The intercalation matrix of loader styles, based on survey data, was subjected o the Stuntman-Lingoes smallest space analysis, which transforms correlation coefficients to distances in an Euclidean space. The hypothesized partial order relations among the different leader styles were accurately reflected in the analysis. This lends support to the potential of facet analysis in studying leadership styles. This paper examines facet analysis (CB. Stuntman, 1959, 1966) as a method for developing and testing hypotheses about structural relations of leadership styles.

Heeler (1971) suggested that leaders’ decision styles can be described along an influence-power continuum. Bass and Valence (1974) followed this approach and defined five leadership styles: direction, negotiation, consultation, participation, and delegation. These styles can be characterized according to the relative degree of authoritativeness that is inherent in the specific leader style. Bass and Valence (1974) postulated that leadership style is a part of a broader management-subordinate system, and that two major variables in this system are the power and the information distribution between the leader and his subordinates.

Hence, the probability that a certain leader style will occur partly depends on the power and information differences between the leader and his braininess. To test the Bass and Valence (1974) postulates on leader styles, the method of facet analysis (CB. Stuntman, 1959, 1966) was employed. According to this approach, a Cartesian space is constructed by the facets that characterize the phenomenon under investigate author wishes to thank B. M. Bass and K. R.

Gabriel for their comments on an earlier draft of this paper. Requests for reprints should be sent to Cur Shapiro, The Jerusalem School of Business, Hebrew University, Jerusalem, Israel. Zion. The elements of the facets define profiles in the Cartesian space, which correspond to the variables employed in the empirical study. The relationships among the variables are predicted according to the facet design and are tested subsequently using an appropriate multivariate procedure.

To express the Bass and Valence (1974) formulations in a facet design, the following facets are proposed: A: Leader behavior a i : Authoritative a->: Democratic B : Locus of power b i : Boss has the power BC: Subordinate has the power C : Locus of information CIA: Boss has the information c->: Subordinate has the information The leader styles constitute the different profiles that are constructed by these facets. For example, abaci is a style that is described by an authoritative boss ho has both the power and the information to make decisions.

Such a style was defined by Bass and Valence (1974) and Heeler (1971) as a directive style. Based on the same facets the following leadership styles can be defined: Profile abaci abaci basics a 2 b 2 CIA Jacob Leader style Directive Negotiation Consultative Participative Delegating A common order exists within the different facets. Namely, in each facet the order shows a progression from a strong to a weak form 136 LEADERSHIP STYLES 137 TABLE 1 of authoritative behavior of the leader toward his subordinates. This can be represented by AI > , bi > be, and CIA > co.

However, no assumptions are made as to a possible rank order among the facets. The entire universe of profiles (i. E. , leader styles) can therefore be represented as a partially ordered set (see Figure 1). The relations among the different leader styles as predicted by the facet design are presented in Figure 1. The objective of this study was to test these predicted relations by using an appropriate multidimensional scaling technique. METHOD Subjects There were 407 subordinates (327 males and 80 females) who responded to the survey questionnaire.

Of these, 214 worked as subordinates n university, industrial, or city libraries. Another 106 were subordinates whose managers were studying part time for masters degree in business administration through an executive development program, and 87 worked for a large industrial organization. INTERRELATIONS pop LEADER STYLES Leader style 1. 2. 3. 4. 5. Direction Negotiation Consultation Participation Delegation 1 1. 00 2 . 25 1. 00 3 . 31 . 02 1 . 00 4 . 28 . 14. 84 5 . 13 . 10 . 66 . 68 Data Analysis Stuntman-Lingoes smallest space analysis (CB.

Augustan, 1968; Lingoes, 1973) was applied to the leader styles’ interrelations matrix. This analysis provides a meteoric representation of the different variables (i. E. , leader styles) as points in an Euclidean space. The distances between pairs of points in the space correspond to the correlations of the variables. Hence, two points are closer if the correlation between the corresponding variables is higher. Let RI I be the correlation between two leader styles i and j, and let d i j be the distance between the corresponding points which represent i and j in the Euclidean space.

Also, let our be the correlation between leader styles k and 1, and let dig be the corresponding distance between k and 1 in the Euclidean space. Smallest space analysis attempts to find the space with the minimum number of dimensions in which the rank order of relations will be preserved, namely: whenever RI j > r k i, then du < dti. The goodness of fit of a solution in a given space is measured by a coefficient of alienation ('l ? r 3 ) , where r is a rank-order correlation between the variables' intercorrelations and their corresponding geometric distances.

The smaller the coefficient of alienation, the better the fit; zero represents a perfect fit. Generally, the greater the number of dimensions, the smaller the coefficient of alienation. Thus, smallest space analysis works in a sequential manner to provide the minimal number of dimensions needed to obtain a geometric representation with a good fit (i. E. , a coefficient of alienation smaller than . 15; see Stuntman, 1968). The Survey Instrument A survey questionnaire was employed. The questionnaire construction was based on separate principal-components analyses with Bavaria rotations run on each section.

Multi-item scales were constructed within each section to measure the factors accounting for the most variance. The scales used dealt with the following variables: five leader styles, five organizational, three work group, vie task, four attitude, two effectiveness, and six within-system variables. The questionnaire involved 131 items, 36 of which relate to leader behavior. (For a detailed description of the questionnaire, see Bass & Valence, 1974). FIGURE 1 . Partial order of the leader styles in the definitional Cartesian space.

RESULTS The intercalation matrix of the different leader styles is presented in Table 1. The smallest-space-analysis solution for the matrix is given in Figure 2. The configuration of points was obtained after seven iterations. For a two- dimensional-space solution the coefficient of alienation was . 0009 and the responding Karakul’s (1964) stress measure was . 0007. Therefore, Figure 2 gives an almost perfect fit of the plotted distances to the actual interrelations. Hence, the structural 138 CUR SHAPIRO ALTER 2. Smallest space analysis of the five leadership styles. Elation’s among the different leader styles can be examined simply by looking at the configuration plotted in Figure 2. DISCUSSION There is a remarkable correspondence between the predicted structure of the leader styles relations and the empirical complaisance-analysis results. This supports the theoretical framework suggested by Bass and Hyaline (1974). The correspondence is not perfectly isomorphic, since the negotiation variable is relatively distant from the direction-delegation axis (see Figure 2). This reflects the low average correlation (. 13) between the negotiation style and the other styles.

The negotiation style, characterized by a combination of high authoritativeness (AI) and power that is located with the subordinate (b 2 appeared to be a unique and distinct leader style. The proximity between consultation and participation on the other hand suggests that the two are very similar. It is likely that if data was available for al the eight profiles, the scatter diagram would fit the predicted configuration outlined in Figure 1 even better. In examining the results, facet design and smallest space analysis should be clearly distinguished one from the other.

Smallest space analysis is a statistical technique that was used by Taylor (1971) to test the four-factor theory of leadership. As a multidimensional technique, smallest space analysis differs from factor analysis in some respects. Practically, factor analysis is usually used in a post hoc manner, while smallest space analysis is frequently employed to test hypothesized structures. Also, smallest space analysis emphasizes the structural relations among the different variables rather than the factor solutions, which are the goals of most studies that employ factor analysis. For discussions of these issues, see Stuntman, 1966; Schlesinger & Stuntman, 1969; Kim & Levin, 1972; Farley & Cohen, 1974. ) In the present analysis, smallest space analysis provided an instrument for a preliminary test of the facet definition of leadership styles, which is the focus of the current stud}’. The three facets provided 2 3 ? 8 profiles out of which data, were available on S profiles only. The facet design can be extended by adding more facets and intensified by increasing the number of elements within each facet.

The present facet definition did not attempt to account for all the variations in leader behavior, but to demonstrate the potential of facet analysis in studying leadership styles. The results indicate that such a previously unexamined approach may be fruitful in further investigations of leader behavior.