DIFFERENT STRATEGIES FOR THE CONSTRUCTION OF A JOB-RELATED, BEHAVIOR- ORIENTED BIG5-PERSONALITY QUESTIONNAIRE: DISTINCT VS. OVERLAPPING BEHAVIORAL ACT SELECTION
 
Hoeft, S., German Aerospace Center (DLR) and Hell, B., University of Hohenheim, Germany
 
The presentation describes the construction of a job-related personality questionnaire based on the NEO-Version of the Five-Factor-Model by Costa and McCrae (1992), including domain and facet level. The conceptual framework of the item generation is formed by the Act Frequency Approach by Buss and Craik (1983): 58 employees of service industries generated 2040 descriptions of job-related behavior (= acts). Subsequent to this, 1377 reworded acts were rated by ten psychologists according to their prototypicality for the different Big5-constructs. 363 highly prototypical acts are used for an empirical convergence study with the NEO-PI-R as a reference instrument. Calculations of this phase are based on data of 278 subjects, mostly bank clerks or students of business administration and economics. Two different item selection procedures for the formation of the final instrument version are discussed: The first approach strives for conceptual distinct construct representation by choosing acts solely prototypical for the target construct. The second approach aims at the best representation of the constructs by choosing acts with the highest prototypicality for the target disposition disregarding information about conceptual overlap with concurrent constructs. Utilizing a Multitrait- multimethod-framework. The results of the empirical study reveal a slightly better convergence for the second item selection strategy, but at the same time worse divergence occurs. A theoretical model for the integration of conceptual and empirical interlinks of Big5-constructs is presented, and the different applications for the two item selection approaches are discussed.