«RATIONAL AND INTUITIVE STYLES: COMMENSURABILITY ACROSS RESPONDENTS' CHARACTERISTICS' By: W. M. Taggart, Enzo Valenzi, Lori Zalka, and Kevin B. Lowe ...»
RATIONAL AND INTUITIVE STYLES: COMMENSURABILITY ACROSS RESPONDENTS'
By: W. M. Taggart, Enzo Valenzi, Lori Zalka, and Kevin B. Lowe
Taggart, B., Valenzi, E., Zalka, L., & Lowe, K. B. (1997) Rational and intuitive styles: Commensurability
across respondent characteristics. Psychological Reports, 80(1), 23-33.
Made available courtesy of Ammons Scientific: http://www.ammonsscientific.com/
*** Note: Figures may be missing from this format of the document
This study was designed to examine differences in responses to the six rational/intuitive scales of the Personal Style Inventory in relation to gender, age, ethnic group, birth country, occupation, and industry. Data were collected from 495 participants in training programs in Australia, England, New Zealand, and the United States.
Multivariate analysis of variance indicated no differences among groups on the six scales which then are not sensitive to the characteristics so separate norming scores are not indicated. Lack of differences between sexes contrasts with the finding that women score more intuitive than men on other style assessment tools. Findings are not, however, consistent. And, since characteristics other than gender may show similar disparate results, further study of rational-intuitive commensurability is needed.
It is well known that people differ in their relative preferences for rational and intuitive ways of dealing with situations. Measurement of personal style frequently is used in personnel development for individual counseling and in group training programs to assess participants' preferences on relevant attributes. Scoring norms for assessment tools may differ from one group to another. Ideally, the meaning and interpretation of scores should be the same, that is commensurable, for the various groups of participants in an organization or a program. Then the same raw-score conversion table applies for all participants regardless of their gender, age, ethnic group, birth country, industry, or occupation. This commensurability of scale scores across participants' characteristics facilitates the completion and interpretation of personal results in real-time settings.
Recently, assessment of the intuitive dimension of cognitive style has emerged as a significant theme in development of management. A compelling case for the world-wide implications of this emerging emphasis appears in the report of a recent study of intuition in management (Parikh, Neubauer, & Lank, 1994). In a nine nation survey of senior and top managers, they found that 54% were guided equally by intuition and rationality while 8% said they were guided more by intuition (p. 63). The Personal Style Inventory (Taggart & Hausladen,
1993) provides an effective and efficient assessment of preferences for rational/intuitive styles.
The early assessment tool designed to help individuals in management understand their preferences for rational/intuitive styles was published as the Human Information Processing Survey (Taggart & Torrance, 1984). The three scales provide scores for an individual's left-dominant (rational), right-dominant (intuitive), and integrated (rational/intuitive) behavior preferences. The Personal Style Inventory evolved as a secondgeneration measure from extensive field experience with the survey. For the inventory an idea pool of 500 behavior and assessment terms was generated and sorted into six scales (Taggart & Valenzi, 1990). This pool was used to express behaviors and preferences for an original set of 90 items grouped into 15 paired items for each of the three paired scales. Factor analysis was used to select the final five items for each scale.
The inventory includes 30 behavioral items (five for each scale) to assess six information-processing modes classified as either a rational or an intuitive style. Responses are based on a 6-point rating scale of frequency anchored by 1 (never) and 6 (always). Adverbial anchors for the numerical scale were selected based on a magnitude-estimation scale procedure (Bass, Cascio, & O'Conner, 1974). The scales are paired on three "how
do you" themes with contrasting rational and intuitive styles relative to each theme:
How do you prepare for the future?
Rational planning by developing proposals or Intuitive vision by generating scenarios How do you solve problems?
Rational analysis as a specialist or Intuitive insight as a generalist How do you approach work?
Rational control procedure-oriented or Intuitive sharing people-centered Using this framework, the inventory presents a more detailed assessment of individuals' rational/intuitive preferences than the earlier study and thereby increases their self-awareness.
The present study was designed to examine differences in responses to the inventory that are related to typical respondents' characteristics of gender, age, ethnic group, birth country, occupation, and industry. The broad question we sought to answer was whether the mean scores vary by characteristic. While no a priori hypotheses were developed, the outcomes have clear implications for the use of the inventory. Significant evidence of noncommensurability would require group norms by characteristic for individual interpretation and understanding of the scores. On the other hand, the presence of commensurability across characteristics would facilitate the interpretation and use of scores for the purposes described above.
METHODSubjects The 495 subjects were participants in 40 supervisory and management training programs using the Personal Style Inventory in Australia (n=102), England (n=124), New Zealand (n=109), and the United States (n=160).
The programs were public offerings with self-selected participation. The average workshop size was 13 participants, with a range from 5 to 30. The six characteristics of respondents were gender, age, ethnic group, birth country, industry, and occupation. In the Appendix (pp. 31-33) are the values of the frequencies, means, and standard deviations for each scale on each characteristic.
Analysis Respondents' scores to the 30-item survey were summed into the six scales identified by Taggart and Valenzi (1990). For each characteristic such as age, gender, etc., a separate multivariate analysis of variance was performed on scores on the six scales. The analysis was used because intercorrelations among the scales were substantial. Interactions between respondents' characteristics were not posited because there were no a priori reasons to expect them. Nevertheless, a series of ad hoc multivariate analyses of variance included multiple independent variables to test for interactions. The number of significant interactions was fewer than the number expected by chance.
Statistically significant results from the multivariate analysis of variance were explored further in two ways to assess more precisely the nature of group differences. First, a multiple descriptive discriminant analysis was performed for each characteristic in which the scales were the discriminant or outcome variables and the characteristics were the grouping variables. Stepwise discriminant analyses using the Wilks method were performed in SPSS Release 4.1 (SPSS, 1990). This method causes all variables to be entered in the order of their contribution to group separation. The stepwise analysis was not done as a variable selection procedure but to provide F-to-remove statistics to assess importance of variables for the discriminant functions as recommended by Huberty and Morris (1989; Huberty, 1994). The higher the F-to-remove value the more the variable contributes to group separation (Huberty, 1994; Huberty & Morris, 1989). These were obtained after Step 6 when all outcome variables of the inventory are in the equation.
Second multivariate contrasts between groups were examined to assess whether any group pairs were significantly different. The discriminant analysis was not cross-validated with an independent sample because our purpose was to examine the commensurability of scale scores across groups rather than to predict group membership for classification. Using the entire sample provided a more accurate and powerful test of group differences.
RESULTSCoefficients alpha and intercorrelations of scores on the six scales of the inventory were compared with those of the original study (Taggart & Valenzi, 1990). The lowest alpha in the original study was.53 for the Control scale. Data from the present study measured coefficient alpha at.35 for Control and from.64 to.73 for the remaining five scales. Because the reliability of the Control scale was lower in this study, we decided to retain the variable in the analyses but only for exploratory purposes. The question of the unacceptable coefficient alpha for Control is being addressed in a separate study in which improvement of the reliabilities of all six scales is undertaken. Also, the correlations among the scales were similar in size and sign to those of the original study. The correlations, descriptive statistics, and coefficients aloha for the study variables are reported in Table 1. As shown in Table 2, the multivariate analysis of variance gave statistically significant effects (p.05) for four of the six characteristics, ethnic group, birth country, industry, and occupation.
The multivariate analysis of variance for the significant characteristics of respondents was examined with discriminant analysis to describe the group differences on the six scales. Differentiating group membership based on this inventory would suggest that separate norms should be developed as a function of the respondents' characteristics. The variance explained and canonical correlations for the discriminant functions are provided in Table 3 Box's M test for the equality of group covariance matrices indicated that the assumption was not violated.
For occupation, one discriminant function was statistically significant at the.05 level and another at.01.
Functions falling short of statistical significance (p =.05) were identified for ethnic group, birth country, and industry. For these five functions, ethnic group had the highest (.33) and birth country the lowest (.18) canonical correlation. These low canonical correlations suggest that these inventory variables account for little betweengroup variance.
For descriptive purposes, univariate F tests were used to identify statistically significant differences between group means. Group means on the scales of the Personal Style Inventory were different (p=.05) in five of 42 comparisons. These are Analysis (p=.006) and Vision (p=.03) for ethnic group, Analysis (p =.001) for industry, and Analysis (p =.02) and Vision (p=.03) for occupation. The multivariate analyses of variance for the significant and for the grouping variables which fell just short of significance were followed up with a multivariate test of the pairwise contrasts between all pairs of groups as recommended by Huberty (1994). None of the p values were significant at a.=.05 after a Bonferroni correction to protect against Type I error across all tests.
Summary of Analyses The results of the multivariate analysis of variance and discriminant analyses for statistical significance of differences among inventory scales by the six characteristics of respondents suggest the following conclusion.
From a global perspective, Table 2 shows that three of the six characteristics fell short of significance, ethnic group, birth country, and industry, while occupation was significant (p =.01). The rank order of the Fs-toremove in Table 3 suggests Analysis is the most likely contributor to significance for ethnic group, Insight for birth country, Analysis for industry, and Vision for occupation. The coefficients of the first discriminant function for each characteristic confirm the F-to-remove rank for ethnic group and country but suggest Analysis instead of Insight for birth country and Analysis instead of Vision for occupation.
The data from Tables 2 and 3 suggest that only one or two at the most of the six inventory variables would account for the possible significant differences for the four characteristics. Within each characteristic, differences between all pairs of groups were tested with an F test. After the Bonferroni correction is applied to the p values for the individual F statistics, none remain significant. Only one reaches p =.06 for the public administration versus manufacturing pair contrast for the industry characteristic. Due to the absence of statistical significance after the omnibus data from Table 2 are analyzed in more detail, we conclude that norming the Personal Style Inventory as a function of the respondents' characteristics studied here is unnecessary.