E. Demerouti et al.: Burn out Instruments EJPA 19 (1), �� 2003 Hogrefe & Huber Publishers The Convergent Validity of Two Burnout Instruments A Multitrait-Multimethod Analysis Evangelia Demerouti1, Arnold B. Bakker2, Ioanna Vardakou3, and Aristotelis Kantas4 1Carl von Ossietzky University of Oldenburg, Germany, University of Nijmegen, The Netherlands, and Utrecht University, The Netherlands, 2Utrecht University, The Netherlands, HCMG Human Capital Management Group, Utrecht, 3London School of Economics, United Kingdom, 4University of Patras, Greece Keywords: Burnout, instruments, confirmatory factor analysis, multitrait-multimethod analysis European Journal of Psychological Assessment, Vol. 19, Issue 1, pp. 12���23 Summary: This study examined the factorial and convergent validity of the OLdenburg Burnout Inventory (OLBI) and the Maslach Burnout Inventory ��� General Survey (MBI-GS) among 232 Greek employees from different occupational groups (e. g., banking and insurance, chemical industry). Results of confirmatory factor analyses supported the proposed factor structure for both instruments. In addition, the convergent and discrim- inant validity of the OLBI and MBI-GS was supported by the results of multitrait-multimethod analyses. Since the OLBI subscales contain both positively and negatively phrased items, it is suggested to increase the psycho- metric properties of the MBI-GS by including the positively phrased items of the OLBI. Over the past two decades, the assessment of burnout has attracted the interest of many researchers, particularly those who study the syndrome in the human services sector. Researchers who want to assess burnout outside this occupational domain are faced with a problem since the most popular instrument for the measurement of burnout, the Maslach Burnout Inventory (MBI Maslach & Jackson, 1981a, 1981b, 1986 Maslach, Jackson, & Leiter, 1996), was developed exclusively for use in hu- man services professions. Accordingly, the three sub- scales of the MBI ��� Emotional exhaustion, Depersonal- ization, and Personal accomplishment ��� are applicable only to employees who work with people. Hence, when studying burnout outside the human services, either the MBI has to be adapted or an entirely new instrument has to be developed. The central goal of the present study is to introduce and test a new instrument for the measurement of burn- out ��� the OLdenburg Burnout Inventory (OLBI Demer- outi, 1999 Demerouti & Nachreiner, 1998). We examine the convergent and discriminant validity of the OLBI vis-��-vis the Maslach Burnout Inventory ��� General Sur- vey (MBI-GS Schaufeli, Leiter, Maslach & Jackson, 1996). The MBI-GS is an adaptation of the MBI-Human Services Survey, in that it assesses burnout in a generic sense. The OLBI is an entirely new instrument that can be used in virtually every occupational context. This in- strument aims to overcome one major psychometric shortcoming of the MBI and the MBI-GS, namely, the one-sided wording of the items. Whereas the MBI-GS was initially developed for English-speaking populations, the OLBI was developed in Germany. In the current study, we administer both instruments to a Greek sample after translating the orig- inal items of both burnout instruments. Thus, our second goal was to investigate the psychometric quality of both instruments in Greece, a country where employees��� burnout has not received much research attention in the past. It is important to note that the factorial validity of the MBI is not completely beyond question as some stud- ies have reported fewer or more factors than the three proposed (Schaufeli, Enzmann, & Girault, 1993). In ad- dition, validation of the factor structure of the MBI-GS and the OLBI in other cultures is still needed. DOI: 10.1027//1015-5759.19.1.12 EJPA 19 (1), �� 2003 Hogrefe & Huber Publishers

Instruments for the Measurement of Burnout Several adaptations of the MBI have been proposed in order to enable its application outside the human servic- es. Among other changes, Golembiewski and colleagues (e. g., Golembiewski, Boudreau, Munzenrider, & Luo, 1996 Golembiewski, Munzenrider, & Carter, 1983) sub- stituted the term ���co-worker��� for ���recipient.��� According- ly, ���depersonalization��� and ���personal accomplishment��� were assessed in terms of how participants felt toward work colleagues instead of recipients. This substitution wrongly assumes that the two types of relationships ��� service provider and recipient resp. worker and co-work- er ��� are conceptually equivalent, which is unlikely to be the case (Garden, 1987). Interpersonal relations with re- cipients are essential in the human services because they are the main vehicles for achieving professional goals, whereas relations with co-workers have a different meaning and are indeed qualitatively different. Other re- searchers have substituted ���recipients��� in MBI-items with ���subordinates��� or ���teachers��� when studying burn- out among managers (Lee & Ashforth, 1990, 1993) or students (Fimian, Fastenau, Tashner, & Cross, 1989 Powers & Gose, 1986), respectively. Not surprisingly, the internal consistency, particularly of the rephrased de- personalization subscale, is rather modest (i. e., Cron- bach���s �� is about .60). Moreover, Evans and Fisher (1993) tested the equivalence of the original MBI and an adapted co-worker version and found, as predicted by Garden (1987), that depersonalization did not form a coherent or meaningful factor in a sample of information technology employees, whereas the three-factor solution of the MBI was replicated among public-school teachers. This result casts doubts on the elaborate work of Golem- biewski and colleagues, which hinges on the generality of the depersonalization and personal accomplishment dimension across occupational fields. Thus, adapting the MBI for use outside the human services by simply sub- stituting other terms for ���recipients��� does not seem to be an encouraging solution. Recently, Schaufeli et al. (1996) developed the Mas- lach Burnout Inventory ��� General Survey (MBI-GS). Based on the notion that emotional exhaustion, deper- sonalization, and personal accomplishment can be broadened beyond the interpersonal domain characteris- tic to the human services, they included three more ge- neric burnout dimensions, labeled exhaustion, cynicism, and professional efficacy. Clearly, these MBI-GS sub- scales parallel those of the original MBI. However, con- trary to the earlier slight adaptations in the wording of items, the MBI-GS includes different items that refer to more general, nonsocial aspects of the job. Studies in Canada (Leiter & Schaufeli, 1996) and in The Nether- lands (Bakker, Demerouti, & Schaufeli, in press) sup- ported the invariance of the MBI-GS���s factor structure across various occupational groups, including mainte- nance workers, nurses, software engineers, and manag- ers. Also, its factorial structure seems to be invariant across workers from the same company who are em- ployed in three different countries ��� The Netherlands, Sweden, and Finland (Schutte, Toppinen, Kalimo, & Schaufeli, 2000). Moreover, initial studies with the MBI- GS suggest that the phenomenon and process of burnout takes the same form in occupations within and outside human service professions (Leiter & Harvie, 1998 Leiter, Harvie, & Frizzell, 1998 Leiter & Schaufeli, 1996). However, the MBI-GS did not overcome one impor- tant psychometric shortcoming of the original version of the MBI, namely, that the items in each subscale are all framed in the same direction. Accordingly, all exhaustion and cynicism items are phrased negatively, whereas all professional efficacy items are phrased positively. From a psychometric point of view, such one-sided scales are inferior to scales that include both positively and nega- tively worded items (cf. Anastasi, 1988). It can, for in- stance, lead to artificial factor solutions in which posi- tively and negatively worded items are likely to cluster (cf. Doty & Glick, 1998). Indeed, some authors (Demer- outi & Nachreiner, 1996 Lee & Ashforth, 1990) have criticized the MBI on this point. For instance, Lee and Ashforth (1990) argued that the item wording can be seen as a problem since it offers an alternative explana- tion for the strong associations of exhaustion and deper- sonalization with psychological strain (generally nega- tively worded) and of personal accomplishment with perceptions of performance (generally positively word- ed). Besides the MBI-GS, a handful of researchers have developed an entirely new instrument to measure burn- out in other than human service occupations. Although this approach seems most promising at first glance, re- sults thus far have been rather disappointing. For in- stance, Sonnentag, Brodbeck, Heinbokel, and Stolte (1994) investigated burnout among software developers and came up with two burnout scales, which they labeled ���Lack of identification��� and ���Perceived pressure.��� Al- though they claim that these scales correspond with Per- sonal accomplishment and Emotional exhaustion in the MBI, inspection of the items reveals that this is not the case. For instance, Lack of identification is contaminated with items that are usually found in organizational com- mitment inventories (e. g., ���I like to stay at my work- place���). In a similar vein, Perceived pressure is con- founded with cognitive representations of the working conditions and does not refer to feelings of exhaustion. E. Demerouti et al.: Burnout Instruments 13 EJPA 19 (1), �� 2003 Hogrefe & Huber Publishers