a suo $ wk so + ‘re Soyouepogy mses we eoQupom 2 a a siseseuag om 1 punseay puna ‘tg eK POON SIMO} JO ABAMAN, UO UaWon pub UaWy 20) suOsIZeduo’y astmaiey PUB SV AONV 10 Silnsay pue daosy W>e Jo SULT cava, 176 SPENCE AND ROBBINS TABLES Means for Each Group and Results of ANOVAs and Pairwise Comparisons for Men and Women on Heaith Complaints Scale Mer Women 1 Work Eochusias 2. Workahotics a 3 Enchuslastic Workaholics 4 Relaxed Workers 5 Disenchanted Workers 2250, € Unengaged Workers 15.04. 154, RS, 105) = 3.67 HG, HT} = 4.98 p< W p< O1 Nowe” The sabecripar indicate the number of the profile fom which the given group difers significantly, p< 05. than Work Enthusiasts and Relaxed Workers. Disenchanted Workers have in common with Worksholics above average scores on the Driven scaie and below average scores on the Enjoyment scale. DISCUSSION, The outcomes of our analyses demonstrate the utility of assigning men and women who are highly invested in work to subcategories rather than lumping them all together under a single label such as workaholism. When amount of enjoyment and the feeling of being compelied or driven to work are teken into account, cluster analyses revealed that subgroups exhibiting various profiles with respect to these additional qualities actually occur in both sexes. Farther- more, the subgroups differ, as predicted, in a number of other attributes. For example, the high degree of perfectionism and unwillingness co delegate respon- sibiliry to others that observers have attributed to workaholics the. an undifferentiated group of work-involved individuals) are found in those we have labeled as Workaholies but not in those we have iabeled as Work Enthusiasts. Also, the former but not the latter are elevated in che amount of stress chey experience in their jobs and have more health complaints. ‘Although these results are highly suggestive, 2 number of important issues remain unaddressed. High on the list is the question of the relationship between, the various profiles and adequacy of performance. If scores on the Work Involvement Scale are considered in isolation, it seems highly likely chat the correlation will be positive. This contention is based on the results of ongoing studies with students, in which we have demonstrated that scores on the Work Involvement Scale are substantially corzelated in both sexes with several scales developed in previous research, namely Achievement Strivings (Spence et al., WORKAHOLISM 177 1987) and the Mastery and Work Orientation scales from the Work and Family Orientation Questionnaire (Spence & Helmreich, 1983). These scales have been found to be related to measures of real-life performance in a number of different samples (e-g., grades in college students; Spence & Helmreich, 1983; Spence et al., 1987), number of publications, and citations to their work in academic scientists (Heimreich, Spence, Beane, Lucker, & Matthews, 1980, Helmreich, Spence, & Pred, 1988). The question of central concern is whether workaholism, as we have defined the concept, is associated with poorer perfor- mance than work enthusiasm because such characteristics as perfectionism, nondelegation, and job stress that are found in workaholies have deleterious effects. Several stadies are currently being initiated to test this possibility. ‘A second question concerns the implications of the greater number of health complaints reported by those exhibiting the workaholic and the disenchanted profiles. In prior studies (e.g., Costa & McCrae, 1987; Smith, Pope, Rhodewalt, & Poulton, 1989), ic has been established that individuals exhibiting anxiety and other neurotic symptoms report more physical symptoms than others. It is likely chat the workaholice and others elevated on the Job Stress scale of our study tend to be more generally anxious than others. Some investigators (¢.8., Costa & McCrae, 1987; Watson & Pennebaker, 1989} dispute the validity of such reports by anxious persons, both as indicators of current health status, determined by objective means, and as predictors of fueure illness. On the other hand, other investigators (e-g., Fried man & Booth-Kewley, 1987; Taylor, 1980) claim that chronic trait anxiety increases vulnerability to physical illness. The significance of the health complaints data thus remains probiematic, The causal dizection in the data we obtained also remains uncertain; itis possible, at least in some instances, that individuals who chronically suffer minor health prob- Jems may as @ result enjoy their werk less, experience more job stress, and so forth. We are currently initiating studies that may throw some light on the meaning of our correlational findings. ACKNOWLEDGMENT Research for this article was supported in part by 2 grant to Janet T. Spence from the University of Texas Research Institute. REFERENCES Boronson, W. (2976). The workaholic in vou. Money, §, 32-35, ‘Cantarow, E. 0979). Women worksholics. Mother Jones. 6, 56. ‘Cherrington, D: (29803, The work ethic: Working wales and values that work, New Yori: AMACOM, (Costs, B.T..}e.,& MeCrae, RB. (987), Neuroticism, somatic complaines, and diceace: ie the bark 178 SPENCE AND ROBBINS Journal of Personality, $$, 299-336 PP. (£986), Workaholsm, sex, and sex role ssereotyping among female professionals. Sex Roles, 14, 352-560. Farnsworth, 7. 987). Test your execwtive sills. London: Ebury. Friedman, H. S.. & Booch-Kewiey, S. (1987), The “disease-prone personality”: A metwanalytice view of the construct. Amenian Peyhologist, 42, 538-555. Games, P. A., Keseiman, H[.,S Rogen, }C. (198), Simltancous pairwise multiple comperison procedures for means when sample sizes are unequal. Prychological Bulletin, 90, 594.598, Helmreich, R. L., Spence, }.T., Beane, W. E., Lucker, G. W., & Matthews, K, A. (1980). Making 5 im academic paychology: Demographic and personality correlates of attainment. fowai of ogy. 39, 896-908. J Fy & Pred, R. S, <1988). Making x without losing it: Type A, ion, and scientific attainment revisited. Personality and Socal Poychology Personality and Helmreich, RoE, So achievement mot Bullen, 14, 495-508 Jaccard, J. Becker, M. A., & Wood, G. (1984). Pairwise maleipie comparison procedures: A review Prychological Bullets, 9, 585-526. ‘Machlowits, M.(1977, October). Workaholics. Acros the Board, pp. 30-32, 11980), Werkahotics: Living sith ther, working with them. Reading, MA: Addi Meyer, H. (2974. The boss ought to take more time off Fortune. 89, 40-142, 229-250. Minit, F, Meier, P. Wi Brewer, B., & Skipper, S. (2981) The workahatte and hs jamal. ‘Grand Rapids, Ml. Baker Bock House. Mosier,S.K. (29851 Workaholis: An acai ef she stes, success an proities, Unpublished masters thesis, University of Texas at Astin ‘Outes, W. 097!!, Conlesions of @ workaholic: The facts about work addiction. New York: Word Smith, T. W., Pope, M. K.. Rhodewalt, F, & Poulton, } (989). Optimise, nestotiism, coping, ‘and symptom reports: An alternative interpretation of che Life Orientation Tos. Journal of Personality and Socal Pascolog, #8448. Spence, -T..& Helmevich, RL. (1963). Achievement-related moxives and behavior. In}. T-Spence B8), Achicement and achievement motives: Pachologica! and socoogical appraackes ipp. 10-742 Sen Francisco: Freeman, Spence}. T.,Helmreich, R. L., & Pret, RS. 2987), Impatience versus achievement strivings in the Type A pactern: Differencial effects on stodents’ health and academic achievement. foumal of Applied Poycholon, 75, 522-528. Sprankic, ..& Ebel. H. (1987), The workaholic syndrome. New York: Walker. evlor, S.E, (£950), Health pevchology: The science and the ficid, American Prucholagis, 45, 40-50. Ti Martin, } 1987}, Com }.M, Davlev Eds. “The compiea: acaemic” A practical guide forthe begin socal scents. Hiidale, N}. Lawrence Erlbaum Associ Ward, 1H. 15963) fhe American P Spence Department of Psychology University of Texas at Austin Austin, TX 78712 Received January 17, 1991

You might also like