The degree of correspondence between various inventories has been investigated by directly comparing an individual's interests as measured by various instruments. Several such studies yield apparently contradictory results. Triggs (1943, 1944) and Wittenborn, Triggs, and Feder (1943) reported similarities in the Strong Vocational Interest Blank (SVIB) and the Kuder Preference Record by considering overall profiles and correlational configurations. In more studies, King, Norrell, and Powers (1963), O'Shea and Harrington (1971), Wilson and Kaiser (1968), and Zytowski (1968) reported low correlations between corresponding scales of the Strong and various Kuder forms, and Kuder (1969) discussed methodological differences in construction of the instrument which could lead to low correlations. In still other comparisons, Rose and Elton (1970) and Wall, Osipow, and Ashby (1967) used intermediate variables (such as vocational choice or major field) to demonstrate the relationship between two inventories-this time the Strong, and Holland's Vocational Preference Inventory (VPI).
Although the direct approaches have produced different conclusions, many writers have proposed categories of vocational interest which indirectly suggest correspondence between inventories. Super and Crites (1962) compared the factors of interests reported in several factor analytic studies, including those of Turstone (1931), Strong (1943), and Guilford, Christensen, Bond, and Sutton (1954). The following factors were commonly found: scientific, social, language-literary, mechanical, business, and artistic factors. Cottle (1950) and Harrington (1970) also reported correspondence of many factors of the Strong and different Kuder forms.
Many classifications of occupations have used categories which resemble the factors listed. Roe (1956) proposed eight groups (Technology, Outdoor, Science, General Cultural, Arts and Entertainment, Service, Business Contact, and Organization) in her two-way classification of occupations. Super (1957) named similar groups: Technical or Material, Scientific, Musical or Artistic, Literary, Humanistic or Social Welfare, Business Contact, and Business Detail. Further similarities are found in the families of scales of the Strong (Darley and Hagenah, 1955; Strong, 1943). Holland (1966) described six major occupational and personality types (Realistic, Intellectual, Artistic, Social, Enterprising, and Conventional) and used the types to classify occupations (Holland, 1966).
These classes of occupations and interests have been used primarily as discrete and independent categories. However, recent considerations of the relationships among the categories suggest a basis for the comparison of interest inventories which may help to unify previous results. Roe (Roe, 1956; Roe and Klos, 1969) suggested that her interest categories were related to each other in a circular ordering in which classes adjacent in the circle were most closely related while those most widely separated were the least related. Several studies supported the proposed continuum of interests (Jones, 1965; Roe, Hubbard, Hutchinson, and Bateman, 1966). Holland examined the relationships of the six scales of the Vocational Preference Inventory and empirically demonstrated a circular arrangement of the scales (Holland, Whitney, Cole, and Richards, 1969) from Realistic to Intellectual to Artistic to Social to Enterprising to Conventional and back to Realistic. Cole, Whitney, and Holland (1971) used a statistical analysis of spatial configuration to relate occupational groups to the six Holland scales. The results also demonstrated that the circle of Holland scales can be considered as a continuum of interests with each location in the circle representing a different mix of the six Holland interests.
The similarities of the many proposed categories of occupational interests and the recent demonstrations of relations among the categories suggest that a simple configuration or structure (such as that demonstrated by Holland) may underlie the vocational interests assessed by various inventories. If such a configuration does exist, the inventories may be sampling interests from a common underlying interest domain, and the configuration thus may provide a useful perspective for research and interpretation.
The purpose of this paper is to determine whether the circular configuration of interests proposed and demonstrated by Roe and Holland is common to other interest inventories. Since Holland's six interest categories represent one of the simplest skeletal versions of the interest categorizations, it is the basis of comparison. The following inventories were studied: the Strong Vocational Interest Blank (SVIB), the Kuder Occupational Interest Survey (Kuder OIS), Holland's Vocational Preference Inventory (VPI), the Minnesota Vocational Interest Inventory (MVII), and a new instrument, the American College Testing Program's Vocational Interest Profile (ACT VIP). If the circular arrangement is common to the internal structure of several instruments, the arrangement would provide a basis for comparisons of the instruments that might resolve the conflicting results of previous research and provide a basis for counselor interpretation.
Method
The Analysis of Spatial Configuration
Cole and Cole (1970) described an analysis of spatial configuration which used the method of principal components to examine variable relationships. This analysis, when performed on Holland's Inventory, yielded a circular configuration of the six Holland scales (Cole et al., 1971). The same analysis was applied to the Strong, the Kuder OIS, the Minnesota Inventory, and the ACT instrument to determine if the scales on each of these inventories are also arranged in a corresponding circle.
The analysis of spatial configuration uses the correlation matrix as the source of information about relationships among the variables. The geometric configuration of vectors (or points) representing variables (in this case scales of an inventory) is then reduced by fitting a smaller space to the vector points. If the variables lie predominately in a smaller space, then the relationships among the variables may be more easily understood. Often, in practice, the smaller space is two dimensional (a plane) and the variables may be plotted and visually represented as in the analysis of Holland's instrument.
Two results of the configural analysis are of special importance to our considerations. The first is the degree to which the relations among the scales on an interest inventory may be represented by a plane and the second is the particular arrangement of the scales in that plane. The percentage of the trace for the first two components of the second stage analysis can be interpreted as indicating the percentage of the variance of the scale points accounted for by the fit of the plane. The projection of the interest scales onto the plane provides the particular scale configuration.
Comparisons of Configurations
In order to compare the planar configuration of one instrument with that of another, differences in types and numbers of scales used in the instruments must be accommodated. Holland's simple system of six categories, which is similar in many ways to other categorizations and has the circular configuration under consideration, is used as the basis of comparison. Scales of the Strong and the Kuder were classified into Holland's system on the basis of a classification of 400 occupations reported in Holland, Viernstein, Kuo, Karweit, and Blum (1970). (Scales without a corresponding classification in that report were left unclassified.) Then the positions in the configural analysis for all the scales in each Holland class were averaged, resulting in mean planar locations representing the six Holland categories. These mean planar points could then be examined for indication of the circular configuration in the Holland ordering.
Data
Separate correlation matrices of the scales in each of the interest inventories were submitted to the analysis described. The inter-correlations of 50 SVIB occupational scales for 301 men were reported in Campbell; those for the 22 SVIB basic scales for 647 men were found in Campbell, Borgen, Eastes, Johansson, and Peterson. The Kuder Occupational Interest Survey Manual gave the inter-correlations of the 23 core scales for 276 men. Inter-correlations of the nine homogeneous keys of the Minnesota Inventory for 400 men were obtained from Clark. The inter-correlations of the 12 scales of the ACT instrument for 311 male students are given in the Handbook for the ACT Career Planning Profile. The analysis of a correlation matrix for men on Holland's inventory was adapted from Cole et al. (1971) in order to facilitate comparisons with the other inventories.
The Planar Configurations
The scale points were projected onto the best-fitting planar surface for each of the inventories. Since the location of the principal component axes is of no special interest in this configural analysis, the planar configurations have been oriented in the same general way for each inventory to simplify visual comparisons.
Discussion
The purpose of this study was to examine the degree to which Holland's circular configuration of interests is common to various inventories. The results of the analysis of spatial configuration provide answers to two basic questions. The first question is: To what extent can the relationships among the scales be described in a plane (in two dimensions)? The answer, provided by the goodness of fit of the planes, is that the interrelations among the scales can be described to a large and important degree by two dimensions. The second question is: To what extent are the particular planar configurations of scales common from one instrument to another? Investigation of the configurations demonstrates that Holland's (and Roe's) configuration is, to a large degree, common to all the instruments investigated. This second result would have been an important finding even if present only on less dominant dimensions. That the circular configuration always appeared on the first two dimensions of variation among the scale points and that these two dimensions accounted for a large part of variation provide even stronger evidence for the practical importance of the circular ordering of interests.
Relation to Previous Research
A related purpose of the study was to use the results about the configuration of the scales to help reconcile the conflicting results of previous research. The demonstration of a common, underlying, circular ordering of vocational interest scales for several instruments provides a new basis for comparison of different inventories which may make the results of previous studies more understandable.
In one group of previous studies low correlations on same-named or similarly named scales were reported and it was concluded that the inventories measured very different things. However, a new look at the results of these studies using the common configuration of interest provides evidence of great similarities between instruments.
In a second group of studies, a general correspondence between inventories was reported. In those studies the use of patterns of profiles or patterns of correlations as the bases for comparison meant that many aspects of the circular configuration were being implicitly used. Thus the results provided evidence for similarities between the inventories. Similarly the factor analytic studies, using correlational configurations, usually demonstrated in similarities.
Although the comparisons of the present study were indirect in the sense that correlation matrices for different groups were used, reconsideration of the results of studies making direct comparisons provides confirmation of the direct relation of one instrument's circular configuration to that of another. That is, those scales of Interest Inventory A that fall into one of Holland's categories, as a whole, tend to correlate most highly with the scales of Inventory B that fall into the same category. Furthermore, the scales in an adjacent category (in the circular ordering) tend to provide the next highest correlations.
Implications for Counseling
Since much research on comparing interest inventories has been motivated by counselors who were frustrated and confused by contradictory results, consideration of the relevance of these findings to counseling practice is appropriate. The results of this study have implications for the interpretation of scores from a single inventory as well as for comparison of scores from more than one inventory. By using the circular structure, a counselor may determine patterns of interest by observing scores on groups of scales rather than considering scales individually. For example, a person's high scores on scales such as Farmer, Carpenter, Forester, and Engineer together indicate Realistic interests which may be applicable to many other occupations. These results also suggest that correspondence of the scores on two or more interest inventories can best be determined by considering patterns of scales rather than individual scales. If a person scores high on several Intellectual scales on one instrument, he will likely score high on Intellectual scales on another. However, he may be highest on Chemist on one instrument and on Physicist on the other.
The discrepancies that occur in the comparison of scores on similarly named scales of different instruments deserve further consideration. This study suggests some reasons for the discrepancies which may be helpful to counselors. Most occupational titles, such as engineer, physician, lawyer, and salesman, do not represent narrowly defined occupations. Instead many different types of activity occur under the same occupational title. The activities of an engineer may range from building bridges to teaching electrical engineering. While the activities have common aspects and reflect many common interests, in the Holland scheme the bridge builder represents almost typical Realistic interests while the electrical engineering professor combines Intellectual interests with the Realistic ones. Similarly, the physician may be a pediatrician or a surgeon. Both share the Intellectual interests common to all physicians but the pediatrician likely reflects strong Social interests as well as Intellectual interests and the surgeon, Realistic interests. If we consider the circular structure of interests we can better understand why a scale based on one physician group (composed of pediatricians) would correlate more highly with a teacher scale, say, than with another physician scale (based on surgeons). Thus, the circular configuration of interests provides a possible explanation for differences in performance on scales having the same name but possibly based on groups reflecting somewhat different vocational interests.
