Behavioral expressions of ́Big Five ́-dispositions at the workplace

The research project “multimethod construct measurement” is concerned with on construct validity analysis of personality test use in occupational settings. For this context, it investigates possibilities of a synthesis of construct-based and simulation-based personnel selection methods. The bases of the project are outlined and the results of a first study are presented. In this study, behavioral expressions of personality dispositions at the workplace are investigated applying the concept of Buss and Craik’s (1983) “act frequency approach” and the Five Factor Model of Personality as supposed by Costa and McCrae (1992), including domains and facets. 58 employees of service industries generated 2040 descriptions of vocational behavior (=acts). Two raters preassessed these descriptions referring to formulation errors, definiteness of act generation and general quality of act generation. Subsequently, ten assessors assigned the 1377 reworded acts to the most plausible domains and facets using a multiple disposition act sorting system. In addition, they judged the job relatedness and the occupational desirability for every act. Results show a uniform pattern of formulation error independent of domains and facets. Allocation errors are specific for dimensions, facets and poles. Contrary to expectations, no homogenous domain pattern was found. “Conscientiousness” is the only domain with uniform job relevance. All other dimensions show heterogeneous relevance depending on specific facets and the criteria used. An overwhelming take-over effect was found for the assignment of acts to domains and facets unaware of their original source domains and facets. “Achievement striving” as a facet of “conscientiousness” profited most from this effect. Possible causes are discussed. ‘Big 5’-Expressions at the Workplace 3 Behavioral expressions of ‘Big Five’-dispositions at the workplace Relevance of personality descriptors for personnel selection purposes In the past decade, several meta-analyses (e.g. Barrick & Mount, 1991; Tett, Jackson & Rothstein, 1991; Hough, Eaton, Dunnette, Kamp & McCloy, 1990; Salgado, 1997) have examined the criterion-related validity of personality variables in personnel selection. These analyses confirm older reviews (e.g. Guion & Gottier, 1965) in their conclusion that personality inventories rarely exceed a validity of r=.30 for overall job performance. However, the results are interpreted more optimistically, because they are based on a more rigorous preselection of primary studies and elaborated meta-analytic techniques: “... [The] controversy over whether personality tests can be useful for prediction of employee performance is no longer pertinent; the more pressing need at this time is to clarify when, for what, and how they are useful.” (Kanfer, Ackerman, Murtha & Goff, 1995, p.597) Although all meta-analyses used the ‘big five’-framework or an extended version of it (Hough et al.), their specific results differ in validity estimates and even rankings of the different Big5dimensions for job performance. In addition, several moderator variables are discussed, first of all the kind of performance criteria used and the kind of occupation analyzed. Objective of the entire research project “multimethod construct measurement” In contrast to the “meta-research” concentrating on criterion-related validity our research project “multimethod construct measurement” is focusing on the “micro-level” of individual items, starting with some basic questions of personality theory. The two major project targets could be summarized as follows: • Construct validity analysis of personality test use in occupational settings In our opinion, substantial amplification of knowledge concerning the prognostic value of personality variables for job success could only be reached by further investigations on the boundary conditions of construct expressions. Our primary operating point for this is the analysis of well-defined fundamental research personality constructs with respect to their principal occupational pertinence. One of the most widespread personality theories, the Five Factor Model of Personality in the domain and facet version of Costa and McCrae (1992), serves as reference for this work. Job activities are analyzed based on this “personality viewpoint” for potential indicators of Big5-expressions. Only after a systematic investigation of possible personality-based behavior styles in work environments, it seems reasonable to ask which of these expressions are related to job performance. • Synthesis of construct-based and simulation-based personnel selection methods Refering to Wernimont and Campbell’s (1968) distinction between interpreting behavior as “sign” or as “sample”, personnel selection methods can be divided into two conceptual groups: “Sign” methods (like general mental ability tests or traditional personality tests) interpret executed behavioral acts as expressions of stable dispositions. “Sample” methods (e.g. the classical assessment center tasks like role play, group discussion, in-basket-test) concentrate on the representativeness of the selection tasks for the target job activities. The registered behavioral expressions in these simulations serve as predictive sample for future activities on the job. Both groups show complementary validity profiles: Construct-based methods rely heavily on construct validity results and employ well-defined dispositions. But their criterion-related 1 The terms ”dimension“ and “domain“ are used synonymously in this paper. ‘Big 5’-Expressions at the Workplace 4 validity is sometimes ambiguous (personality tests) and their content validity is often not evident. Simulation-based methods reflect for the most part an obvious content validity and show moderate criterion-related validity. The most serious drawback is the unclear conceptual status of the measured constructs in these exercises (cf. Neidig & Neidig, 1984; Sackett & Dreher, 1982, 1984). An attempt to integrate these different assessment approaches should ideally result in an instrument that would benefit from the advantages of both conceptions – or inevitable incongruences may be uncovered. Significance of the present study for the entire research project The work presented herein belongs to a series of four studies. Their relationships are shown in figure 1. study 1 : (presented) compilation of Big5-expressions at the workplace study 2 : (in progress) identification of Big5-requirements for work purposes study 3 : (in progress) variation of Big5-measurement within traditional paper&pencil item formats study 4 : (in progress) variation of Big5-measurement within simulation-based techniques base for construction base for construction extendend to informs about possible moderators