Consensual Definition of the Big Five 1 Toward a Consensual Definition of the Big Five

The Abridged Big Five-Dimensional Circumple~ (ABSC; Hofstee, de Raad, & Goldberg, 1992) was used to locate consensus across different conceptualizations of the "Big Five" or Five-Factor ModeL Trait rating data from instruments representing four different versions of the Big Five (Goldberg, 1992; Hogan & Johnson, 1981; McCrae & Costa, 1985; Norman, 1963) were subjected to separate ABSC analyses for 2148 American and 128S German subjects. Replicated results formed standard designation codes representing trait adjectives' primary and secondary factor loadings. These codes unveiled both the core meaning of each factor common to all instruments and also the unique coloring imparted by secondary loadings to different scales proffered by researchers to represent the five factors. 3 Consensual Definition of the Big Five Toward a Consensual Definition of the Big Five Recent reviews (Digman, 1990; John, 1990; McCrae & John, 1992) have indicated that the universe of personality traits can be represented at an abstract level by five broad dimensions. Although these reviews clearly intend to stress commonality, consensus, and agreement of thinking about the "Big Five" or Five-Factor Model (FFM), a prominent section of each review discusses disagreements concerning the precise nature of, and best labels for, the five factors. This paper suggests that consensus can be reached by describing personality terms and scales with Hofstee, de Raad, & Goldberg's (1992) Abridged Big Five-Dimensional Circumplex (AB5C). Hofstee, et al. hypothesize that differences in secondary factor loadings may have contributed to the equivocal interpretation of the Big Five factors across research programs. The present study tests this hypothesis explicitly by mapping both primary and secondary loadings of items across diverse data sets according to Hofstee, et al.'s AB5C model. Method The current study reanalyzes 15 data sets from previous studies (total N = 3422; see Table 1). These data sets contain trait ratings data from instruments representing four different versions of the Big Five (Goldberg, 1992; Hogan & Johnson, 1981; McCrae & Costa, 1985; Norman, 1963). These ratings were factor analyzed to yield a five-factor, varimax-rotated principle components matrix. Insert Table 1 about here A FORTRAN program was written to postmultiply the varimax-rotated matrices from each data set by the submatrices described by Hofstee et al. (1992). The program locates each term's highest loading among the 90 factor poles within 10 circumplexes generated by the five factors. Terms are 4 Consensual Definition of the Big Five depicted according to their highest primary/secondary loadings. For example, in most data sets, the highest loading for analytical was found in the circumplex defined by Factor III (x-axis) and Factor V (y-axis) on the factor inserted at 60°. This term therefore received the ABSC code V+III+ (primarily Factor V, secondarily Factor III). A composite designation for each term was computed by averaging designations across data sets. The composite ABSC depictions, based on multiple samples responding to inventories based on different conceptions of the FFM, provide a description of personality terms that is more reliable and theory-neutral than depictions derived from a single inventory administered to a single sample. After establishing ABSC depictions of all terms, the ABSC character of the authors' scales was determined following the method of Hofstee, et al. (1992, p. 161), whereby two points are assigned for the primary and one point for the secondary part of the designation of each term on the scale. The scale designation in AB5C space is simply the unweighted sum of the deSignations of each term on that scale. An examination of the overall AB5C character of different scales proffered by researchers to represent the five factors shows, through the secondary aspect of each scale, exactly how different researchers conceptualize and operationalize the factors differently. On the other hand, the consensually derived "factor-pure" terms (I+I+, II+II+, etc.) indicate the "core meaning" of each factor, Le., interpretations of the factors upon which the various research groups agree. Results Table 2 summarizes the final AB5C depictions for all 140 personality terms. The 24 factor-pure terms represent the consensus across research groups. 5 Consensual Definition of the Big Five Insert Table 2 about here Table 3 indicates the AB5C structure of the scales proffered by the four research groups for assessing the FFM. Only three scales contain mostly factor-pure terms; the remaining scales are heterogeneous factor blends. The AB5C differences in nonpure scales point to conceptual differences underlying the various measures. Insert Table 3 about here Discussion The present analyses demonstrate how different interpretations of the FFM are reflected in scales designated by divergent ABSC codes. McCrae & Costa's view that positive emotions and warmth belong to Factor I causes their Factor I and II scales to take on a 1+11+ and 11+1+ character, respectively. Other researchers' Factor I scales are 1+1+, while their Factor II scales are II+IV+. McCrae and Costa's Factor III scale received a III+V+ designation, reflecting their interpretation of Factor III as organized purposefulness or intellectual achievement. In contrast, the other researchers' scales were all 111+11+, reflecting a view of Factor III as interpersonal maturity and impulse control. Finally, McCrae and Costa's Factor V scale, which they say measures openness to experience, received a V+I+ designation. The other researchers, who favor an intellect interpretation of Factor V, employ V+III+ scales. An important question is whether alternative interpretations of the factors are equally valid, or whether certain interpretations are, in some way, more "correct" or better. Some would regard this question as intractable, because interpretation of factors depends upon rotation, and rotational decisions are arbitrary (see McCrae & Costa, 1989, p. 591). The present findings are not offered as the correct interpretation of factors, but 6 Consensual Definition of the Big Five the "factor-pure" terms (1+1+, 11+11+, etc.) in the present study do represent a standard upon which a number of research programs converge. Based on the available data, one might draw the following tentative conclusions about the core meanings of each factor. The pure 1+1+ items extraverted, frank/open, funloving, sociable, talkative, and straightforward suggest that the core of Factor I is social expressiveness QX communicativeness. The 11+11+ cell contains acquiescent, mild/gentle, and softhearted. Because past research has indicated that Factor II is highly desirable, perhaps these terms suggest the type of softness valued by Taoists (Waley, 1958): Virtue through yielding, pliancy, suppleness, and receptivity. Factor III is defined by the following pure items: careful, fussy/tidy, hardworking, neat, punctual, scrupulous, thrifty, and well-organized. Of all these terms, only one--hardworking--points to an achievement interpretation of Factor III. The other terms describe an abstemious, exacting, orderly, prudent, restrained, temperate person, a person who avoids excesses and pays close attention to detail. perhaps Tellegen's (1982) label Constraint fits the 111+111+ terms most aptly. Factor IV, a relatively noncontroversial factor, is defined by one pure IV+IV+ item, calm. This suggests that the core of IV is freedom from negative affect (pace Tellegen, 1982). For Factor V, we find five V+V+ items, artistic, creative, imaginative, learned, and progressive. This suggests that creativity is not merely related to Factor V (cf. McCrae, 1987) but rather lies at the core of the meaning of Factor V. Goldberg (1992) has stressed the need for a set of relatively pure markers for the Big 5. The present research shows that scales proposed by different researchers to assess the FFM often tend to be blends rather than pure markers of these factors. Through the methodology of the AB5C, however, one can locate a commonality in the way Goldberg, Hogan, McCrae and Costa, and 7 Consensual Definition of the Bi9 Five Norman conceptualize and measure the Bi9 S. This provides us with a consensual definition of the core of each factor: I, social communicativeness; II, softness; III, constraint; IV, freedom from negative affect, and V, creativity. Future research can assess whether the inclusion of measures offered by other researchers would substantially change these findin9s • Consensual Definition of the Big Five