The five-factor model (FFM) is unquestionably the best-known model of human personality (Funder, 2001). Its popularity is evident in its recent application to a wide variety of areas, such as, but not limited to, the adoption of technology (Devaraj, Easley, & Crant, 2008), aggression toward an intimate partner (Hines & Saudino, 2008), college students’ health-related behaviors (Raynor & Levine, 2009), motivation to learn and develop (Major, Turner, & Fletcher, 2006), organizational commitment (Erdheim, Wang, & Zickar, 2006), organizational justice (Shi, Lin, Wang, & Wang, 2009), relationship infidelity (Orzeck & Lung, 2005), and retirement decisions (Robinson, Demetre, & Corney, 2010). Despite the popularity and versatility of the model, the nature and appropriateness of the FFM has been the source of substantial debate (see e.g., DeYoung, 2010; Srivastava, 2010). Moreover, researchers’ efforts to explicitly clarify the number of factors that best describe human personality have generated differing results (see e.g., Almagor, Tellegen, & Waller, 1995; Ashton, Jackson, Helmes, & Paunonen, 1998; Bowler, Bowler, & Phillips, 2009; Jackson, Paunonen, Fraboni, & Goffin, 1996; Simms, 2007; Zuckerman, Kuhlman, Joireman, Teta, & Kraft, 1993).

The impact social perceptions have on the FFM is one of the most relevant issues in this debate. As noted by Srivastava (2010), it is impossible to separate measures of personality from the errors of human perception. Either the measures themselves require human perception when they are being completed (e.g., self-report measures) or human perception is introduced at the measure’s creation (e.g., Funder & Sneed, 1993; Yamagata et al., 2006). Regarding the need for human perception, Saucier and Goldberg (1996) noted that the FFM represents “dimensions of perceived personality” (p. 124). Similarly, Fiske (1994) noted that the FFM is useful for understanding “how people perceive people” (p. 124) and is based on “interpretations or small generalizations from perceived behavior” (p. 123). Along these lines, recent evidence suggests that cognitive complexity (the dimensionality of an individual’s social perceptions) has an impact on the structure of five-factor measures (Bowler et al., 2009). In this study we sought to test this assumption further by addressing the following question: Does the personality structure of individuals with higher levels of cognitive complexity differ from the personality structure of individuals with lower levels of cognitive complexity?

The Five-Factor Model

The history of the development of the FFM has been thoroughly described by numerous researchers (e.g., Wiggins, 1996). What began as a two-factor model of intellect and will (Webb, 1915) quickly expanded into three- and four-factor models (Cattell, 1933; Garnett, 1919). From these, Fiske (1949) developed the basic foundation for what would become the FFM as it is now constituted. During the following decades, the majority of researchers supported the overall stability of the five-factor structure (e.g., Aluja, García, García, & Seisdedos, 2005; Costa & McCrae, 1992; Goldberg, 1992). Although there are currently numerous five-factor measures (e.g., Costa & McCrae, 1992), all of them share the same fundamental factors: agreeableness, conscientiousness, emotional stability, intellect, and surgency (extroversion). Additionally, in the most common measures some kind of variation of typical self-report items is used. For example, Goldberg’s (1992) unipolar marker measure constitutes a series of single-word traits (such as generous and relaxed) and respondents are instructed to indicate the degree to which they believe the trait accurately describes them on a 9-point scale, ranging from 1 = extremely inaccurate to 9 = extremely accurate. Similarly, Costa and McCrae’s revised neuroticism, extroversion, openness personality inventory (NEO PI-R) constitutes a series of short statements (e.g., “I am easily frustrated”) and respondents indicate the degree to which they agree with each statement on a 5-point scale, ranging from 1 = strongly disagree to 5 = strongly agree.

Criticisms of the Five-Factor Model

In addition to the problematic issues of acquiescence and response distortion (Barrick & Mount, 1996; Holden, 2008) that are commonly associated with the self-report item format, there are several specific criticisms of the FFM. These have ranged from Mischel’s (1968) argument that the five factors are superficial stereotypes that have little or no relationship with actual behavior, to Block’s (1995) argument that the dimension definitions lack precision and do not provide insight into human personality. Other criticisms of the FFM relate to the lack of content validity (Fiske, 1994), lack of construct-related validity (Waller & Ben-Porath, 1987), and of criterion-related validity (Ashton et al., 1998). Some of the most substantial criticisms of the FFM focus on the application of factor analysis to the development of its factor structure (e.g., Block, 1995). Specifically, although this method has existed for over a century, there remains no clear method for determining the appropriate number of factors to extract (i.e., the subjectivity of scree plot interpretation). Irrespective of these criticisms, a substantial number of researchers support the general FFM. With some exceptions noted below, factor analysis regularly yields a five-factor structure that parallels the typical dimensions of the FFM for self, peer, supervisor, and teacher ratings (Goldberg, 1992; Miller, Pilkonis, & Morse, 2004; Ployhart, Lim, & Chan, 2001).

Despite this consistency, some researchers suggest that the factor structure may range from three to seven factors (see e.g., Almagor et al., 1995; Jackson et al., 1996; Simms, 2007; Zuckerman et al., 1993). In addition to the irregularity of these results, we are unaware of any researchers who have provided an explanation for the differing number of factors identified and used in measures and, more specifically, the individual differences that may engender these different factor structures. However, in a recent study Bowler et al. (2009) suggested that an individual’s cognitive complexity may have an impact on the factor structure of an individual’s personality.

Cognitive Complexity

Cognitive complexity is a measure of an individual’s ability to incorporate and differentiate multiple environmental elements (Kelly, 1955; Labouvie-Vief & Diehl, 2000; Vannoy, 1965). Specifically, (a) individuals can perceive and organize a finite number of social behaviors, and (b) this finite number is variable across individuals. Those who demonstrate high levels of cognitive complexity are able to distinguish numerous social elements and proceed to investigate the connections among those elements. In contrast, individuals who demonstrate low levels of cognitive complexity distinguish fewer social elements. There is a substantial body of research to support the assertion that cognitive complexity reflects the dimensionality of an individual’s thoughts (see e.g., Bieri et al., 1966; Feixas, Moliner, Montes, Mari, & Neimeyer, 1992). Thus, as cognitive complexity constrains the dimensionality of an individual’s thought, it may also restrict the dimensionality of his or her personality.

Cognitive complexity and the five-factor model. Although studies have been conducted in which the relationship between cognitive ability and the factor structure of personality measures has been examined (e.g., Austin et al., 2002; Toomela, 2003), we are aware of only one other study in which the impact of cognitive complexity on the FFM has been directly assessed (i.e., Bowler et al., 2009). Thus, using a novel data set, we sought to replicate the findings of Bowler et al. by directly evaluating the impact that differing levels of cognitive complexity have on the factor structure of a conventional five-factor measure – in this instance Saucier’s (1994) Mini-Markers. Specifically, our aims were to determine whether individuals with below average cognitive complexity demonstrated simpler personalities (i.e., fewer than five factors) than did individuals with average cognitive complexity and whether individuals with above average cognitive complexity demonstrated more complex personalities (i.e., more than five factors) than did individuals with average cognitive complexity.

Method

Participants

Participants were 753 undergraduate students (64% freshmen, 26% sophomores, 7% juniors, and 3% seniors) enrolled in a research participation pool at a large southeastern university in USA. The mean age of the participants was 18.82 years (SD = 2.59, range 17 to 53 years) and 71% were female. Among the group of participants 77% identified themselves as Caucasian, 16% as African American, 2% as Asian American, and 3% as Hispanic, with the remaining 2% preferring not to provide this information.

Measures

Five-Factor Model. We used Saucier’s (1994) Mini-Markers to measure the FFM. This measure is composed of 40 adjectives, with eight adjectives measuring each of the factors of extroversion, agreeableness, conscientiousness, emotional stability, and intellect or openness. The extroversion factor includes adjectives such as bashful, energetic, and talkative; the agreeableness factor includes adjectives such as cooperative, kind, and warm; the conscientiousness factor includes adjectives such as careless, efficient, and sloppy; the emotional stability factor includes adjectives such as envious, fretful, and jealous; and the intellect factor includes adjectives such as complex, deep, and philosophical. For each of the 40 adjectives, respondents are instructed to describe themselves as accurately as possible using a 9-point Likert-type scale, ranging from 1 = extremely inaccurate to 9 = extremely accurate.

Cognitive complexity. As in the study by Bowler and colleagues (2009), cognitive complexity was measured via a computer-based version of the Construct Repertory Test (Rep Test). Originally developed by Bieri et al. (1966), the Rep Test begins with respondents identifying 10 individuals in their lives who correspond to 10 predefined roles that are specified in the test as “yourself, a person you dislike, your mother, a person you would like to help, your father, a friend of the same sex, a friend of the opposite sex, the person with whom you feel the most uncomfortable, a person in a position of authority, a person who is difficult to understand”. Using a 6-point Likert-type scale, participants rate each of the individuals on 10 bipolar adjective pairs: outgoing to shy, maladjusted to adjusted, decisive to indecisive, excitable to calm, interested in others to self-absorbed, ill humored to cheerful, irresponsible to responsible, considerate to inconsiderate, dependent to independent, and interesting to dull. Based on the scoring procedure developed by Johnson (1994), overall cognitive complexity is calculated by summing the number of matching ratings that are assigned for each role (two points each) as well as the number of ratings that are within one point of each other (one point each). This results in 450 total comparisons that generate scores ranging from 230 (above average cognitive complexity) to 900 (below average cognitive complexity). In order to streamline data collection and scoring, the Computer-administered Rep Test (CART) was used to measure cognitive complexity in this study. When developing this computer-based measure, Woehr, Miller, and Lane (1998) noted both its measurement equivalence with the paper-and-pencil Rep Test as well its strong test-retest reliability (r_t1t2 = .75).

Results

Cognitive Complexity and Subgrouping

The mean cognitive complexity score was 314.58 (SD = 44.11) and ranged from 237 (high level of cognitive complexity) to 582 (low level of cognitive complexity). Overall, cognitive complexity was unrelated to participants’ sex, F(2, 485.79) = 1.85, p = .16, ethnicity, F(4, 744) = 1.39, p = .23, or academic year, F(3, 176.64) = .77, p = .50. Furthermore, cognitive complexity was unrelated to participants’ age (r = -.047, ns).

To remain consistent with previous research in which the impact of cognitive complexity on the FFM has been examined, the 25th percentile cut-off points employed by Bowler et al. (2009) were used to create the three subgroups used for comparison in this sample. Thus, the sample was split at 281.5 and 329.5 in order to classify participants into three groups of low level of cognitive complexity, average level of cognitive complexity, or high level of cognitive complexity. These quartiles were originally chosen in an attempt to maintain equal sample sizes for the low cognitive complexity and high cognitive complexity groups. As in the study by Bowler and colleagues, when these cut-off points were applied to this sample, three subgroups were established with significantly different levels of cognitive complexity, F(2, 325.09) = 902.21, p < .001.

Factor Analyses

In order to maintain consistency with the development of this particular FFM measure, we employed the same basic analytical techniques as Saucier (1994). Specifically, we employed a principal components factor analysis with varimax rotation. Additionally, all analyses were based on correlation matrices (i.e., ipsatized data) in order to control for response bias (Rammstedt, Goldberg, & Borg, 2010). The first set of analyses was conducted on the entire dataset (N = 753) in order to ascertain whether this sample produced the expected factor structure (i.e., an FFM) regardless of cognitive complexity. Overall, acceptable values were obtained for both the Kaiser-Meyer-Olkin (KMO; Kaiser & Rice, 1974) measure of sampling adequacy (.80) and Bartlett’s test of sphericity (Bartlett, 1950), χ²(780, N = 753) = 10070.39, p < .001. Unlike in previous research (e.g., Bowler et al., 2009), interpretation of the scree plot from this initial analysis was somewhat ambiguous. Following the conventional standard promoted by Cattell (1966), this plot suggested either five or six primary factors (i.e., there was a clear break in the decrement of eigenvalues between components five and six and between components six and seven). However, using Nelson’s (2005) R² test (the sequential regression of the eigenvalues onto their relative factors and evaluating the resultant change in R² as eigenvalues and factors are removed) yielded a value of .81 for the FFM. When the fifth factor is removed this value exceeds Nelson’s suggested threshold of .80. Thus, it can be said that a relatively straight line fits the remaining eigenvalues and components, and that no additional substantial factors remain.

An additional factor extraction procedure was used in order to determine how many factors to extract. Specifically, Velicer’s Minimum Average Partial (MAP; Velicer, 1976; Velicer, Eaton, & Fava, 2000) test was conducted. This method has been shown to be quite an accurate way of correctly identifying the number of factors to extract (Zwick & Velicer, 1982). Unlike many of the more common methods for extracting factors (e.g., scree plot), the MAP test provides a specific value based on the average squared partial correlation reaching a minimum value when components are extracted. Using this method (see O’Connor, 2000), four additional analyses were conducted, one on each of the noted correction matrices. When we applied the MAP test to the entire sample we concluded that there were six factors. This is one more factor than suggested by scree plot interpretation and the R² test. However, in order to remain consistent with the methods by which the five-factor nature of this measure was first constructed, we continued our analyses based on a five-factor solution.

The varimax-rotated factor loadings from this FFM were then examined. All of the items displayed their highest loading on the factor for which they were expected to be markers. Moreover, of these 40 items, all possessed factor loadings greater than |.30| (the traditional standard used by Goldberg, 1992) and 92.5% possessed factor loadings that were at least double their loading on any other factor (the conservative standard used by Saucier, 1994). Overall, the FFM appears to be well represented by this data. Thus, disregarding cognitive complexity, responses on Saucier’s Mini-Markers appear to support the traditional structure of the FFM that was suggested during its development.

Average cognitive complexity subgroup. The next analysis was a principal components analysis of the subgroup of participants who demonstrated an average level of cognitive complexity (n = 342). As with the entire sample, acceptable values were obtained for both the KMO measure of sampling adequacy (.77) and Bartlett’s test of sphericity, χ²(750, n = 342) = 5071.09, p < .001. However, unlike the scree plot for the entire sample, the scree plot for this subsample was much easier to interpret. Following the Cattell (1966) standard, this plot clearly suggests the existence of five primary factors with a clear break in the decrement of eigenvalues between components five and six. Additionally, Nelson’s (2005) threshold of .80 was not achieved by the four-factor solution (R² = .75) but was exceeded by the five-factor solution (R² =.89). Similarly, when the MAP test was applied to the correlation matrix of this subgroup, a five-factor solution was suggested. Thus, unlike the MAP analysis of the entire sample, the analysis of the average cognitive complexity subgroup suggested the traditional five factors.

From this five-factor solution, the varimax-rotated factor loadings for this subgroup were then examined, with all of the items displaying their highest loading on the factor for which they were expected to be markers. Moreover, all of these items had factor loadings greater than |.30| and 85% of these items possessed factor loadings that were at least double their loading on any other factor. Thus, the FFM was supported for the average cognitive complexity subgroup and was not distorted by the removal of those with higher and lower levels of cognitive complexity.

Below average cognitive complexity subgroup. We conducted a principal components analysis on the below average cognitive complexity subgroup (n = 231). Acceptable values were obtained for both the KMO measure of sampling adequacy (.75) and Bartlett’s test, χ²(780, n = 231) = 3356.85, p < .001. Unlike the scree plot for the average cognitive complexity subgroup, the scree plot for this subgroup was much more ambiguous and open to interpretation. However, the Nelson (2005) threshold was exceeded after factor three (R² =.90), suggesting that a three-factor solution was the most appropriate. Moreover, when the MAP test was conducted on the correlation matrix for this subgroup, a three-factor solution was suggested.

When examining the factor loadings of the five varimax-rotated factor loadings for the low cognitive complexity subgroup, only 90% of the variables displayed their highest loading on the factor for which they were expected to be markers (i.e., 10% had incorrect primary loadings). Additionally, of these correct loading items, only 77% possessed factor loadings that were at least double their loading on any other factor and 81% had a second factor loading that was greater than |.30|. Thus, overall, individuals with below average levels of cognitive complexity did not display the clear five-factor structure that characterized those with average levels of cognitive complexity.

Above average cognitive complexity subgroup. Finally, a principal components analysis was conducted on the high cognitive complexity subgroup (n = 177). Acceptable values were obtained for both the KMO measure of sampling adequacy (.68) and Bartlett’s test, χ²(780, n = 177) = 2893.15, p < .001. Unlike the average and low cognitive complexity subgroups, when applying the Cattell (1966) standard, this plot suggests six primary factors. Additionally, the Nelson’s (2005) threshold was not achieved by the five-factor solution (R² = .78) but was satisfied by the six-factor solution (R² = .93). Similarly, when we conducted a MAP test on the correlation matrix of this subsample, a six-factor solution was suggested.

Furthermore, when examining the varimax-rotated factor loadings for the high cognitive complexity group, only 92.50% of the variables displayed their highest loadings on the factor for which they were expected to be markers. Moreover, of these correct loading items, only 72% possessed factor loadings that were at least double their loading on any other factor and 92% had a second factor loading that exceeded |.30|. Thus, as with the below average cognitive complexity subgroup, there appeared to be distortion of the factor structure of the above average cognitive complexity subgroup. Individuals demonstrating a higher level of cognitive complexity did not exhibit the clear five-factor structure that was demonstrated by those with an average level of cognitive complexity.

Post hoc Analyses

Three-factor solution. To explore further the possibility of factor distortion we examined the three-factor solution recorded by individuals in the below average cognitive complexity subgroup in order to ensure that the three-factor solution did not fit the average cognitive complexity and above average cognitive complexity subgroups. Thus, a series of principle component analyses were conducted, based on the three-factor solution suggested by the initial results recorded by the below average cognitive complexity subgroup. When examining the varimax-rotated factor loadings, for the average cognitive complexity subgroup, only 80% of the variables displayed their highest loading on the factor for which they were expected to be markers, with only 72% of these variables possessing factor loadings that were at least double their loading on any other factor. Similarly, for the above average cognitive complexity subgroup, only 55% of the variables displayed their highest loading on the factor for which they were expected to be markers, with only 59% of these variables possessing factor loadings that were at least double their loading on any other factor. Thus, the three-factor model suggested by the below average cognitive complexity subgroup did not appear to be appropriate for either the average or the above average cognitive complexity subgroups.

Six-factor solution. As with the three-factor model, in order to further explore the factor distortion of the above average cognitive complexity subgroup and ensure that the six-factor solution did not fit the average or below average subgroups, a series of factor analyses were conducted based on the six-factor solution suggested by the initial above average cognitive complexity subgroup results. When examining the varimax-rotated factor loadings, neither the average nor the below average subgroups provided a good fit. For the average subgroup, only 85% of the variables had their highest loading on the factor for which they were expected to be markers and, of these items, only 76% of these variables recorded factor loadings that were at least double their loading on any other factor. Similarly, for the below average subgroup, only 70% of the variables had their highest loading on the factor for which they were expected to be markers and of these items, 89% of these variables recorded factor loadings that were at least double their loading on any other factor. Thus, as with the three-factor model, the six-factor model suggested by the above average cognitive complexity subgroup did not appear to be appropriate for either the average or the below average cognitive complexity subgroups.

Discussion

In this study we examined the stability of the factor structure of a commonly used five-factor measure developed by Saucier (1994). As noted by Bowler et al. (2009), the universality of the use of five factors in such measures may be questionable. Specifically, these authors concluded that the individual differences in cognitive complexity, that is, the number of social dimensions that an individual can cognitively maintain at any given moment (see Kelly, 1955; Labouvie-Vief & Diehl, 2000; Vannoy, 1965), alters the number of factors that an individual reports when responding to self-report measures of personality. That is, they noted that both the total sample of their participants and the subgroup with an average level of cognitive complexity displayed a five-factor structure. In contrast, individuals with a below average level of cognitive complexity displayed a three-factor structure whereas individuals with above average levels of cognitive complexity displayed a seven-factor structure.

In line with these previous findings, our results in this study further support the notion that cognitive complexity impacts the factor structure an individual displays on self-report measures of the FFM. Specifically, individuals with a lower level of cognitive complexity displayed fewer factors than did individuals with an average or higher level of cognitive complexity. Stated another way, when responding to traditional Likert-based self-report measures, individuals who are cognitively less complex appear to record responses that indicate less complex personalities, whereas individuals who are cognitively more complex record responses that indicate more complex personalities.

The nature of cognitive complexity may provide an answer as to why differences in level of cognitive complexity would impact the factor structure of traditional self-report measures of personality. Bieri (1955) originally conceived of cognitive complexity as how well individuals could differentiate between elements in their environment. Also included in this conceptualization was the ability to differentiate between oneself and others. There may be a relationship between cognitive complexity and measurement formats. When individuals respond to questions using the Likert format, they are required to differentiate between five or seven levels of a construct. To clarify, the respondent must differentiate between the five or seven categories of behavior incorporated in the Likert scale. A cognitively complex individual will be likely to be able to differentiate between these different levels of a construct, whereas a cognitively simple individual may have difficulty in differentiating between these constructs, and, thus, may record responses that show less variability among the levels.

Limitations

The type of analysis that we were able to carry out limited this study. Given the large size of the correlation/covariance matrices being analyzed, a confirmatory factor analysis was impractical. In practice, a confirmatory factor analysis will create item packets by averaging pairs of highly correlated items. The goal in this study was to determine whether the factor structure would change because of differences in the cognitive complexity of the individuals being tested. It was for this reason that principal components analysis was used. Moreover, in order to remain in line with the development of the FFM measure, a varimax rotation was used.

Future Research

One potential avenue for future research in this area is the role played by measures that use frequency-based items (Edwards & Woehr, 2007). Unlike Likert-type items, in which respondents are asked to describe themselves, with frequency-based personality measurement respondents are asked to think about their recent behavior and allot percentages to the amount of time (i.e., percent of time it was inaccurate, percent of time it was neither inaccurate nor accurate, and percent of time it was accurate) that a particular statement (e.g., I am the life of the party) is descriptive of their behavior. Subsequently, respondents are not asked to describe themselves but to think about their behavior. The frequency-based response format is known to be less susceptible to deliberate response distortion, and can increase predictive validity (Fleisher, Woehr, Edwards, & Cullen, 2011). The resiliency of frequency-based personality measurement to changes in factor structure provides added support for more widespread application.

Edwards and Woehr (2007) noted several additional advantages to using frequency-based personality measurement. Most importantly, the frequency-based measure provides data on temporal stability and behavioral consistency in one test administration, and these features are available only with multiple administrations of a Likert-type measure. Secondly, frequency-based and Likert-type formats show similar reliabilities and convergent validity. Third, whereas traditional response formats require respondents to mentally calculate the average amount of a given behavior, the frequency-based measure allows for an easier appraisal of how often a particular behavior is, or is not, present. This easier appraisal may be the feature of the method that diminishes the differences in factor structure as compared to Likert scales. Thus, we suggest that this may be a method by which the distortion of cognitive complexity on the factor structure of common FFMs may potentially be reduced.

An additional avenue for future research is the examination of the impact cognitive complexity has on how an individual rates another person. For example, when rating a peer, do the ratings of the individual doing the rating follow the same pattern as self-ratings? Essentially, this focus would help to ascertain the true significance and nature of the phenomenon that we have noted. Specifically, is this an issue of perception or is this an issue of personality? If it is the former, then we can expect peer ratings to follow the same pattern, so that individuals with a low level of cognitive complexity should perceive others in a more simplistic manner (e.g., three factors) and those with a high level of cognitive complexity should perceive others in a more complex manner (e.g., six factors). Essentially, individuals would be viewing others in the same dimensional manner as they view themselves. If, however, it is an issue of personality, then this would suggest that cognitive complexity may play a role in the development of individual personality and in the consistency of individual behaviors.

Conclusion

Our results in this study further challenge the assumption of the universality of the FFM. As in the study by Bowler and colleagues (2009), although individuals in both the entire sample and the average cognitive complexity subgroup clearly displayed adherence to the five-factor structure, individuals in both the below average and above average cognitive complexity subgroups displayed factor structures that diverged from the traditional model. We do not believe that these results invalidate the FFM, as, according to our results, it is relevant for at least 50% of the population. However, we do question the impact that this may have on the validity of the individual dimensions of this model. For example, in the organizational behavior literature, dimensions of the FFM rarely, if ever, have validities greater than |.30| (see e.g., Barrick & Mount, 1991; Judge, Heller, & Mount, 2002). This could be caused by an incongruence of measurement bandwidth between predictor (i.e., five-factor measure) and criterion (supervisor). Thus, in addition to evaluating the impact of cognitive complexity on alternative measures of the FFM (e.g., the NEO-FFI) with different response formats such as frequency-based items, future researchers should explore the potential of any differential prediction engendered by the difference factor structures produced by cognitive complexity.

Almagor, M., Tellegen, A., & Waller, N. G. (1995). The Big Seven model: A cross-cultural replication and further exploration of the basic dimensions of natural language trait descriptors. Journal of Personality and Social Psychology, 69, 300-307. http://doi.org/b6fxr4

Aluja, A., García, O., García, L. F., & Seisdedos, N. (2005). Invariance of the ‘NEO-PI-R’ factor structure across exploratory and confirmatory factor analyses. Personality and Individual Differences, 38, 1879-1889. http://doi.org/btph6r

Ashton, M. C., Jackson, D. N., Helmes, E., & Paunonen, S. V. (1998). Joint factor analysis of the Personality Research Form and the Jackson Personality Inventory: Comparisons with the Big Five. Journal of Research in Personality, 32, 243-250. http://doi.org/d35383

Austin, E. J., Deary, I. A., Whiteman, M. C., Fowkes, F. G. R., Pedersen, N. L., Rabbitt, P., … McInnes, L. (2002). Relationships between ability and personality: Does intelligence contribute positively to personal and social adjustment? Personality and Individual Differences, 32, 1391-1411. http://doi.org/cqfpxk

Barrick, M. R., & Mount, M. K. (1991). The Big Five personality dimensions and job performance: A meta-analysis. Personnel Psychology, 44, 1-26. http://doi.org/hjv

Barrick, M. R., & Mount, M. K. (1996). Effects of impression management and self-deception on the predictive validity of personality constructs. Journal of Applied Psychology, 81, 261-272. http://doi.org/b6db97

Bartlett, M. S. (1950) Tests of significance in factor analysis. British Journal of Mathematical and Statistical Psychology, 3, 77-85. http://doi.org/dgx

Bieri, J. (1955). Cognitive complexity-simplicity and predictive behavior. The Journal of Abnormal and Social Psychology, 51, 263-268. http://doi.org/ch3sds

Bieri, J., Atkins, A. L., Briar, S., Leaman, R. L., Miller, H., & Tripodi, T. (1966). Clinical and social judgment. New York: Wiley.

Block, J. (1995). A contrarian view of the five-factor approach to personality description. Psychological Bulletin, 117, 187-215. http://doi.org/cjzm9t

Bowler, M. C., Bowler, J. L., & Phillips, B. C. (2009). The Big-5 + 2? The impact of cognitive complexity on the factor structure of the five-factor model. Personality and Individual Differences, 47, 979-984. http://doi.org/cph8m2

Cattell, R. B. (1933). Temperament tests. II: Tests. British Journal of Psychology, 24, 20-49. http://doi.org/ff7g5h

Cattell, R. B. (1966). Handbook of multivariate experimental psychology. Chicago, IL: Rand-McNally.

Costa, P. T., Jr., & McCrae, R. R. (1992). NEO Personality Inventory professional manual. Odessa, FL: Psychological Assessment Resources.

Devaraj, S., Easley, R. F., & Crant, J. M. (2008). How does personality matter? Relating the five-factor model to technology acceptance and use. Information Systems Research, 19, 93-105. http://doi.org/dgvsjn

DeYoung, C. G. (2010). Toward a theory of the Big Five. Psychological Inquiry: An International Journal for the Advancement of Psychological Theory, 21, 26-33. http://doi.org/b6ktdt

Edwards, B., & Woehr, D. J. (2007). Examination and evaluation of frequency estimation based personality measurement. Personality and Individual Differences, 43, 803-814.

Erdheim, J., Wang, M., & Zickar, M. J. (2006). Linking the Big Five personality constructs to organizational commitment. Personality and Individual Differences, 41, 959-970. http://doi.org/csbkbs

Feixas, G., Moliner, J. L., Montes, J. N., Mari, M. T., & Neimeyer, R. A. (1992). The stability of structural measures derived from repertory grids. International Journal of Personal Construct Psychology, 5, 25-39. http://doi.org/fbqd86

Fiske, D. W. (1949). Consistency of the factorial structure of personality ratings from different sources. The Journal of Abnormal and Social Psychology, 44, 329-344. http://doi.org/bfxmws

Fiske, D. W. (1994). Two cheers for the Big Five! Psychological Inquiry, 5, 123-124. http://doi.org/dsnrxx

Fleisher, M. S., Woehr, D. J., Edwards, B. D., & Cullen, K. L. (2011). Assessing within-person personality variability via frequency estimation: More evidence of a new measurement approach. Journal of Research in Personality, 45, 535-548. http://doi.org/b4d4vx

Funder, D. C. (2001). Personality. Annual Review of Psychology, 52, 197-221. http://doi.org/dgvxw8

Funder, D. C., & Sneed, C. D. (1993). Behavioral manifestations of personality: An ecological approach to judgmental accuracy. Journal of Personality and Social Psychology, 64, 479-490. http://doi.org/crbmv8

Garnett, J. C. M. (1919). General ability, cleverness, and purpose. British Journal of Psychology, 9, 345-366. http://doi.org/bt65zq

Goldberg, L. R. (1992). The development of markers for the Big Five factor structure. Psychological Assessment, 4, 26-42. http://doi.org/cd8

Hines, D. A., & Saudino, K. J. (2008). Personality and intimate partner aggression in dating relationships: The role of the “Big Five”. Aggressive Behavior, 34, 593-604. http://doi.org/bgzmh3

Holden, R. R. (2008). Underestimating the effects of faking on the validity of self-report personality scales. Personality and Individual Differences, 44, 311-321. http://doi.org/bx3nrc

Jackson, D. N., Paunonen, S. V., Fraboni, M., & Goffin, R. D. (1996). A five-factor versus six-factor model of personality structure. Personality and Individual Differences, 20, 33-45. http://doi.org/cq845d

Johnson, J. W. (1994, April). The effect of rater cognitive complexity on performance rating effectiveness. Paper presented at the 9th annual meeting of the Society for Industrial and Organizational Psychology, Nashville, TN.

Judge, T. A., Heller, D., & Mount, M. K. (2002). Five-factor model of personality and job satisfaction: A meta-analysis. Journal of Applied Psychology, 87, 530-541. http://doi.org/dpzz84

Kaiser, H. F., & Rice, J. (1974). Little jiffy, mark IV. Educational and Psychological Measurement, 34, 111-117. http://doi.org/ftb8rg

Kelly, G. A. (1955). The psychology of personal constructs. New York: Norton.

Labouvie-Vief, G., & Diehl, M. (2000). Cognitive complexity and cognitive affective-integration: Related or separate domains of adult development. Psychology and Aging, 15, 490-504. http://doi.org/ddpkqx

Major, D. A., Turner, J. E., & Fletcher, T. D. (2006). Linking proactive personality and the Big Five to motivation to learn and development activity. Journal of Applied Psychology, 91, 927-935. http://doi.org/ds6gqj

Miller, J. D., Pilkonis, P. A., & Morse, J. Q. (2004). Five-factor model prototypes for personality disorders: The utility of self-reports and observer ratings. Assessment, 11, 127-138. http://doi.org/frx29x

Mischel, W. (1968). Personality and assessment. New York: Wiley.

Nelson, L. R. (2005). Some observations on the scree test, and on coefficient alpha. Thai Journal of Educational Research and Measurement, 3, 1-17.

O’Connor, B. P. (2000). SPSS and SAS programs for determining the number of components using parallel analysis and Velicer’s MAP test. Behavior Research Methods, 32, 396-402. http://doi.org/d8zf2h

Orzeck, T., & Lung, E. (2005). Big Five personality differences of cheaters and noncheaters. Current Psychology, 24, 274-287. http://doi.org/dtkvd8

Ployhart, R. E., Lim, B.-C., & Chan, K.-Y. (2001). Exploring relations between typical and maximum performance ratings and the five-factor model of personality. Personnel Psychology, 54, 809-843. http://doi.org/dkxvmd

Rammstedt, B., Goldberg, L. R., & Borg, I. (2010). The measurement equivalence of Big Five factor markers for persons with different levels of education. Journal of Research in Personality, 44, 53-61. http://doi.org/fvrgvx

Raynor, D. A., & Levine, H. (2009). Associations between the five-factor model of personality and health behaviors among college students. Journal of American College Health, 58, 73-81. http://doi.org/ctfv36

Robinson, O. C., Demetre, J. D., & Corney, R. (2010). Personality and retirement: Exploring the links between the Big Five personality traits, reasons for retirement and the experience of being retired. Personality and Individual Differences, 48, 792-797. http://doi.org/czznjx

Saucier, G. (1994). Mini-markers: A brief version of Goldberg’s unipolar Big Five markers. Journal of Personality Assessment, 63, 506-516. http://doi.org/cfq

Saucier, G., & Goldberg, L. R. (1996). The language of personality: Lexical perspectives on the five-factor model. In J. S. Wiggins (Ed.), The five-factor model of personality: Theoretical perspectives (pp. 21-50). New York: Guilford.

Shi, J., Lin, H., Wang, L., & Wang, M. (2009). Linking the Big Five personality constructs to organizational justice. Social Behavior and Personality: An international journal, 37, 209-222. http://doi.org/chd6q6

Simms, L. J. (2007). The Big Seven model of personality and its relevance to personality pathology. Journal of Personality, 75, 65-94. http://doi.org/bzg64t

Srivastava, S. (2010). The five-factor model describes the structure of social perceptions. Psychological Inquiry, 21, 69-75. http://doi.org/d2fqcg

Toomela, A. (2003). Relationships between personality structure, structure of word meaning, and cognitive ability: A study of cultural mechanisms of personality. Journal of Personality and Social Psychology, 85, 723-735. http://doi.org/cv4d69

Vannoy, J. S. (1965). Generality of cognitive complexity-simplicity as a personality construct. Journal of Personality and Social Psychology, 2, 385-396. http://doi.org/dhjjb2

Velicer, W. F. (1976). Determining the number of components from the matrix of partial correlations. Psychometrika, 41, 321-327. http://doi.org/czsxbw

Velicer, W. F., Eaton, C. A., & Fava, J. L. (2000). Construct explication through factor or component analysis: A review and evaluation of alternative procedures for determining the number of factors or components. In R. D. Goffin & E. Helmes (Eds.), Problems and solutions in human assessment (pp. 41-71). Norwell, MA: Kluwer.

Waller, N. G., & Ben-Porath, Y. S. (1987). Is it time for clinical psychology to embrace the five-factor model of personality? American Psychologist, 42, 887-889. http://doi.org/bbrwvq

Webb, E. (1915). Character and intelligence: An attempt at an exact study of character. Cambridge, UK: Cambridge University Press.

Wiggins, J. S. (1996). The five-factor model of personality: Theoretical perspectives. New York: Guilford.

Woehr, D. J., Miller, M. J., & Lane, J. A. S. (1998). The development and evaluation of a computer- administered measure of cognitive complexity. Personality and Individual Differences, 25, 1037-1049. http://doi.org/bsh784

Yamagata, S., Suzuki, A., Ando, J., Ono, Y., Kijima, N., Yoshimura, K., … Jang, K. L. (2006). Is the genetic structure of human personality universal? A cross-cultural twin study from North America, Europe, and Asia. Journal of Personality and Social Psychology, 90, 987-998. http://doi.org/cfxnwc

Zuckerman, M., Kuhlman, D. M., Joireman, J., Teta, P., & Kraft, M. (1993). A comparison of three structural models for personality: The Big Three, the Big Five, and the Alternative Five. Journal of Personality and Social Psychology, 65, 757-768. http://doi.org/b33ntc

Zwick, W. R., & Velicer, W. F. (1982). Factors influencing four rules for determining the number of components to retain. Multivariate Behavioral Research, 17, 253-269. http://doi.org/b53mfs

Mark C. Bowler, Department of Psychology, East Carolina University, Rawl 104, Greenville, NC, 27858, USA. Email: [email protected]