UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A comparison of Likert and Q-sort scaling techniques in the assessment of personality Hanson, Carrie Elizabeth 1996

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1996-0517.pdf [ 2.27MB ]
Metadata
JSON: 831-1.0099046.json
JSON-LD: 831-1.0099046-ld.json
RDF/XML (Pretty): 831-1.0099046-rdf.xml
RDF/JSON: 831-1.0099046-rdf.json
Turtle: 831-1.0099046-turtle.txt
N-Triples: 831-1.0099046-rdf-ntriples.txt
Original Record: 831-1.0099046-source.json
Full Text
831-1.0099046-fulltext.txt
Citation
831-1.0099046.ris

Full Text

A COMPARISON OF LIKERT AND Q-SORT SCALING TECHNIQUES IN THE ASSESSMENT OF PERSONALITY by CARRIE ELIZABETH HANSON B.A., University of California, Riverside, 1993 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES Department of Psychology We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA June 1996 © Carrie Elizabeth Hanson, 1996 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia Vancouver, Canada Department DE-6 (2/88) II Abstract This study involved a comparison of two scaling techniques commonly used in the area of personality assessment: the Likert scale, which is a normative, free response technique, and the Q-sort, which is an ipsative, forced-choice technique. Comparisons were made in three major areas. Firstly, the internal consistencies of scales were compared across the two formats. Secondly, the distribution of participants' responses were examined in the two formats. Finally the factor structures of measures in Likert and Q-sort format were compared. Participants were 130 students (92 women and 38 men). Mean age of subjects was 19. Three measures were used: the California Q-set (CAQ) , the NEO-Five Factor Inventory (NEO) , and the Interpersonal Adjective Scales (IAS). The C A Q is a 100-item Q-sort measure derived by clinicians to capture the full range of personality traits. The N E O and IAS are empirically derived, well-validated personality scales. The N E O is a 60-item measure of the Five-Factor Model , and the IAS is a 64-ilem measure of the interpersonal domain of Dominance and Love. A l l three measures were administered both in Likert and Q-sort format. Participation was spread over two sessions, separated by approximately three weeks, with either Likert or Q-sort format given in each session. Results suggest several differences between the two scaling techniques. Internal consistencies were higher for Likert versions of all three measures. Distribution of responses for Likert measures departed substantially from the forced nature of the Q-sort. In terms of factor structure, a clear structure was obtained for the IAS and N E O in both Likert and Q-sort formats. A moderately clear structure was obtained for C A Q in Likert format. The structure of the C A Q in Q-sort format was found to be quite ambiguous. Across all three measures Likert solutions were found to account for a greater percentage of variance. Implications for the application of these two scaling techniques are discussed. T A B L E OF C O N T E N T S in Abstract ii List of Tables v List of Figures vi Acknowledgments vi i Introduction 1 Ipsative versus Normative Measurement 2 Limitations in Interpretation of Ipsative Measures 5 Limitations in Statistical Application of Ipsative Measures 5 The Q-sort Method 7 The California Q-Set 8 Comparison of Subjects' Q-sorls with a Prototype Q-sort 9 Factor Analysis of the C A Q 10 Overview of Study 10 Method 11 Participants 11 Measures 11 C A Q (Forced-Q) 11 C A Q ( L i k e r t ) 12 NEO-Five Factor Inventory (Likerf) 12 NEO-FFI (Forced-Q) 13 I V Interpersonal Adjective Scales (Likert) 13 IAS (Forced-Q) 13 Design and Administration 14 Results 14 Comparison of Scores at the Scale, Item, and Profile Levels 14 Comparison of Likert Value Distributions with Q-sort Category Distributions 15 Factor Analyses 16 Structure of C A Q 16 Structure of N E O 18 Structure of IAS 18 Circumplex Analysis of the IAS 19 The "Optimally Adjusted Personality" Q-sort Prototype 19 Discussion 20 Factor Structure 22 Directions for Future Research 24 Conclusions 25 References 27 LIST OF T A B L E S Table 1. Scale means and, internal consistencies, and correlations between Likert and forced-Q data 32 Table 2. Correlations between item means for Likert and forced-Q 33 Table 3. Mean correlations between subject profiles for Likert and forced-Q 34 Table 4. Percentage of variance accounted for by factors of IAS, N E O , and C A Q items in forced-Q And Likert format 35 fable 5. California Q-set items defining five factors (in Likert format) 36 fable 6. California Q-set items defining live factors (in forced-Q format) 39 Table 7. Correlations of C A Q factors in Likert and forced-Q format with NEO-FFI domain scales 42 Table 8. N E O items defining live factors (in Likert format) 43 Table 9. N E O items defining live factors (in forced-Q format) 46 Table 10. IAS items defining four factors (in Likert format) 49 Table 11. IAS items defining four factors (in forced-Q format) 52 Table 12. Factor scores for IAS Likert scales on the interpersonal factors Love and Dominance 54 Table 13. Factor scores for IAS forced-Q scales on the interpersonal factors Love and Dominance 55 Table 14. Factor loadings of C A Q items on five factors (from McCrae, Costa, & Busch, 1986) 56 vi LIST OF F I G U R E S Figure 1. Example distribution of Q-sorl ratings 59 Figure 2. Distribution of Likert ratings for the N E O 60 Figure 3. Distribution of Likert ratings for the C A Q 61 Figure 4. Distribution of Likert ratings for the IAS 62 Figure 5. Similarity scores to the "Optimally Adjusted Personality" for Likert and forced-Q data 63 Vll Acknowledgments I would like to express thanks to several people, without whose help and encouragement this undertaking would have been a great deal more difficult, or possibly impossible. First, I would like to thank my Mom and Dad and other sundry family members for their ever-gentle telephonic incitements. I would also like to thank my friends/relatives Toni, Mary, and Eliot for their endless Internet inspiration, as well as their humorousness. Sincere gratitude also goes out to several Vancouver friends: Steven for his technical transformations, Jodi for her moral support, Daria for her computer expertise, David for his motivation. Monica for her helpful feedback, Shawn for his reality checks. 1 would especially like to thank Paul Trapncll and Michelle Yik for their constant techno-emotional support. Many thanks also to my committee members, .lames Steiger for his very useful feedback on my manuscript, and .Mm Russell for his encouragement in the earlier stages of my thesis. Finally, I would like to thank my advisor, Jerry Wiggins, for his patience and support, and for all that he has taught me. 1 Introduction In the social sciences the way we ask our questions strongly influences the responses that we get and the way we interpret out data. For example, in memory research older adults and amnesics show similar performance on implicit memory tasks when compared to young adults, but impaired memory performance on explicit memory tasks (Graf. 1990; Graf & Schacter, 1985). These tasks measure memory for the same information but differ in the type of memory retrieval; explicit memory tasks require conscious or deliberate recollection, whereas implicit memory tasks involve memory without awareness. Another example is in the area of recognition of emotion from facial expressions. When asked to identify expressions of emotions in photographs, respondents give very different answers depending on whether they are asked an open-ended question (e.g., "What emotion is this person experiencing"?) or given a list of possible emotions from which to choose (Russell, 1994). In the area of personality assessment the way we ask our questions influences such things as how honest people are, how much they try to present themselves in a favorable light, and how careful they are in completing the task at hand. For this reason there is considerable debate about whether personality measures should be objective or projective, empirically or theoretically derived, and so forth. In objective personality assessment the issue of how to ask questions is intimately associated with scaling techniques. Should I use a True/False format, should 1 ask subjects to rank order traits, or should I use a Likert scale? If I use a Likert scale, should I use live places, seven places, or 10 places? Vclicer and his colleagues, in the late 1970s and early 1980s (e.g.. Vclicer & Stevenson, 1978; Oswald & Velieer, 1980), were parlieulary interested in the issue of how sealing techniques influence the structure of personality measures and hence their reliability and validity. Until this time True/False scaling formats were used almost invariably in personality assessment, due to ease of administration and ease of scoring. Velieer and colleagues found substantial differences in structure and reliability of personality measures when True/False versus Likert scales were used. Specifically, they found cleaner structure, a larger percentage of variance accounted for, and better internal consistencies with the Likert format than the True/False format. A related issue in the assessment of personality is whether participants should be allowed to respond freely to items or whether their responses should be forced. This study compares some characteristics of two types of scaling methods commonly in use in the area of personality psychology: the Likert method (which is a normative, unforced method), and the Q-sort method (which is an ipsative, forced choice method). Ipsative versus Normative Measurement Ipsative measures are those in which the total sum of scores across measures for each subject is a constant, such as paired comparisons and ranking procedures. For example, if subjects are asked to rank order 10 traits, then each subject will have the same total score (1+2 +... + 10). This is in contrast to normative measures, such as rating scales, in which there is (presumably) considerable variability in subjects' total scores. For example, with a Likert scale, subjects are free to give any value within a specified range. There are several differences between ipsative and normative measures. Ipsative measures examine a number of traits in relation to each other within a particular individual, so 3 ipsative scores are distributed about the mean of the individual. In ipsative measurement, an individual's scores on the various measures are dependent on his/her other scores, and are independent of anyone else's scores. Normative scores, on the other hand, examine a trait or traits with reference to a population, and are therefore distributed about the mean of that population. In normative measurement, an individual's score on any given measure is independent of his/her other scores, but is statistically dependent on the scores of everyone else in the population, as is the case with standardized scores. In addition to the definition of ipsativity given above, there is a more lenient one (Guilford, 1952; Cattell, 1944) related to the dependency inherent in such measures. According to this more lenient criterion, "any test is ipsative which produces intraindividual assessment of variables of a type such that a score elevation on one attribute necessarily produces a score depression on other attribute(s)" (Micks, 1970, p. 170). Measures which meet this definition are termed partially ipsative measures, whereas those which meet the more stringent criterion are called purely ipsative measures. Ipsativity can also be conceptualized along a continuum, where the degree of ipsativity is presumed to be an inverse function of the amount of variability in the scores (Hicks, 1970). Normative measures are sometimes criticized because of various biases associated with them (Saville & Willson. 1991). Critics claim that such biases as acquiescence, socially desirable responding, and central tendency responding are inherent in normative measures. In fact, forced choice tests originally came about to mitigate alleged biasability in non-cognitive psychological instruments (Hicks, 1970; Zavala, 1965). Because of the lack of variability between subjects in ipsative measures, they presumably are not as susceptible to such biases 4 as are normative measures. However, there are several problems associated with ipsative data as well. First, forced choice may not be the most natural way for subjects to respond. Some favor ipsative measures because they claim that "life is aboul choices" (Saville & Willson, 1991, p. 221). In market research it is often difficult for subjects to rate different products numerically, but it is usually easy lor them to rank-order products (e.g.. Green & Tul l , 1982). In geography it is possible to get a good estimate of interval distances between cities by obtaining several paired comparisons or rankings (Shephard, 1966). However, it may not be possible to generalize these findings to the area of personality research. A free response format may be more natural for respondents than a format which uses forced choice. Secondly, although proponents of ipsative measures believe them to have better reliability and validity than normative measures, because of the alleged response sets in normative measures, there does not seem to be any strong evidence of this. In the ease of partially ipsative measures, there is sometimes a fairly high degree of predictive validity, and in some cases it is even higher than in respective normative measures. However, there is no systematic explanation for this, and it may be due to chance alone. In general, as ipsativity increases, validity decreases. In the case of purely ipsative measures, there is no evidence that they have ever had belter validity than normative measures (Hicks, 1970). As for reliability, normative measures are generally found to be at least as high as ipsative measures (Merritt & Marshall, 1984), i f not higher (Thompson, Levitov, & Miederhoff, 1982). A final problem with ipsative measures is that there are limitations associated with their interpretation and statistical application. Although the use of ipsative measures is widespread. Hicks (1970) claims that "an examination of careful methodological studies of purely ipsative 5 measurement techniques indicates that many researchers are unaware of the narrow limits within which interpretation may validly proceed and standard statistical techniques may legitimately be applied" (p. 181). Presumably, this is also true for partially ipsative techniques, although the effects should be less extreme. Limitations in Interpretation of Ipsative Measures It must be stressed that ipsative measures are only useful for looking at differences in trait levels within each individual rather than between individuals. There is no universal scale, so although Participant A and Participant B might give a particular item top ranking, in relation to the other items in the set, the item may be extremely characteristic of Participant A and only moderately characteristic of Participant B. The same is true with ipsative data in non-psychological settings. If respondents are asked to rank several brands of a product in the order of preference, one subject who gives an item the highest ranking may greatly prefer it, whereas another subject who gives the same product the highest ranking may not like any of the choices, but believe it is the best choice by default. Because of this lack of a single scale, it is impossible to compare subjects using ipsative measures. A high score on a particular item would have a very peculiar meaning, something like "this individual is higher on this variable relative to his scores on other variables than are other individuals 1 scores on this variable relative to their scores on other variables" (Hicks, 1970, p. 168; also sec Katz, 1962). Limitations in Statistical Application of Ipsative Measures 6 There are problems associated with ipsative data when they are factor analyzed. The interdependence of items in ipsative measures presents some serious problems in intercorrelation matrices. Because of the dependence in the scores, there tends to be an unrealistic proportion of negative values in ipsative intercorrelation matrices. In fact, it is highly unlikely that any really large positive correlations wil l be found, even among items measuring practically the same construct. A data matrix with many negative correlations wil l tend to produce bipolar factors, and this is what happens when a factor analysis is performed on an ipsative correlation matrix. However, these factors represent artifactual, rather than true relationships in the data. Any true relationships wil l be obscured by the artifactual factor pattern, as it wi l l "overwhelm any factor structure seen with normative factor analysis" (Dunlap & Cornwell, 1994, p. 116). The conclusion of Johnson and his colleagues (1988) is that "manipulating ipsative measures as i f they were normative measures is an exercise in futility, like cheating at patience" (p. 161). Hicks (1970) goes as far as to conclude that ipsative measures should not be used unless significant response bias exists, the bias reduces validity, and an ipsative format successfully diminishes bias and increases validity to a greater extent than do non-ipsative controls for bias. He warns that even in the unlikely event that these criteria are met, the researcher still needs to realize the limitations of his/her ipsative data and proceed accordingly, rather than treating the ipsative data as i f they were normative. Despite such cautions, many researchers have treated ipsative data as i f they were normative. Many have factor analyzed ipsative measures, despite the fact that they violate the assumption of independence in intercorrelation matrices. Certain areas of psychology have been researched almost entirely with purely ipsative measures, and some areas which have used both ipsative and normative measures have come up with contradictory results, which are probably 7 due at least in part to the different methods of measurement. Hicks (1970) calls for a re-evaluation of the research which has relied on purely ipsative measures, and/or has used statistical techniques with assumptions that are violated with ipsative data. He suggests that ipsative measures be readministered, along with respective normative versions, so that comparisons can be made. It must be pointed out that the problems cited above are associated with purely ipsative measures. The characteristics of partially ipsative measures are not as clear, although any problems associated with them are probably not as severe. This study is concerned with comparing one particular form of partially ipsative measurement, the Q-sort method, with the Likert or rating method, a normative form of measurement. The Q-Sorl Method The Q-sort method was originally developed by Stephenson (1953) in his research on Q factor analysis, in which persons as opposed to scales serve as variables. Stephenson invented the Q-sort method as a scaling technique suitable to this form of factor analysis. Although the Q-sort method was originally associated with Q factor analysis, it is now considered a scaling technique in its own right (Block, 1978). The technique consists of arranging items into categories to describe an individual. Items uncharacteristic of the individual are given low values, items characteristic of the individual are given high values, and neutral items are placed in middle categories. The difference between the Q-sort method and traditional Likert ratings is that with the Q-sorl method certain constraints are placed on the rater. Specifically, there is a predetermined number of items for each category, with few items allowed in extreme categories, and many items allowed in middle categories. Because of these constraints, it is necessary that 8 the rater be able to rearrange items until the desired ordering is obtained. For this reason, Q-sorts are traditionally printed on cards to make it easier to move items around. The California Q-Sel Probably the most well-known and widely used Q-sort measure is the California Q-set ( C A Q ; Block, 1978). The C A Q was designed to enable observers to have a common language with which to describe a subject's personality (Block, 1978). The C A Q was constructed by numerous panels of clinical psychologists over lime, and the aim was an instrument with comprehensive coverage of the domain of personality. The C A Q was originally intended for use by professional raters, but i l has since been modified for use by non-professional observers and for self-description (Bern and Funder, 1978). Block (1978) justifies the forced nature of the C A Q for several reasons. First, he claims, i f the Q-sort had an unforced distribution, the degree of similarity between raters would be obscured. That is, i f two raters agreed on the ordering oi'all items but used the scale differently, the degree of similarity between their ratings would be attenuated. Secondly, he believes that the Q-sort in unforced form is more susceptible to the Barnum effect, the tendency to say very generally true things about a subject. Third, according to Block, the Q-sort in unforced format is no more reliable than in forced format even though the former requires finer discriminations. Fourth, Block asserts that the unforced Q-sort does not provide more information than the forced Q-sort. Finally, he claims that Q-sort data in unforced form are unwieldy and difficult, whereas data in forced form are convenient and accessible. As for the rationale for the shape of the C A Q , Block (1978) believes that 9 a symmetrical distribution should be used simply because it is neutral and uncommitted and a skewed distribution is too difficult. He claims that the distribution of Q-sort items should be unimodal as opposed to rectangular or any other shape for the following reasons: (1) One study (Livson & Nichols, 1956) showed that judges prefer a unimodal distribution on average, (2) Items placed in middle categories are psychologically less important, (3) Placements in middle categories are most difficult and time consuming for judges, and (4) Comparisons made between Q-sorts using correlations (see next section) are more sensitive to extreme placements than to middle placements. Comparison of Subjects ' Q-sorts with a Prototype Q-sort According to Block, a unique feature of the CAQ is that two Q-sorts can be correlated, using a Pearson product moment correlation, in order to get an index of similarity between the sorts. For example, the similarity between two judges' Q-sorts on one subject can be found. A special application of the Q-sort, according to Block, is the comparison of subjects' sorts with a prototype or criterion Q-sort. CAQ prototypes are constructed by having a number of people, such as clinicians or professional researchers, construct a Q-sort for a particular type of personality. Then all of their Q-sorts are "requed", meaning that all of the items for all the sorters are given points depending upon the categories in which they were placed. The five items with the most points are placed in Category Nine, the eight with the next highest points are placed in Category Eight, and so on. The resulting Q-sort is a prototype for that particular type of personality. Once a prototype is developed, subjects' Q-sorts can be correlated with the prototype Q-sort, yielding an index of similarity between the subject and the prototype. Some 10 examples of prototypes that have been developed are Hysteria (Block, 1978), Narcissism (Wink, 1991), and the "optimally adjusted personality" (Block, 1978). Fuel or Analysis of I he CAQ McCrae, Costa, & Buseh (1986) report recovering the dimensions of the Five-Factor Model from C A Q sell-report data. The live factors Extraversion, Agreeableness, Neuroticism, Conscientiousness, and Openness to Experience have repeatedly emerged from normative measures, and this model is the most agreed-upon model of personality structure currently in the Held (see Method section for a more complete discussion). McCrae et al. conclude that the "correspondences are...remarkable when it is recalled that the Q-sort method assesses personality through the relative salience of traits within the individual. When treated normatively, these ipsative data show the same structure as do conventional questionnaires" (pp. 442-443). However, different analyses have found conflicting results, and it seems there are several possible interpretations of the factor structure in the C A Q . Lanning (1994) found five factors to be necessary but not sufficient in the structure of the C A Q . Using composite expert Q-sorts, he found that between 3 and 7 additional factors are necessary to account for the variance in C A Q items. Other studies have found varied results (e.g., Lieberman & Tobin, 1983; Ilaan, 1981; Lorr, 1978). Overvie w of Study As mentioned above, Velieer and his colleagues were interested in the structure of the same instrument in True/False versus Likert format. This study is quite similar, except that I examine the structure of the same instrument in Likert versus forced-Q format. Since previous 11 factor analyses of the C A Q in its original forced format have yielded varied results, I thought it would be interesting to compare the factor structure of the C A Q items in their original format to the structure of the same items when they are given in Likert format. Any differences that are found could be attributed to the two different response formats. Because the C A Q was derived theoretically by clinicians, a study of this kind is as much an examination of the psychometric robustness of the C A Q as it is a study of response format. For this reason I thought it would be useful to compare the structure of Likert and forced-Q data using instruments that were derived factorially and which have been well-validated. 1 chose therefore to examine the structure of the NEO-Five Factor Inventory (Costa & McCrae, 1989); McCrae & Costa, 1987) and the Interpersonal Adjective Scales (Wiggins, 1995). The NEO-F ive Factor Inventory is a measure of the domains of the Five-Factor Model, and the IAS is a measure of the interpersonal domain of Dominance and Love (please see next section). Method Participants Participants were 130 University of British Columbia students (92 women and 38 men). Mean age of subjects was 19. Subjects were solicited from a subject pool and completed the experiment for course credit. Measures CAQ (Forced-Q). The C A Q was given in its original format. The C A Q contains 100 cards, each containing a personality description. The rater sorts the cards into nine categories, 12 ranging from least salient or characteristic of the subject (Category 1) to most salient or characteristic of the subject (Category 9). A specified number of cards is placed in categories one through nine, as follows: 5 8 1 2 1 6 1 8 1 6 1 2 8 5. The C A Q does not contain inherent scales, but several researchers have developed post hoc scales from it. An example used in this study is from Wink (1992), who developed three Narcissism scales for the C A Q by factor analyzing the highest-rated items in a conceptually-derived Narcissism prototype. These scales are Willfulness, Hypersensitivity, and Autonomy. Wink (1992) reports internal consistencies ranging from .87 to .92 for these scales. CAQ (Likert). A paper-and-pencil version of the C A Q was also used. This measure contains the same items as the original C A Q , but was rated by subjects on a 9-place scale, ranging from "Extremely Uncharacteristic" to "Extremely Characteristic". NEO-Five Factor Inventory (Likert). The NEO-Five Factor Inventory (NEO) is a short version of the NEO-Personality Inventory, which operationalizes the Five-Factor Model (Costa & McCrae. 1989; McCrae & Costa, 1987). The Five-Factor Model represents personality structure in terms of five broad dimensions: Extraversion (E), Agreeableness (A), Neuroticism (N), Conscientiousness (C), and Openness to Experience (O). The Five-Factor model had its origins in peer rating studies of officer effectiveness (Tupes & Christal, 1961), that involved ratings of adjectives which Caltell (1945) had clustered into synonyms based on the original English language trait taxonomy of Allport and Odbert (1936). The Five-Factor Model has since been found to account for the structure underlying a number of personality tests. The N E O - F F I contains 60 items, 12 from each scale. Internal consistencies for the five scales range from .74 to .89 (McCrae and Costa, 1989). In this study the N E O was rated on a 9-place scale to keep the scaling consistent across measures. 13 NEO-FFI (Forced-Q). The N E O was also given in Q-sort format. The 60 items in the original FF1 were placed onto Q-cards and subjects were instructed to place them into nine piles from least to most characteristic, according to the following distribution: 3 5 7 9 12 9 7 5 3. This distribution is prorated from the C A Q distribution. Inter personal Adjective Scales (Likert). The Interpersonal Adjective Scales (IAS) was designed to measure two factors of interpersonal behavior, dominance and love (Wiggins, 1995). The IAS was based on the interpersonal system of Leary (1957) and colleagues (Freedman, Leary, Ossorio, & Coffey, 1951), which is an operationalization of the interpersonal theory of Sullivan (1953). Dominance and Love form the axes of the interpersonal circumplex, a circular representation of interpersonal behavior. The IAS breaks the circumplex into eight octant scales and measures the interpersonal behavior associated with each of these octants. These scales (and their associated abbreviations) are: Assured-Dominant (PA), Gregarious-Extraverted (NO), Arrogant-Calculating (BC), Cold-hearted (DE), Aloof-Introverted (FG), Unassured-Submissive (Ml), Unassuming-Ingenuous (JK), Warm-Agreeable ( L M ) , and Gregarious-Extraverted (NO). Reported internal consistencies for these scales range from .75 to .86 (Wiggins, Trapnell, & Phillips, 1988). The current version of the IAS contains 64 adjectives, eight from each octant. The eight octant scores yield a profile of subjects' interpersonal behavior. The IAS was rated on a 9-place scale. IAS (Forced-Q). The IAS was also given in Q-sort format. The 64 items from the IAS were put onto cards and subjects were asked to place them into nine piles from least to most characteristic, as follows: 2 5 7 1 1 1 4 1 1 7 5 2. This distribution was prorated from the distribution of the C A Q . 14 Design and Administration A within-subjects, fully crossed design (cross-method and cross-instrument) was employed. Participation was spread over two sessions, separated by approximately three weeks, with either Likert format or forced-Q format given in each session. The order in which the two formats were administered was counterbalanced. Half of the subjects took the instruments in Q-sort format in the first session and rating format in the second; the other half took the measures in the opposite order. The order in which the instruments within each session were administered was also counterbalanced. Subjects were randomly assigned in the first session to one of six orders in which to complete the three measures. In the second session subjects were assigned to a group corresponding to their first group. Specifically, the order of the instruments in the second session consisted of the one they took second in the first session, followed by the first and the third. For example, i f in the first session a subject took the C A Q first, followed by the IAS and the N E O , in the second session s/he would take the IAS first, followed by the C A Q and the N E O . Results Comparison of Scores at the Scale, Item, and Profile Levels fable 1 presents means and internal consistencies (alpha values) for the scales of the N E O . IAS, and C A Q . Columns one and two present means and alphas for Likert data, while columns three and four present forced-Q data. A n asterisk next to a scale indicates a significant difference between means (p < .05) for the two formats. Note that for the majority of scales, the 15 means for Likert and forced-Q data are significantly different. Note also that alpha values are consistently higher for Likert data than for forced-Q data. The final column of Table 1 presents the correlation between the Likert and forced-Q versions of each scale. The mean correlation for N E O scales (.79) is substantially higher than that of the IAS and C A Q scales (.65 and .66, respectively). Table 2 presents correlations between Likert and forced-Q item means (i.e., the mean was found for each item in Likert and forced-Q format, and these scores were correlated). Although there appear to be significant differences between Likert and forced-Q data at the level of the scale, correlations at the item level are extremely high. Correlations between Likert and forced-Q profiles of scores were also found. For each subject a correlation was found between respective Likert and forced-Q scores across all items of each measure (e.g., for the IAS a correlation between items 1-64 in Likert format and items 1-64 in forced-Q format was found). The means of these correlations were then found. These means are presented in fable 3. Comparsion of Likert Value Distributions with Q-sort Category Distributions Figure 1 gives an example of the distribution of category values (i.e., one through nine) for the forced-Q distribution. Since this distribution is predetermined to be normal, it is possible to examine how closely Likert data conform to the normal distribution of the Q-sort. Figures 2 through 4 present histograms containing the distribution of Likert values for the N E O , C A Q , and IAS, respectively. Note the similarity in shape between the N E O and C A Q . It is clear that when subjects are allowed to respond freely to items, the distribution of their ratings is very different from the forced distribution of the Q-sort. 16 Factor Analyses 1 compared the factor structures of the three instruments in Likert and forced-Q format at the item level. I extracted five principal components from the N E O (representing the domains of the F F M ) and four from the IAS (representing the two primary axes and the two diagonal axes of the IAS cireumplex) (see Lorr & Strack, 1990). I extracted live components from C A Q data for two reasons. First, I wanted to compare my results with past research which has recovered a live-factor structure from C A Q items (e.g., McCrae et al. 1986). Secondly, an examination of scree plots for C A Q data in both formats showed a break after Five components. In all cases I used Varimax rotation. Table 4 presents the percentage of variance accounted for by each of these factors, as well as the total variance accounted for in each solution. Note that in all cases these factors account for a larger percentage of the variance in Likert format than they do in Q-sort format. Using a Likert format has the apparent advantage of producing components that account for a higher proportion of the total variance. Structure of CAQ. Table 5 and Table 6 present the highest loading items (.33 or greater) and their communalities for the first five factors of the C A Q in Likert rating and forced-Q formats, respectively. A visual inspection of these tables suggests that when five factors are extracted from C A Q item data, the structure is clearer with Likert ratings than in forced-Q format. Factors I through V for Likert data seem to be good approximations of N , E. C , A , & O, respectively. The positive end of the N factor contains items associated with anxiety, vulnerability, irritability, and low self-esteem. The negative end includes such items as calm, consistent, cheerful, and satisfied with self. The E factor contrasts talkativeness, poise, and charm with over-control and aloofness. The C factor contrasts intelligence, high aspiration level. 17 productivity, and ethical behavior with submissiveness, and emotional blandness. The A factor contrasts sympathetic and giving behavior with condescending, deceitful, and critical behavior. The O factor includes unusual thought processes, non-conforming behavior, and fantasy at one pole, and moralistic behavior, consistency, and conservatism at the opposite pole. Although these five factors are only an approximation of the five-factor model, the structure seems to be clear and logical. When five principal components are extracted from the C A Q in forced-Q format, however, the structure seems less clear, as can be seen in Table 6. Factor I seems to include high N items at one pole (anxious, low in self-esteem) and high C items at the opposite pole (productive, dependable). Factor II seems to be a close approximation to A , contrasting giving and warm behavior with guileful and hostile behavior. Factor III seems to tap Extraversion, with one pole containing items suggestive of blandness, submissiveness, and emotional distance, and the other pole suggesting poise and talkativeness, as well as physical attraction items. However, the positive end also contains O and N items. Factor IV appears to be a blend of high E/low N at one pole (gregarious, calm, relaxed), and high N/high O at the opposite pole (irritable, philosophical). It is unclear what Factor V is measuring. Clearly the positive pole is associated with humor, but why this should contrast with a lack of morality and a concern with one's body is unclear. Since factor interpretation is a subjective process, I also examined the factor structure of the C A Q empirically. For both Likert and forced-Q data, I correlated the five C A Q factors with N E O - F F I domain scales. Results are presented in Table 7. I assigned the C A Q factors to respective F F M designations for convenience only, not because I claim to have recovered a pristine five-factor structure. 18 For Likert F F M correlations with the NEO-FF1 there are reasonably strong to large correlations supporting construct validity for N , E, and A . The C and O factors were less supported, as they correlate equally strongly with other factors. For forced-Q data, the C factor is very poorly fitted (.04, compared to .52 in Likert format). Other factors are less dramatically attenuated in forced-Q format. N correlates .60 with NEO-FFI N , whereas in Likert format the correlation is .74. There is also a highly significant correlation of N with C , making it an ambiguous factor. E in forced-Q format correlates .53 with N E O E, whereas the respective correlation for Likert data is .74. The A factor correlates about equally for forced-Q and Likert data (.62 and .67. respectively). The O factor also correlates about equally for the two methods (.45 for forced-Q, .42 for Likert), and has equally high or higher correlations with other factors for both methods. Structure of NEO. Tables 8 and 9 present the highest loading items (.33 or greater) and their communalities for the first live factors of the N E O for rating and Q-sort versions, respectively. Although these factors account for a smaller percentage of the variance in Q-sort than in Likert format (see Table 4), the multivariate structure of the N E O seems to be maintained in both formats. Letters in brackets next to items indicate the scale and direction for "misses" (cases in which a significant loading occured on a factor for an item that belongs to a different scale). Although there are fewer items with significant loadings for the forced-Q version of the N E O , there are fewer misses in this format. Structure of IAS. Tables 10 and 11 present IAS items with the highest loadings (.33 or greater) and their communalities for Likert and forced-Q, respectively. At the item level, the factor structure of the IAS seems to be maintained in the Q-sort version. In fact, as with the 19 N E O . there are fewer "misses" for forced-Q data than for Likert data (for these items, the correct scale is placed in parentheses next to the item). Circumplex Analysis of IAS. Because internal consistencies for scales differed across format, I decided to examine the factor structure of one of the instruments at the scale level. I thought it would be interesting to examine the structure of the IAS in the two formats, because of its unique circumplex structure. I found correlations with the interpersonal factors Dominance and Love for each of the eight IAS scales. Dominance ( D O M ) and Love ( L O V ) factor scores were found for Likert and forced-Q data by entering the IAS scales in standardized form into the following formulas (from Wiggins. 1995): (1) D O M = .3 [PA - HI) + .707 (NO + B C - F G - JK)] (2) L O V = .3 | L M - D E ) + . 7 0 7 ( N O - B C - F G + .IK)|. Once D O M and L O V scores were computed for Likert and forced-Q versions of the IAS. these scores were correlated with the respective IAS scales. This yielded a matrix of circumplex coordinates, presented in Tables 12 and 13 for rating and Q-sort data, respectively. Table 12 shows a clear circumplex patterning for IAS Likert data. As seen in Table 13, IAS forced-Q data have a circular patterning as well, although loadings are generally smaller. The circumplex properties of the IAS seem to be maintained across format. The " Optimally Adjusted Personality" Q-sort Prototype The C A Q prototype for the "optimally adjusted personality" ( O A P ; Block, 1978) was devised by a panel of nine judges, and its reliability is .97. The underlying mechanism associated with the O A P is ego-resiliency. Block and Block (1980) described ego-resiliency as 20 "the dynamic capacity of an individual to modify his/her modal level of ego-control,...as a function of the demand characteristics of the environmental context" (p. 48). At one extreme, ego resiliency is the "resourceful adaptation to changing circumstances and environmental contingencies, analysis of'goodness of fit' between situational demands and behavioral possibility, and flexible invocation of the available repertoire of problem-solving strategies" (p. 48). A t the other extreme, it involves "little adaptive flexibility, an inability to respond to the dynamic requirements of the situation, a tendency to perseverate or to become disorganized when encountering changed circumstances or when under stress, and a difficulty in recouping after traumatic experiences" (p. 48). A n index of similarity to the O A P prototype was found for each subject by computing a Pearson correlation between his/her C A Q Q-sort data and the O A P prototype (Block. 1978). This procedure was repealed for each subject's C A Q Likert data. Finally, a correlation was computed between correlations obtained under the two formats. This correlation was .88. Figure 5 graphically illustrates the close correspondence between O A P scores for Q-sort and rating data. The mean score for rating data was .45, and the mean for Q-sort data was .41. Although a t-test found these means to be reliably different (t = 2.30, p < .05), it is clear that there is a very close correspondence between the two sets of scores. Discussion As noted earlier. Block (1978) has given several reasons why forced Q-sort ratings are preferable to unforced ratings. One of these is that reliabilities are no higher for Likert data than for Q-sort data. In this study, however, internal consistencies were found to be higher for rating 21 data in all cases. It may be the case that Q-sort data are simply not conducive to scoring for scales, because of the limitations in interpretation associated with the method. Since there is no single scale across subjects, it may not make sense to score forced-Q data in this way. This may account for the attenuated internal consistencies. Block (1978) also argues that unforced rating data is susceptible to response biases. While it is clear that the distribution of Likert values for rating measures departs substantially from the symmetric, unimodal distribution of the Q-sort, it is not clear whether these distributions or those for the Q-sort are more "correct" or better. Although the forced-normal distribution may lessen the chances of biases such as social desirability or preference for one part of the scale, it may be the case that a distribution with more items in the middle, neutral category than any other category is not a natural one. The proportion of ratings in Category 5 for these Likert data suggest that it is used relatively less than other categories. It may be that people consider most traits either characteristic or uncharacteristic of themselves, to varying degrees, and that there are few traits they would consider themselves to be "neutral" on. It should be noted that the C A Q was originally meant to be a device for other-ratings, so it may be that scaling preferences are different for self-ratings than they are for those of professional raters. Block (1978) also argues that Q-sort data are more convenient and less unwieldy than rating data. One major reason for this is that forced data allow for comparisons between sorts. This study involved comparison of forced-Q and Likert data to a prototype Q-sort for the optimally adjusted personality. Results indicated that the usefulness of finding similarity indices to a prototype for a particular type of personality is not limited to Q-sort data. Although scores for the two sets of data were on a different scale, they were very highly correlated. It is unclear 22 whether making comparisons between raters using similarity indices is applicable to Likert data. Further research would be necessary to clarify this. Although Block claims that forced-Q data is less unwieldy in terms of application, it is not less unwieldy in terms of administration. The Q-sort takes longer to administer, is more difficult for subjects to understand, and requires much more time to record and transform into an analyzable format. The C A Q has traditionally been used for expert ratings, and presumably judges take less time Q-sorting once they become familiar with the procedure. However, the C A Q is now commonly used for self- and peer- ratings, with subjects who have no experience with the Q-sort method. For self-ratings, much time and hassle would unquestionably be spared by having subjects answer on a Likert scale. Even when expert ratings are used, it is unclear whether a Q-sort format is preferable. The C A Q data obtained from expert ratings in studies at I P A R (The Institute for Personality Assessment and Research) are unquestionably quality data, but it is unclear whether this is due to the Q-sort method itself or to the expertise of the judges. It is possible that data of similar quality could be obtained simply by having judges rate the C A Q items on a Likert scale. Factor Structure Despite successful attempts at recovering a clear five-factor structure with C A Q items (Lanning, 1994; McCrae et al., 1986), in the present study the live-factor solution of the C A Q items in Q-sort format is difficult to interpret. The present study is probably not comparable to that of Lanning (1994), because he used expert ratings, most of which were composites of several judges. 23 The McCrae et al. study (1986) is probably the best study to compare to my findings, since it also involved extraction of live factors from self-ratings. The factor loadings those authors obtained are reported in Table 17 for comparison (note that they report items with loadings of .30 or greater). It is puzzling that they obtained a clearer structure with Q-sort data and one that closely approximates the Five-Factor Model. One possible explanation is the larger N in their study (403. compared to 130 in the present study) and the fact that they used an adult rather than a college student sample. However, a close examination of the factor loadings from their study suggests that the factors, while definitely more interpretable than mine, are not entirely clear or straightforward. McCrae et al. demonstrated comprehensiveness in the C A Q items in terms of their containing the five factors, but the structure they found is somewhat ambiguous. For example, their first factor, Neurotieism, contains at the negative pole the items "verbally fluent'\ "intelligent", "behaves ethically", "personally charming", "socially poised", and "is interesting person", which are not considered part of the Neurotieism factor. In fact their Neurotieism factor is very similar in content to the corresponding factor in my data. The items at the positive pole are clearly associated with Neurotieism, but whether the negative pole can be construed as the opposite of Neurotieism is a matter of interpretation. The Conscientiousness factor that McCrae et al. found contains at the negative pole the items "engages in fantasy", "dissociative tendencies", "enjoys sensuous experiences", "interested in opposite sex", and "eroticizes situations". McCrae et al. claim that the last three of these items add new information to the conceptualization of Conscientiousness. However, whether this finding is replicable or only an artifact of the Q-sort format remains to be seen. Interestingly, the items "interested in opposite sex" and "enjoys sensuous experiences" had high loadings on the 24 negative pole of the third factor in my Q-sort data. However, the positive pole of this factor was very weakly associated with Conscientiousness, i f at all . Overall, the factor structure obtained by McCrae et al. is more easily interpretable than my Q-sort data. M y rating data, however, seem no less interpretable than their Q-sort data, although interpretability varies by factor. M y Neuroticism factor seems clearer, while their Openness factor has more items with significant loadings. The Agreeableness and Extraversion factors seem to be similar in terms of interpretability. The respective Conscientiousness dimensions seem to be tapping two different things: mine seems to be tapping a kind of Ambition, while theirs seems to be contrasting ethical, dependable behavior with erotic interest. Overall, the factor structure of the C A Q seems to be more interpretable in Likert format. This may be due to the problems associated with factor analysis of ipsative measures, or it may be due to a lack of clear structure in the C A Q items. Some of the items are not singular in meaning (e.g., "Is guileful, deceitful, manipulative, opportunistic") and some have multiple interpretations (e.g., "Favors conservative values in a variety of areas"). Also, in contrast with the N E O and IAS, there is an uneven number of items marking each dimension in the C A Q . Because of the clear structures of the N E O and IAS, these instruments were chosen for a comparison of structure in Likert and forced-Q format. The underlying structure of these instruments seemed to be maintained across format. However, the variance in all three measures was substantially attenuated in forced-Q format. Direct ions for Future Research One possible area for future research is an examination of the psychological process involved with the forced-choice nature of the Q-sort. One way to do this would be to get 25 respondents' feedback during or after the Q-sorting procedure. Another way would be to develop a computerized version of the Q-sort which would actually track all of a subject's sorting moves. The aim in either case would be to examine the cognitive processes that are occurring with subjects as they sort. For example, when a subject has more than the allotted number of cards in the extreme-most category, how does s/he decide which cards to move down to less extreme categories? Does the respondent in reality consider the items s/he moves down to be less characteristic of him or her, or is the process of selecting less characteristic items more or less random? Likewise, how does a subject go about deciding which items are more uncharacteristic of him than other items? The answers to such questions would enable a more comprehensive understanding of the Q-sorting process as well as answer important reliability and validity issues raised in this paper. Conclusions The Q-sort method and the rating method set out to measure two different things, so neither should be considered superior to the other. The Q-sort method was designed to measure the salience of a number of traits within an individual, without reference to a population. The Likert method was designed to assess how much of a trait a person has in relation to others. These are both useful pursuits and both are associated with their own problems and biases. The major problem seems to lie in what data are used for, once they are collected. The limitations of ipsative data must be kept in mind, both in terms of interpretation and statistical application. When ipsative measures arc used to compare traits within an individual, this is line. However, when they are used to compare subjects or in multivariate techniques such as factor analysis, problems seem to arise. Specifically, internal consistencies are substantially reduced 26 for scales, and factor structure is greatly affected. The fact is. however, that Q-sort data are often treated as i f they are normative, even though they are an entirely different type of data with different properties. Although the Q-sort is partially, as opposed to purely ipsative, researchers should take caution when interpreting and analyzing Q-sort data. Since Q-sort data are often analyzed in the same manner as normative data, it might make sense in most cases to collect data in Likert format. This of course depends on the research purpose; i f ipsative data are needed, the Q-sort would be preferred over rating data. However, all other things being equal, there seem to be few practical advantages to the Q-sort. The Likert method takes much less time, and the data are easier to handle. In terms of comparability of subjects' scores, thought to be a major advantages of the Q-sort, more research needs to be done to see whether this application can be used with Likert data. At least one such use, comparison of subjects' Q-sorts with a prototype Q-sort, does seem to be applicable to data in Likert format. A final point is that, while it is not appropriate to examine ipsative measures interindividually, because of the lack of a single scale, it is possible to examine normative scores intraindividually. This can be done by ipsatizing scores, for example by standardizing each subject's scores about his/her mean. Because of the limitations involved in ipsative measurement, it may be best to collect data in normative form and ipsatize should the need arise. 27 References Alden, L . E . , Wiggins, J.S.. & Pincus, A . L . (1990). Construction of circumplex scales for the Inventory of Interpersonal Problems. Journal of Personality Asssessment. 55. 521-536. A l l port, G .W. , & Odbert, M.S. (1936). Trait names: A psycholexical study. Psychlogical Mongraphs, 47, (1, Whole No. 211). Bern, D.J., & Funder, D.C. (1978). Predicting more of the people more of the time: Assessing the personality of situations. Psychological Review, 85. 485-501. Block. J. (1978). The Q-sort method in personality assessment and psychiatric research. Palo Alto, C A : Consulting Psychologists Press. Block,.(. , & Block, .1.1-1. (1980). The role of ego-control and ego resiliency in the organization of behavior. In W . A . Collins (Ed.). The Development of Cognition, Affect, and Social Relations. Millsdale, N.J . : L . Erlbaum Associates. Cattell, R .B . (1944). Psychological measurement: Normative, ipsative, interactive. Psychological Review. 5J_, 292-303. Cattell, R .B . (1945). The principal trait clusters for describing personality. Psychological Bulletin. 42. 129-161. Costa. P.T., Jr., & McCrae. R.R. (1989). The N E O Personality Inventory manual. Odessa, F L : Psychological Assessment Resources. Dunlap, W.P., & Cornwell, J . M . (1994). Factor analysis of ipsative measures. Multivariate Behavioral Research. 29(1). 115-126. Freedman, M . B . , Leary, T.F., Ossorio, A . G . , & Coffey, M.S. (1951). The interpersonal dimension of personality. Journal of Personality, 20, 143-161. 28 Goldberg. L .R . (1977, August). Language and personality: Developing a taxonomy of trait-descriptive terms. Invited address to the Division of Evaluation and Measurement at the Annual Meeting of the American Psychological Association, San Francisco. Guilford, J.P. (1952). When not to factor analyze. Psychological Bulletin, 49. 26-37. Haan, N . (1981). Common dimensions of personality development: Early adolescence to middle life. In D . H . Eichorn, J .A. Clausen, N . Haan, M.P . Honzik, & P .H. Mussen (Eds.).. Present and past in middle life. New York: Academic Press. Hicks, L . E . (1970). Some properties of ipsative, normative, and forced-choice normative measures. Psychological Bulletin, 74, 167-184. Hofslee, W . K . B . , de Raad, B . , & Goldberg, L . (1992). Integration of the Big Five and circumplex approaches to trait structure. Journal of Personality and Social Psychology, 63, 146-163. Johnson, C .E . , Wood, R., & Blinkhorn, S.F. (1988). Spuriouser and spuriouser: The use of ipsative personality tests. Journal of Occupational Psychology, 61, 153-162. Katz, M . (1962). Interpreting Kuder Preference Record scores: Ipsative or normative. Vocational Guidance Quarterly, 10, 96-100. Lanning, K . (1994). Dimensionality of observer ratings on the California Adult Q-set. Journal of Personality and Social Psychology, 67( 1), 151 -160. Leary, T. (1957). Interpersonal diagnosis of personality. New York: Ronald Press. Lieberman, M . A . , & Tobin, S.S. (1983). The experience of old age: Stress, coping, and survival. New York: Basic Books. Livson, N . H . . & Nichols, T.F. (1956). Discrimination and reliability in Q-sort personality descriptions. Journal of Abnormal and Social Psychology, 52, 159-165. 29 Loi r , M . (1978). The structure of the California Q-set. Multivariate Behavioral Research, 13. 387-393. Loi r . M . . & Struck, S. (1990). Wiggins" Interpersonal Adjective Scales. Personality and Individual Differences, U, 423-435. McCrae, R.R., & Costa, P.P., .Ir. (1987). Validation of the five-factor model of personality across instruments and observers. Journal of Personality and Social Psychology, 52, 81-90. McCrae, R.R., & Costa, P.T., Jr. (1989). The structure of interpersonal traits: Wiggins' circumplex and the live-factor model. Journal of Personality and Social Psychology. 56. 586-595. McCrae, R.R., Costa. P. I., Jr., & Busch, C M . (1986). Evaluating comprehensiveness in personality systems: The California Q-set and the five-factor model. Journal of Personality, 54. 430-446. Merritt. S.L., & Marshal. J.C. (1984). Reliability and construct validity of ipsative and normative forms of the Learning Style Inventory. Educational and Psychological Measurement. 44, 463-472. Norman, W.T. (1963). Toward an adequate taxonomy of personality attributes. Journal of Abnormal and Social Psychology, 49. 710-721. Russell, J .A. (1994). Is there universal recognition of emotion from facial expression? A review of the cross-cultural studies. Psychological Bulletin. 115( 1), 102-141. Saville, P., & Willson, E. (1991). The reliability and validity of normative and ipsative approaches in the measurement of personality. Journal of Occupational Psychology, 64, 219-238. 30 Stephenson, W. (1953). The study of behavior. Chicago: University of Chicago Press. Sullivan, M.S. (1953). 'The interpersonal theory of psychiatry. New York: Wiley. Thompson, 13., Levitov, .I.E., & Miederhoff, P .A. (1982). Validity of the Rokeach Value Survey. Educational and Psychological Measurement. 42, 899-905. Tupes, E .C. . & Christal, R.E. (1961). Recurrent personality factors based on trait ratings. U S A D A S P Technical Report, No. 61-97. Wiggins, J.S. (1979). A psychological taxonomy of trait-descriptive terms: The interpersonal domain. Journal of Personality and Social Psychology, 37, 395-412. Wiggins, J.S. (1991). Agency and communion as conceptual coordinates for the understanding and measurement of interpersonal behavior. In W . M . Grive & D. Cicchetti (Eds.), Thinking clearly about psychology. Volume 2: Personality and psychopathology (pp. 89-113). Minneapolis: University of Minnesota Press. Wiggins, J.S. (1995). The Interpersonal Adjective Scales manual. Odessa, Florida: Psychological Assessment Resources, Inc. Wiggins, J.S., Phillips, N . , & Trapnell, P. (1989). Circular reasoning about interpersonal behavior: Evidence concerning some untested assumptions underlying diagnostic classification. Journal of Personality and Social Psychology, 56, 296-305. Wiggins, J.S., Steiger, J .H. , & Gaelick, L . Evaluating circumplexity in personality data. Multivariate Behavioral Research, 16, 263-286. Wiggins, J.S., Trapnell, P., & Phillips, N . (1988). Psychometric and geometric characteristics of the revised Interpersonal Adjective Scales (IAS-R). Multivariate Behavioral Research, 23,517-530. 31 Wink, l \ (1991). Sel l - and object-direcledness in adult women. Journal of Personality, 59 (4), 769-791. Wink, P. (1992). Three narcissism scales for the California Q-set. Journal of Personality Assessment, 58( 1), 51 -66. Table 1 Scale Means, Internal Consistencies, and Correlations between Likert and Forced-Q Data Scale Mean a (Likert) Mean a (Forccd-Q) r between (Likert) (Forccd-Q) Likert and Forced-Q N E O : N 4.9 .87 4.9 .85 .81 E* 6.1 .86 5.7 .77 .81 0* 6.0 .72 5.7 .70 .77 A 6.1 .76 6.2 .70 .78 C* 6.2 .83 5.8 .80 .77 IAS: PA* 5.5 .81 4.9 .78 .80 I3C 3.6 .88 3.8 .78 .70 DE* 2.4 .82 2.8 .68 .50 FG* 3.3 .86 3.7 .76 .65 HI 4.4 .83 4.3 .79 .68 .IK* 5.4 .77 4.8 .72 .53 LM* 7.0 .84 5.9 .74 .57 NO* 6.6 .88 5.6 .79 .75 C A Q : W* 4.5 .57 4.0 .52 .55 H* 4.3 .78 4.1 .69 .74 A* 6.6 .68 6.2 .58 .68 Note: For NEO, N = neuroticism; E = extraversion; O = openness; A = agrecableness; C = conscientiousness. For IAS, PA = assured-dominant; BC = arrogant-calcuating; DE = cold-hearted; FG = aloof-introverted; HI = unassured-submissive; JK = unassuming-ingenuous; LM= warm-agreeable; NO = gregarious-extraverted. For CAQ Narcissism, W - willfulness; 11 = hypersensitivity; A = autonomy. A "•*" indicates a significant difference (p <. 05) between the rating and Q-sort scale means. All correlations between Q-sort and rating data are significant at p < .001. Table 2 Correlations between Item Means for Likert and Foreed-0 Measure r C A Q " ~ " . 9 8 IAS .97 N E O .98 Table 3 Mean Correlations between Subject Profiles lor Likert and Forced-Q Measure Mean r C A Q ~ . 6 6 N E O .67 IAS .69 35 Table 4 Percentage of Variance Accounted for by Factors of IAS, N E O . and C A Q Items in Forced-Q and Likert Format Measure Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Total IAS 16 12 6 5 N / A • " 3 9 " (Forced-Q) IAS 21 16 6 5 N / A 48 (Likert) N E O 13 8 7 6 4 38 (Forced-Q) N E O 15 9 7 6 6 43 (Likert) C A Q 11.4 5.7 4.6 3.6 3.5 28.8 (Forced-Q) C A Q 14 8 6 4 4 36 (Likert) Table 5 California Q-Set items defining five factors (in Likert format) Item description Loading h 2 Factor I: 68. Basically anxious .71 .65 78. Feels victimized .67 .55 40. Generally fearful .65 .53 13. Thin-skinned .64 .49 55. Self-defeating .63 .55 45. Brittle ego defenses .63 .51 79. Tends to ruminate .55 .34 34. Irritable .55 .50 9. Uncomfortable with uncertainty .55 .38 19. Seeks reassurance from others .54 .36 72. Concerned with own adequacy .53 .38 82. Has fluctuating moods .52 .33 22. Feels lack of personal meaning .50 .45 87. Overinterprets simple situations .49 .35 47. Guilt-prone .46 .26 30. Gives up under frustration .46 .34 99. Self-dramatizing .41 .44 12. Self-defensive .41 .29 10. Psychosomatic symptoms .39 .20 63. Judges in conventional terms .39 .45 46. Engages in fantasy .39 .29 50. Unpredictable .38 .32 69. Sensitive to demands .38 .21 16. Introspective .38 .29 89. Compares self to others .37 .21 49. Distrustful .36 .42 21. Arouses nurturant feelings .35 .24 42. Delays action .33 .24 92. Socially poised -.34 .58 84. Cheerful -.40 .70 33. Calm, relaxed -.42 .19 75. Has consistent personality -.46 .47 74. Satisfied with self -.50 .52 Factor II: 88. Personally charming .77 .61 4. Talkative .72 .55 57. Is an interesting person .69 .51 92. Socially poised .67 .58 37 81. Is physically attractive .65 .50 84. Cheerful .64 .70 31. Regards self as attractive .64 .46 28. Arouses liking in people .56 .40 58. Enjoys sensuous experiences .55 .40 54. Gregarious .52 .38 52. Behaves assertively .51 .54 18. Initiates humor .51 .32 74. Satisfied with self .48 .52 35. Warm, compassionate .45 .41 80. Interested in opposite sex .43 .27 43. Facially, gcsturally expressive .37 .34 56. Responds to humor .33 .36 93. Sex-role stereotyped behavior .33 .32 59. Concerned with own body .33 .22 25. Tends toward over-control -.34 .36 48. Keeps people at a distance -.50 .35 Factor III: 98. Verbally fluent .58 .44 8. Appears intelligent .58 .44 71. Has high aspiration level .57 .36 26. Productive .55 .42 70. Behaves ethically .54 .38 60. Insight into own motives .52 .33 3. Has a wide range of interests .49 .36 2. Dependable, responsible .47 .29 96. Independent; autonomous .47 .28 51. Values intellectual matters .46 .33 64. Socially perceptive .43 .27 52. Behaves assertively .43 .54 83. Sees to the heart of problems .41 .23 6. Fastidious .39 .31 66. Enjoys esthetic impressions .37 .27 77. Straightforward .36 .25 41. Moralistic .35 .42 22. Feels lack of personal meaning -.34 .45 61. Exploits dependency in people -.34 .46 14. Submissive -.41 .25 97. Emotionally bland -.42 .30 Factor IV: 27. Condescending .70 .52 37. Guileful, deceitful .67 .46 36. Subtly negativistic .62 .43 61. Exploits dependency in people .55 .46 1. Critical, skeptical .51 .49 91. Power oriented .49 .35 76. Projects feelings onto others .45 .28 73. Eroticizes situations .41 .31 63. .fudges in conventional terms .40 .45 34. Irritable .39 .50 25. Tends toward over-control .38 .36 38. Has hostility toward others .37 .27 65. Tries to push limits .36 .28 49. Distrustful .35 .42 12. Self-defensive .34 .29 99. Self-dramatizing .33 .44 56. Responds to humor -.38 .36 1 7. Sympathetic, considerate -.40 .34 5. Behaves in a giving way -.49 .41 Factor V : 39. Has unusual thought processes .60 .38 62. Non-conforming .60 .46 15. Skilled in play, humor .38 .22 46. Engages in fantasy .34 .29 49. Distrustful .33 .42 75. Has consistent personality -.36 .47 41. Moralistic -.42 .42 93. Sex-role stereotyped behavior -.45 .32 7. Has conservative values -.54 .37 Table 6 California Q-Set items defining live factors (in Foreed-Q format) Item description Loading h 2 Factor \: 45. Brittle ego defenses .55 .38 53. Low self-control .52 .32 19. Seeks reassurance from others .50 .33 68. Basically anxious .44 .41 55. Self-defeating .41 .33 46. Engages in fantasy .41 .20 78. Feels victimized .39 .31 34. Irritable .39 .49 63. Judges in conventional terms .38 .19 79. fends to ruminate .37 .25 14. Submissive .36 .47 13. Thin-skinned .36 .29 89. Compares self to others .33 .29 72. Concerned with own adequacy .33 .37 82. Has fluctuating moods .33 .26 22. Feels lack of personal meaning .33 .18 92. Socially poised -.36 .47 8. Appears intelligent -.37 .17 77. Straightforward -.42 .32 51. Values intellectual matters -.42 .38 83. Sees to the heart of problems -.43 .22 96. Independent; autonomous -.45 .32 71. Has high aspiration level -.46 .30 2. Dependable, responsible -.48 .36 98. Verbally fluent -.51 .42 52. Behaves assertively -.52 .49 26. Productive -.52 .30 60. Insight into own motives -.52 .30 75. Has consistent personality -.54 .35 70 . Behaves ethically -.59 .41 Factor II: 28. Arouses liking in people .60 .49 5. Behaves in a giving way .48 .25 84. Cheerful .47 .56 35. Warm, compassionate .42 .21 88. Personally charming .36 .33 54. Gregarious .35 .36 56. Responds to humor .33 .43 40 65. Tries to push limits -.33 .19 44. Evaluates motivation of others -.38 .17 39. Has unusual thought processes -.38 .29 62. Non-conforming -.39 .22 28. Condescending -.39 .21 1. Critical, skeptical -.44 .27 49. Distrustful -.47 .31 38. I las hostility toward others -.53 .32 37. Guileful, deceitful -.53 .38 36 . Subtly negativistic -.54 .38 Factor III: 97. Emotionally bland .57 .42 14. Submissive .51 .47 48. Keeps people at a distance -.47 .35 25. Tends toward over-control .47 .23 7. Has conservative values .44 .31 9. Uncomfortable with uncertainty .36 .23 13. Thin-skinned .33 .29 98. Verbally fluent -.38 .42 80. Interested in opposite sex -.40 .45 52. Behaves assertively -.43 .49 57. Is an interesting person -.44 .39 43. Facially, gcsturally expressive -.47 .25 58. Enjoys sensuous experiences -.48 .32 4. Talkative -.52 .39 92. Socially poised -.55 .47 31. Regards self as attractive -.59 .51 81. Is physically attractive -.66 .52 Factor IV: 67. Self-indulgent .54 .41 84. Cheerful .53 .56 80. Interested in opposite sex .46 .45 54. Gregarious .42 .36 33. Calm, relaxed •41 .36 95. Proffers advice .40 .18 74. Satisfied with self .33 .38 82. Has fluctuating moods -.37 .26 16. Introspective -.42 .20 51. Values intellectual matters -.43 .38 66. Enjoys esthetic impressions -.45 .21 6. Fastidious -.50 .27 34. Irritable -.54 .49 90. Philosophical -.60 .39 Factor V : 17. Sympathetic, considerate .66 .38 18. Initiates humor .66 .49 15. Skilled in play, humor .64 .48 56. Responds to humor .48 .43 50. Unpredictable .44 .25 77. Straightforward .35 .32 89. Compares self to others -.34 .29 93. Sex-role stereotyped behavior -.38 . .24 59. Concerned with own body -.40 .22 41. Moralistic -.47 .27 42 Table 7 Correlations of C A Q Factors in Likert and Forced-Q Format with NEO-FFI Domain Scales C A Q Factor N E O A C Likert N 74** -.28* .01 -.18 -.11 E -.31* .74** .02 .19 .08 0 -.08 -.15 .42** -.41 ** -.41 ** A .05 .01 -.22* -.67** -.18* C -.27* .03 42** -.04 .52** Forced-Q: N .60* .15 -.11 -.16 -.48** E .20* .53** -.26* .06 -.05 0 -.41 ** -.37** -.45** -.02 -.15 A -.10 -.21* -.02 .62** .12 C -.11 -.05 .04 .07 .04 Note: * p < .05, two-tailed. p < .001, two-tailed. 43 Table 8 N E O items defining five factors (in Likert format) Item description Loading h 2 Extraversion: 37. 1 am a cheerful, high-spirited person. .77 .69 1 7. 1 really enjoy talking to people. .71 .54 22. 1 like to be where the action is. .70 .56 2. 1 like to have a lot of people around me. .69 .57 52. I am a very active person. .61 .66 32. 1 often feel as i f Em bursting with energy. .61 .41 7. I laugh easily. .57 .43 34. Most people I know like me. .53 .40 47. M y life is fast-paced. .51 .43 16. 1 rarely feel lonely or blue. (N-) .44 .39 49. I generally try to be thoughtful and considerate. (A+) .35 .56 12. I don't consider myself especially "lighthearted". -.36 .34 57. I would rather go my own way than be a leader of -.44 .25 others. 27. I usually prefer to do things alone. -.45 .30 42. I am not a cheerful optimist. -.54 .55 Neurotieism: 26. Sometimes I feel completely worthless. .75 .65 4 1 . Too often, when things go wrong, I get discouraged .75 .62 and feel like giving up. 51.1 often feel helpless and want someone else to solve .68 .53 my problems. 1 1. When I'm under a great deal of stress, sometimes I .65 .47 feel like I'm going to pieces. 21.1 often feel tense and jittery. .63 .56 6. 1 often feel inferior to others. .58 .40 56. At times I have been so ashamed I just wanted to .57 .42 hide. 29. I believe that most people will take advantage of you .45 .45 i f you let them. 42. 1 am not a cheerful optimist. (E-) .35 .55 36. I often get angry at the way people treat me. .35 .30 49. I generally try to be thoughtful and considerate. (A+) .34 .56 52. 1 am a very active person. (E+) -.39 .66 16. 1 rarely feel lonely or blue. -.42 .39 1. 1 am not a worrier. -.50 .41 46. 1 am seldom sad or depressed. -.55 .39 31.1 rarely feel fearful or anxious. -.60 .45 .44 Conscientiousness: 1 0. I'm pretty good at pacing myself so as to get things .73 .54 done on time. 35. 1 work hard to accomplish my goals. .69 .58 50. I am a productive person who always gets the job .68 .60 done. 20. I try to perform all the tasks assigned to me .63 .48 conscientiously. 25. I have a clear set of goals and work toward them in .61 .46 an orderly fashion. 5. I keep my belongings clean and neat. .55 .40 60 I strive for excellence in everything I do. .50 .36 40. When I make a commitment I can always be counted .46 .37 on to follow through. 3. I don't like to waste my time day-dreaming. (O-) .36 .21 52. I am a very active person. (E+) .35 .66 8. Once I find the right way to do something, I stick to it. .34 .14 (O-) 47. My life is fast-paced. (E+) .34 .43 15.1 am not a very methodical person. -.43 .23 45. Sometimes I'm not as dependable or reliable as I -.45 .41 should be. 55. I never seem to be able to get organized. -.58 .38 30. I waste a lot of time before settling down to work. • -.64 .46 Agreeableness: 49. I generally try to be thoughtful and considerate. .47 .56 4. I try to be courteous to everyone 1 meet. .42 .28 40. When I make a commitment, I can always be counted .38 .37 on to follow through. (C+) 42. I am not a cheerful optimist. (E-) -.34 .55 45. Sometimes I'm not as dependable or reliable as I should be. (C- -.35 .41 ) 36. 1 often get angry at the way people treat me. (N+) -.36 .30 54. If I don't like peole, I let them know it. -.37 .17 12. I don't consider myself especially "lighthearted". -.38 .34 (E-) 29. I believe that most people will take advantage of you if you let -.46 .45 them. 44. I'm hard-hearted and tough-minded in my attitudes. -.48 .27 9. I often get into arguments with my family and co-workers. -.50 .33 39. Some people think of me as cold and calculating. -.64 .50 59. If necessary, 1 am willing to manipulate people. -.67 .58 24. I tend to be cynical and skeptical of others' intentions. -.68 .49 45 14. Some people think I'm selfish and egotistical. -.68 .51 Openness to experience: 43 Sometimes when I am reading poetry or looking at a work of art, I feel a chill or wave of excitement. .78 .63 13 I am intrigued by the patterns I find in art and nature. .76 .59 53. 1 have a lot of intellectual curiosity. .70 .59 58. 1 often enjoy playing with theories or abstract ideas. .66 .50 48 I have little interest in speculating on the nature of the universe or the human condition. -.55 .33 18 I believe letting students hear controversial speakers can only confuse and mislead them. -.48 .36 33 1 seldom notice the moods or feelings that different environments produce. -.51 .28 23. Poetry has little or no effect on me. -.76 .62 Note: Letters in brackets indicate the actual scale items are from for cases in which there is a discrepancy. N = neuroticism; E = extraversion; O = openness; A = agreeableness; C = conscientiousness. Table 9 NEO items defining live factors (in Forced-Q Format) Item description Loading Ir Neurotieism: 26. Sometimes I feel completely worthless. .68 .55 41. Too often, when things go wrong, I get discouraged and feel .67 .58 like giving up. 51.1 often feel helpless and want someone else to solve my .60 .53 problems for me. 11. When I'm under a great deal of stress, sometimes I feel like I'm .59 .41 going to pieces. 6. I often feel inferior to others .56 .40 21.1 often feel tense and jittery. .54 .35 36. I often get angry at the way people treat me. .43 .32 42. I am not a cheerful optimist. (E-) .41 .39 56. At times I have been so ashamed I just wanted to hide. .41 .29 37. I am a cheerful, high-spirited person. (E+) -.51 .52 46. I am seldom sad or depressed. -.58 .38 16. I rarely feel lonely or blue. -.67 .50 31.1 rarely feel fearful or anxious. -.69 .49 1. 1 am not a worrier. -.74 .59 Conscientiousness: 35. 1 work hard to accomplish my goals. .71 .53 50. I am a productive person who always gets the job .65 .47 done. 25. I have a clear set of goals and work toward them in .61 .47 an orderly fashion. 10. I'm pretty good at pacing myself so as to get things .59 .37 done on time. 20. I try to perform all the tasks assigned to me .57 .41 conscientiously. 60. I strive for excellence in everything I do. .50 .32 5. I keep my belongings clean and neat. .48 .31 40. When I make a commitment, I can always be counted .41 .24 on to follow through. 8. Once I find the right way to do something, I stick to it. .41 .18 (O-) 38.1 believe we should look to our religious authorities .36 .28 for decisions on moral issues. (O-) 37. I am a cheerful, high-spirited person. (E+) .33 .52 45. Sometimes I'm not as dependable or reliable as I -.45 .34 47 should be. 15. I am not a very methodical person. -.48 .28 55. I never seem to be able to get organized. -.54 .32 30. 1 waste a lot of time before settling down to work. -.56 .39 Agreeableness: 49. I generally try to be thoughtful and considerate. .56 .34 4. I try to be courteous to everyone 1 meet. .56 .38 1 2. 1 don't consider myself especially "lighthearted". -.36 .19 ( E - ) 54. If 1 don't like people, 1 let them know it. -.38 .24 24. I tend to be cynical and skeptical of others" -.44 .29 intentions. 9. I often gel into arguments with my family and co- -.46 .36 workers. 14. Some people think I'm selfish and egotistical. -.54 .35 39. Some people think of me as cold and calculating. -.54 .39 59. If necessary, I am will ing to manipulate people. -.56 .40 Extraversion: 22. I like to be where the action is. 2. I like to have a lot of people around me. 17. I really enjoy talking to people. 52. I am a very active person. 32. I often feel as i f I am bursting with energy. 47. M y life is fast-paced. 34. Most people I know like me. 37. I am a cheerful, high-spirited person. 7. I laugh easily. 42. I am not a cheerful optimist. 57 . 1 would rather go my own way than be a leader of others. 27. I usually prefer to do things alone. Openness to experience: 43. Sometimes when 1 am reading poetry or looking at a .73 .59 work of art, 1 feel a chill or wave of excitement. 13.1 am intrigued by the patterns I find in art and nature. .70 .53 53. 1 have a lot of intellectual curiosity. .58 .42 58. 1 often enjoy playing with theories or abstract ideas. .55 .34 33.1 seldom notice the moods or feelings that different -.51 .32 environments produce. 48. I have little interest in speculating on the nature of -.61 .45 the universe or the human condition. 23. Poetry has little or no effect on me. -.76 .58 .69 .56 .57 .42 .56 .44 .53 .41 .49 .34 .48 .37 .42 .30 .37 .52 .36 .27 -.35 .39 -.38 .18 -.53 .32 48 Note: Letters in brackets indicate the actual scale items are from for cases in which there is a discrepancy. N = neuroticism; E = extraversion; O = openness; A = agreeableness; C = conscientiousness. Table 10 IAS items defining four factors (in Likert format) Item description Loading h2 Factor l(NO-FG): 16. Cheerful .79 .67 43. Perky .72 .56 44. Friendly .68 .61 29. .lovial .67 .54 47. Outgoing .65 .56 20. Enthusiastic .64 .52 40. Extraverted .61 .50 21. Self-assured (PA) .60 .48 46. Self-confident (PA) .58 .50 58. Neighborly .50 .45 41. Gentle-hearted (LM) .38 .53 1 1. Coldhearted (DE) -.42 .36 49. Bashful (HI) -.38 .48 36. Timid (HI) -.45 .51 37. Unbold (HI) -.45 .55 45. Unneighborly -.48 .37 25. Meek (HI) -.50 .55 23. Unsparkling -.50 .36 9. Shy (HI) -.58 .53 60. Distant -.66 .46 1. Introverted -.66 .51 13. Dissocial -.73 .54 18. Antisocial -.74 .57 56. Uneheery -.76 .67 52. Unsociable -.79 .66 Factor 11 (JK-BC): 10. Uncunning .72 .53 51. Uncrafty .69 .48 27. Unsly .67 .53 39. Unwily .65 .55 5. Uncalculating .59 .40 31. Boastless .36 .21 6. Accommodating (LM) .34 .28 19. Iron-hearted (DE) -.33 .38 32. Domineering (PA) -.34 .49 59. Warmthless (DE) -.37 .29 12. Ruthless (DE) -.40 .33 61. Cocky -.43 .49 50 48. Boastful -.46 .40 64. Tricky -.64 .45 57. Sly -.70 .60 55. Calculating -.72 .65 54. Wily -.75 .66 30. Crafty -.76 .63 24. Cunning -.77 .66 Factor III ( L M - D E ) : 14. Tender-hearted .71 .56 7. Kind .67 .47 8. Charitable .67 .48 34. Tender .63 .42 62. Sympathetic .62 .47 41. Gentle-hearted .6.1 .53 15. Soft-hearted .58 .38 58. Neighborly (NO) .43 .45 6. Accommodating .40 .28 44. Friendly (NO) .38 .61 42. Persistent (PA) .33 .39 59. Warmthless -.36 .29 1 1. Coldhearted -.36 .36 12. Ruthless -.38 .33 19. lion-hearted -.40 .38 61. Cocky (BC) -.48 .49 53. Hard-hearted -.50 .35 22. Cruel -.61 .51 26. Uncharitable -.63 .41 35. Unsympathetic -.65 .48 Factor IV (HI-PA): 63. Forceless .69 .53 28. Unaggressive .66 .44 33. Unargumentative (.IK) .58 .41 37. Unbold .55 .55 25. Meek .55 .55 2. Undemanding (.IK) .47 .27 36. Timid .44 .51 49. Bashful .40 .48 9. Shy .38 .53 40. Extraverted -.34 .50 21. Self-assured -.35 .48 47. Outgoing (NO) -.36 .56 46. Self-confident -.40 .50 50. Firm -.42 .27 42. Persistent -.50 .39 38. Forceful -.56 .49 32. Domineering -.57 .49 3. Assertive -.71 .52 17. Dominant -.73 .61 Note: Letters in brackets indicate the actual scale items are from for cases in which there is a discrepancy. PA = Assured-Dominant; B C = Arrogant-Calculating; D E = Coldhearted; F G = Aloof-Introverted; HI = Unassured-Submissive; .IK = Unassuming-Ingenous; L M = Warm-Agreeabl N O = Gregarious-Extraverted. Table 11 IAS items defining four factors (in Forced-0 formal) Item description Loading h2 Factor 1 (NO-FG): 16. Cheerful .73 .57 40. Extraverted .70 .61 47. Outgoing .62 .60 29. Jovial .60 .41 20. Enthusiastic .58 .39 43. Perky .57 .40 44. Friendly .55 .35 45. Unneighborly -.33 .19 53. Hard-hearted (DE) -.33 .37 25. Meek (HI) -.35 .45 23. Unsparkling -.36 .25 60. Distant -.37 .17 9. Shy (HI) -.51 .56 13. Dissocial -.58 .39 18. Antisocial -.60 .42 56. Uneheery -.63 .47 1. Introverted -.68 .49 52. Unsociable -.73 .55 Factor II (HI-PA): 28. Unaggressive .70 .49 4. Unauthoritative .62 .50 33. Unargumentalive (.IK) .60 .38 25. Meek .54 .45 63. Forceless .54 .31 37. Unbold .51 .34 9. Shy .40 .56 36. Timid .38 .45 2. Undemanding (.IK) .37 .20 42. Persistent -.35 .15 24. Cunning (BC) -.33 .57 47. Outgoing (NO) -.42 .60 38. Forceful -.50 .36 50. Firm -.56 .38 21. Self-assured -.59 .40 32. Domineering -.60 .43 1 7. Dominant -.64 .43 46. Self-confident -.67 .55 3. Assertive -.74 .60 53 Factor III ( J K - B C ) : 27. Unsly .75 .58 51. Uncrafty .65 .47 39. Unwily .65 .44 10. Uncunning .70 .54 5. Uncalculating .57 .35 31. Boastless .40 .25 8. Charitable ( L M ) .35 .29 48. Boastful -.41 .23 55. Calculating -.58 .49 64. Tricky -.59 .44 54. Wily -.62 .40 24. Cunning -.63 .57 30. Crafty -.64 .48 57. Sly -.65 .45 Factor IV ( L M - D E ) : 14. Tender-hearted .63 .50 34. Tender .58 .37 15. Soft-hearted .55 .39 62. Sympathetic .51 .28 7. Kind .49 .31 41. Gentle-hearted .47 .35 58. Neighborly (NO) .47 .31 36. Timid (HI) .45 .45 8. Charitable .39 .29 49. Bashful (HI) .36 .28 9. Shy (HI) .33 .56 26. Uncharitable -.38 .31 12. Ruthless -.40 .23 53. Hard-hearted -.41 .37 22. Cruel -.42 .27 19. Iron-hearted -.43 .32 59. Warmthless -.45 .26 11. Coldhearted -.57 .39 35. Unsympathetic -.63 .45 Note: Letters in brackets indicate the actual scale items are from for cases in which there is a discrepancy. PA = Assured-Dominant; B C = Arrogant-Calculating; D E = Coldhearted; F G = Aloof-Introverted; HI = Unassured-Submissive; JK - Unassuming-lngenous; L M = Warm-Agreeable; N O = Gregarious-Extraverted. Table 12 Factor Scores for IAS Likert Scales on the Interpersonal Factors Love and Dominance Scale Love Dominance I ' A ~ ~ 7 ) 0 4 9 .8445 B C -.7479 .4787 D E -.8213 .0433 F G -.6462 -.6169 HI -.1651 -.8248 .IK .6468 -.4954 L M .7052 -.0864 N O .6658 .6179 Table 13 Factor Scores for IAS Forced-Q Scales on the Interpersonal Factors Love and Dominance Scale Love Dominance P A -.0733 .8155 B C -.6393 .4526 D E -.7196 .1903 F G -.5608 -.6302 HI .0844 -.8704 .IK .5842 -.5921 L M .7891 -.1865 N O .5888 .5956 Table 14 Factor Loadings of C A Q Items on Five Factors (from McCrae, Costa, and Busch, 1986) Item Description Loading Neuroticism: 13. Thin-skinned .58 68. Basically anxious .58 34. Irritable .53 47. Guilt-prone .52 19. Seeks reassurance .51 12. Self-defensive .48 82. Fluctuating moods .46 72. Concerned with adequacy .46 45. Brittle ego defenses .44 40. Vulnerable to threat .43 58. Self-defeating .42 78. Feels victimized, cheated .42 10. Psychosomatic symptoms .36 50. Unpredictable in behaviors .35 89. Compares self to others .35 23. Extrapunitive .34 30. Gives up under frustration .33 38. Has hostility .31 98. Verbally fluent -.30 8. Intelligent -.31 70. Behaves ethically -.32 60. Has insight into own motives -.32 88. Personally charming -.35 92. Socially poised -.36 83. Sees to heart of problems -.37 84. Cheerful -.42 57. Is interesting person -.44 24. Prides self on objectivity -.44 75. Clear-cut personality -.48 74. Satisfied with self -.51 33. Calm, relaxed -.56 Extraversion: .56 4. Talkative .45 54. Gregarious .45 92. Socially poised .43 52. Behaves assertively .43 15. Skilled in play and humor .41 20. Rapid tempo .41 57 57. Is interesting person .41 99. Self-dramatizing .40 43. Facially, gesturally expressive .37 98. Verbally fluent .36 29. Turned to for advice .33 18. Initiates humor .33 35. Warm, compassionate .32 28. Arouses liking .32 95. Proffers advice .30 79. Has preoccupying thoughts -.33 14. Submissive -.34 45. Brittle ego defenses -.38 30. Gives up under frustration -.38 25. Over control of impulses -.51 48. Avoids close relationships -.51 97. Emotionally bland -.53 Openness: 51. Values intellectual matters .45 62. Rebellious, nonconforming .41 39. Unusual thought processes .38 16. Introspective .36 8. Intelligent .34 66. Aesthetically reactive .34 3. Wide range of interests .32 46. Engages in fantasy, daydreams .30 26. Productive -.30 93. Sex role stereotyped behavior -.33 41. Moralistic -.34 9. Uncomfortable with complexities -.35 63. Judges in conventional terms -.51 7. Favors conservative values -.55 Agreeableness: 17. Sympathetic, considerate .56 35. Warm, compassionate .52 28. Arouses liking .44 5. Behaves in a giving way .37 84. Cheerful .34 56. Responds to humor .33 21. Arouses nurturant feelings .32 88. Personally charming .30 49. Basically distrustful -.30 38. Has hostility -.32 62. Rebellious, nonconforming -.39 48. Avoids close relationships -.41 58 91. Power oriented -.43 94. Expresses hostility directly -.45 65. Tries to push limits -.45 27. Shows condescending behavior -.47 52. Behaves assertively -.48 1. Critical, skeptical -.48 Conscientiousness: 70. Behaves ethically .43 2. Dependable, responsible .42 8. Intelligent .39 26. Productive .36 71. Has high aspiration level .35 83. Sees to heart of problems .33 51. Values intellectual matters .33 46. Engages in fantasy, daydreams -.32 86. Dissociative tendencies -.33 58. Enjoys sensuous experiences -.37 67. Sell-indulgent -.41 53. Unable to delay gratification -.41 80. Interested in opposite sex -.44 73. Eroticizes situations -.53 ONE THREE FIVE SEVEN NINE TWO FOUR SIX EIGHT Figure 1. Example Distribution of Q-sort Ratings 1400-! 60 Figure 2. Distribution of Likert Ratings for the NEO Figure 3. Distribution of Likert Ratings for the CAQ Figure 4. Distribution of Likert Ratings for the IAS Figure 5. Similarity Scores to the "Optimally Adjusted Personality" for Likert and Forced-Q CAQ Data 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0099046/manifest

Comment

Related Items