Search

Research explored methods of providing a coherent framework for the Instructional Systems Development (ISD) model which would promote better system evaluation, internal communications and student testing. Lists of critical tasks for two Air Force specialties were compiled and multidimensional scaling and hierarchical cluster analysis employed to derive their respective job dimensions. A series of advanced measurement techniques was constructed on the basis of the job dimensions and then administered to groups of students. Results showed that: 1) multidimensional scaling analysis provided a useful framework for ISD models; 2) several of the advanced testing techniques were more useful predictors of student success than were the available multiple choice tests; and 3) student and instructor attitudes were positive. It was concluded that the scaling method should be used to order job analytic data and to provide coherency in ISD applications. In addition, the following new testing procedures were recommended as alternatives or adjuncts to the multiple choice format: sequential testing, figural systems, confidence testing, technical words, absurdity recognition, partial knowledge, and signal detection. (Author/PB)

A method for assessing the structure of organization was developed on the basis of the ordinal separation, or proximity, between pairs ot items in recall protocols over a series of trials. The proximity measure is based on the assumption, common to all indices of organization, that items which are coded together in subjective memory units will consistently tend to be recalled contiguously in output. Methods of hierarchical cluster analysis are then employed to determine the structure of organization implied by the proximities between items. An experiment was designed to test the sensitivity of the method to differences in organizational structure. Existing data from several studies of part-whole transfer by Ornstein were reanalyzed to assess the explanatory power of the method of proximity analysis. It was concluded that the method of proximity analysis can be useful in attempts to elucidate the relationship between organization and memory. (Author/DG)

Two hundred twenty-seven organizational units drawn from a variety of industries were cluster-analyzed on the basis of their similarities across 18 behavioral and structural dimensions of effectiveness. Using a multivariate subgrouping procedure, eight homogeneous clusters of units were found, varying in size from 8-65 units, and each characterized by a distinct behavioral profile. Within-cluster differences were described in terms of the 18 dimensions, overall unit effectiveness, and in terms of other "demographic" variables that were external to the clustering process. These variables included unit function, work setting dispersion, growth, technology employed, and skill level of the unit members. Significant relationships were found between the behavioral styles of the clusters on the one hand and each of the demographic variables and the overall criterion of effectiveness on the other. The rationale and benefits of clustering organizational units into groups were discussed, and further research ideas were proposed. (Author)

The design of a proposed occupational clustering system for the Comprehensive Career Education Model (CCEM) was to meet three general criteria: (1) encompass most existing jobs, (2) translate into an entire K-12 curriculum, and (3) show clear advantages over other systems. However, upon examination of existing clustering systems for relevance and possible adaptation, no one system met all the criteria. Therefore, a new clustering system was devised by synthesizing useful features of existing systems. The proposed system has two crucial dimensions--one stressing functions and contents of occupations, the other emphasizing status or levels of occupations. The proposed clustering system was planned to fulfill three instructional functions: (1) inform students about the world of work, (2) assist students in choosing a suitable career, and (3) provide models to shape instructional objectives and learning experiences. (Author/JS)

Information regarding the various publishers' priorities and data to support the supposition that the test publishers do differ in their priorities as reflected in their test characteristics, are presented. Evaluation ratings available for 1,600 published scales appropriate for grades 1, 3, 5, and 6 were from 39 publishers and distributers. A cluster analysis was performed by submitting the mean values for each publisher on each of the four MEAN categories: Measurement Validity; Examinee Appropriateness; Administrative Usability; and Normal Technical Excellence. In addition to the observed differences, three clusters of test publishers appeared. Cluster 1 publishers can be characterized as producing tests highly usable administratively and fairly good in terms of examinee appropriateness; Cluster II is like the first cluster, but its publishers produce tests with greater relevance and validity; and Cluster III, with only one publisher, has emphasized the examinee appropriateness of its tests to the neglect of other test qualities. (Author/LS)

This report describes the results of an experiment assessing the appropriateness of Delphi procedures for formulating group value judgments. Two groups of subjects--upperclass and graduate students from UCLA--were asked to generate and rate value categories relating to higher education and the quality of life. The initial lists (300 and 250 items respectively) were aggregated by the experimental teams to 45 and 48 items respectively. The subjects then rated all possible pairs of these items with respect to their similarity. The average similarity ratings were analyzed by a clustering routine. Subjects from both groups made estimates of the relative contributions of each of the educational categories to each of the quality of life categories. The primary data analyses concerned the importance rating. Three aspects were examined: the quality of the distributions of the responses, the correlation between ratings by different groups and different rating techniques, and the amount of change and degree of convergence upon iteration with feedback. The experiment furnished support for the conclusion that Delphi procedures are appropriate for processing value material as well as factual material. The substantive data concerning higher education and individual life also indicated the possibility that current notions of the role of the university are somewhat loosely tied to the basic interests of the students. A 13-item bibliography is included. (Author/MJM)

This report, one of 21 case studies, describes the history of a recent educational product. The Cluster Concept Program, developed at the University of Maryland, is directed toward the preparation of individuals for entrance into a spectrum of occupations. Three clusters of occupations are included: (1) Construction, (2) Electro-Mechanical Installation and Repair, and (3) Metal Forming and Fabrication. The program is targeted for boys and girls in Grades 11 and 12 who want some occupational preparation while remaining in the main stream of the educational program. It aims at giving students entry level skills for a range of related occupations. Although the 4-year study ended in 1969, the original developers have conducted several related activities since that time. They include: (1) assistance to schools, (2) further evaluation studies, (3) placement and performance research, and (4) development of additional clusters. (GEB)

Thirty-three Parent-Child Centers (PCCs) are grouped into five clusters according to thematic orientation of content or intent for parents' and children's programs in order to provide models which may be viewed as strata in selecting Centers as sampling points for Phase II of the national PCC evaluation--an in-depth study of project impact on low-income member families at the sample Centers. The rationale is presented for the choice of the particular Center to be studied within each cluster. Seven chapters make up the report, the first describing the techniques used to obtain the clusters. Chapters II through VI describe the five clusters identified, the implications for impact which might be expected as a function of each model, and the characteristics of the Center chosen to represent each particular model in Phase II. Chapter VII summarizes characteristics of PCCs selected as sampling points for the Phase II investigation of impact. Appendices are included on focus and leadership items, variables entered in preliminary clustering attempts, and additional data descriptive of the centers within each cluster. (LH)

A technique for testing the hypothesis that a hierarchical sequence of partitions constructed by the max method could have been obtained solely on the basis of "noise" is discussed. The evaluation procedure imvolves comparing a rank-order goodness-of-fit measure to the tabled percentiles obtained from an approximate cumulative permutation distribution of the measure. One of the rank orderings of the object pairs is derived from the original similarity values between the objects to be partitioned; the second rank ordering of the object pairs is obtained from the partition hierarchy itself. (Author/KM)

Various reported measures of clustering in free recall are reviewed under categories of algebraic versus probabilistic approaches. Shortcomings in these measures are outlined and a new multi-dimensional measure is advanced which overcomes many of the deficiencies noted. (Author)

A technique for detecting and studying item (or test) x group interactions independent of differences in level or dispersion of the groups is described. It involves construction of a scatter plot with two groups represented, one on each axis. Each point in the scatter plot represents the coordinates of a measure of a characteristic for one group plotted against a measure of the same characteristic for the other group; and the set of N points in the scatter plot represented the variables under study. The shape of the ellipse represents the degree to which the two groups in question share similar profiles. One method of analysis is to measure the departure of each point from the major axis of the ellipse and to study specific items that are most aberrant. Another is to calculate the variance associated with the item x interaction. Although the technique is not intended as a measure of item or test bias, it is useful in diagnosing cultural differences and comparing different types of groups. (Author/DJ)

The purpose of this paper is to suggest a technique of cluster analysis which is similar in aim to the Interactive Intercolumnar Correlation Analysis (IICA), though different in detail. Two methods are proposed for extracting a single bipolar factor (a "contrast compenent") directly from the initial similarities matrix. The advantages of this general approach are that: (a) It helps avoid certain misclassification problems inherent in IICA; (b) It is related in a straightforward way to conventional techniques of multidimensional scaling and therefore allows a unified treatment of dimensional and "typal" structures; and (c) It provides an interesting solution to the problem of relations among linear contrasts based on different subsets of the stimuli. (Author/DJ)

This report was prepared by the Careers and Curricula Program, which is a career development and guidance program sponsored by The Center for Social Organization of Schools. As one of five Center programs which develop scientific knowledge of the schools' effects upon students and utilize this information to improve educational practices, this report from the Careers and Curricula Program examines the correspondence of various vocational interest inventories with the Holland Occupational Classification. Specifically, Holland's categories are compared with occupational groupings that result from the empirical application of McQuitty's Iterative, Intercolumnar Correlational Analysis to the occupational scales of the Strong Vocational Interest Black (SVIB), the Minnesota Vocational Interest Inventory (MVII), and the Kuder DD Occupational Interest Survey (OIS) Core Scales for Men and for Women. Results indicate that internally consistent occupational clusters do exist, and these usually agree with Holland's classification. The hierarchical structure of the clusters follows Holland's hexagonal ordering of categories, additionally attesting to the validity of Holland's theoretical construct. (Author/AG)

The amount of life stress, as measured by the Schedule of Recent Experience (SRE), has been shown to be related to the onset of illness. This instrument was originally developed with a civilian population, and it became apparent that some questions were inappropriate when it was to be applied to a military population. Furthermore, it was believed that identifying significant dimensions of life stress might provide an indication of how the questionnaire could be revised to gain better results with military personnel. A cluster analysis was performed on the SRE responses made by members of the crews from three Navy cruisers. Four clusters of life events were generated. The item content was described as follows: (1) Personal and Social changes, (2) changes related to Work, (3) Marital changes, and (4) changes related to Disciplinary actions. When the responses to the SRE, given by men aboard an aircraft carrier and a battleship were clustered, esentially the same four clusters evolved for both rated (petty officers) and unrated men. The only notable exception was the lack of a Disciplinary cluster for the rated men. The results showed that the clusters of SRE items were highly stable over very different Navy populations. Thus, they can serve as a guide in determining which items may be utilized or deleted from the questionnaire. Also, it is believed that development of these clusters may lead to an improved method of scoring the SRE items. (Author)

A very general model for the computation of independent cluster solutions in factor analysis is presented. The model is discussed as being either orthogonal or oblique. Furthermore, it is demonstrated that for every orthogonal independent cluster solution there is an oblique analog. Using three illustrative examples, certain generalities are made with respect to the class of independent cluster solutions which are generated by the orthomax criterion. A procedure is presented for comparing the simple structure of an orthogonal independent cluster solution with the simple structure of its oblique analog. (Author)

In this methodological paper two indices are developed: a complexity index and an interpretation index. The complexity index is a positive number indicating on the average how many factors are used to explain each variable in a factor solution. The interpretation index will be positive ranging from zero to unity; unity representing a perfect independent cluster solution and zero representing the poorest factor solution in terms of complexity. Through empirical application to the classic 24 psychological variables it is demonstrated that the indices may be computed by hand and are easily interpreted providing a basis for comparing different factor solutions. (Author)

Using the grade equivalent composite scores on the Iowa Tests of Basic Skills of Iowa fourth grade public school pupils who took the tests in January 1970, a study was made to determine the relative precision with which an estimate could be made of the individual percentile norms from different types of cluster sample designs. Five scores ranging from the 14th to the 93rd percentiles were selected, and the proportions below these five scores became the proportions to be estimated. The variances of the estimates of these five proportions were computed for over 20 different sample designs; results from seven sample designs are presented. Using the error variances that were computed for each of the seven sample designs, the ratio of the error variance based on a cluster sample to the error variance based on a simple random sample of pupils was determined. (DB)

The effects of college environments on student achievement on three GRE area achievement tests, when the initial ability of students was controlled were studied. A wide variety of environmental measures for 95 colleges were included. The method computed within-school regression lines, then colleges were clustered on regression lines, using a taxonomic technique, and multiple group discriminant functions using college descriptive variables were computed. It was found that in all three achievement areas one group or cluster of colleges could be identified as being more effective than the remaining groups of colleges. In the humanities areas the most effective group was characterized by a higher proportion of faculty with doctorates, greater selectivity, and a larger budget. In social sciences and the natural sciences, emphasis on subject matter area as measured by the proportion of majors in the appropriate fields of study was the only variable which demonstrated some discrimination between the most and least effective groups. (Author)

This study attempts to determine whether the Piers-Harris Self-Concept Scale is appropriate for use with varying population types, and whether the dimensions identified by it are sufficiently consistent across samples to allow some clarification of the nature of self-concept in children. (AE)

A factor rotation scheme is presented which has a property that is interesting for examining factor-structure hypotheses. Given several clusters of tests the orthogonal bounds scheme produces factors which contain these clusters in some optimal sense. (Author)

This study was concerned with various schemes for reducing the number of variables in a multivariate analysis. Two sets of illustrative data were used; the numbers of criterion groups were 3 and 5. The proportion of correct classifications was employed as an index of discriminatory power of each subset of variables selected. Of the four procedures using indices that order the variables with respect to contribution to discrimination, the (forward) stepwise procedure yielded the best results. Of the two schemes involving dimensional analysis, that which uses correlations of scores on variables with high maximum likelihood factor loadings against discriminant scores appeared more attractive. (Author)

Rationally defined outcome variables, for use in evaluating youth-work training programs, were incorporated in questionnaires administered to present and former Neighborhood Youth Corp enrollees and analyzed to determine their suitability as criterion measures. Using a factor analytic technique, empirically defined clusters were obtained for immediately available outcomes and longer-term outcomes. The most logical groups of criteria were found for those former enrollees who had full-time employment experience. Relatively clear patterns of job-oriented capability and success were dominant as were two separate factors bearing on personal adjustment to the job and to the community. Some descriptive highlights concerning the vocational behaviors of former trainees are presented and future research needs for better definition and understanding of program objectives are discussed. For the six rating instruments used in the study, see TM 000 866-871. (Author/AG)

DISISS (Design of Information Systems in the Social Sciences) is a research project financed by OSTI, which began in January 1971. The objective of the project is to carry out research necessary for the effective design of information systems in the social sciences. The aim of this part of the DISISS project is the application of statistical techniques to citation data in order to group journal titles in the social sciences. Various statistical techniques exist, some with a fairly long history, for grouping items according to observable attributes. Details of these techniques and the selection of one for use with DISISS data are discussed. This report coveres the use of cluster techniques in bibliography, techniques of clustering, an analysis of the pilot study data, progress with data collection and conversion, and work that is required for the future. (Other reports in the DISISS series are ED 060876, 072815, 072816 and LI 004 401 through 004 403.) (Author/SJ)

The study attempted to demonstrate significant relationships between systematically derived occupational descriptors and selected measures of worker characteristics in the affective domain. The predictor variables were derived from the Occupation Analysis Inventory (OAI) and consisted of work-dimension scores describing occupations in terms of work activities and conditions, and attribute-requirement estimates characterizing occupation in terms of various human requirements. The worker variables were based on responses of job incumbents to selected interest, need, and satisfaction questionnaires. The results of various statistical analyses offered consistent support for the validity of the OAI descriptors. It was concluded that the OAI work-dimension scores and attribute-requirement estimates for occupations possess demonstrated relevance to the interests, needs, and satisfactions of job incumbents and that these occupational variables should prove useful for a number of research and development purposes. (Author/MS)

Ergometrics, the application of psychometric principles and procedures to the study of human work, draws from theories and principles of human behavior as well as from established procedures in psychological measurement and job analysis. The project described employs ergometrics to describe, compare, and group occupations for educational purposes. An Occupation Analysis Inventory (OAI) was generated containing 622 work elements (descriptions of activities and conditions). Ratings on the work elements were obtained by relevance to 103 defined human attributes (resulting in attribute-requirement weights). The OAI work elements were then correlated and subjected to factor analysis. Five test batteries and inventories measuring various abilities, interests, needs and satisfaction were used to discover relationships between these variables and measurable behavioral potentials. Clusters of occupations were found to be discriminable and significant cross-validated multiple correlations between OAI factor scores and aptitude test scores were obtained. Other findings also indicate that a meaningful and potentially useful set of occupational descriptors has been developed and validated. (MS)

Most responses to educational and psychological test items may be represented in binary form. However, such dichotomously scored items present special problems when an analysis of correlational interrelationships among the items is attempted. Two general methods of analyzing binary data are proposed by Horst to partial out the effects of differences in item difficulties: (1) a least square simplex data matrix solution, and (2) a least square simplex covariance matrix solution. Of these, the first was selected for study using (1) a regression approach, (2) a raw data approach, and (3) the computational algorithm for the raw data matrix approach. The results indicate that Horst's modification clearly induces an effect that contaminates the common factor structure of the variables. Further, the findings also indicate that image, alpha, and principal components analysis of correlation matrices obtained from binary data matrices are all satisfactory methods of analysis without the modification. This may be an important finding since it tends to confirm earlier empirical findings concerning the varying difficulties of binary items. (CK)

Dissimilarity Linkage Analysis (DLA) is an extremely simple procedure for developing a typology from empirical attributes that permits the clustering of entities. First the procedure develops a taxonomy of types from empirical attributes possessed by entities in the sample. Second, the procedure assigns entities to one, and only one, type in the taxonomy. This two-step procedure clearly contrasts with many existing clustering techniques that are concerned only with the second step of this two-stage procedure. To develop a taxonomy of attribute types, the method searches for attributes that go together. A statistical test of association is first used to identify all pairs of attributes whose empirical values are significantly associated. Attribute pairs are then linked together to form serpentine clusters, each of which represents an attribute type. The attributes defining each type are not similar. In fact, the method specifically avoids using any criterion of similarity when developing the types. Each entity is then assigned to the type it most closely resembles. An entity may unequivocally fit a type. Or, if an entry does not possess all of the characteristics of a type, it is assigned to the type with which its attribute values best match. Discrete clusters of entities, based on their attribute types, are thus formed. In short, this method moves from types defined by dissimilar attributes, to clusters of similar entities in each type of the taxonomy. (Author)

THIS STUDY EXAMINED THE JOB SATISFACTIONS OF REHABILITATION COUNSELORS USING A SAMPLE OF 143 COUNSELORS FROM SIX "DVR" (STATE-FEDERAL GENERAL VOCATIONAL REHABILITATION PROGRAM) AGENCIES. A REVISED FORM OF THE JOB SATISFACTION INVENTORY, ORIGINALLY DEVELOPED BY JOHNSON (1955), ASSESSING SATISFACTIONS WITH EIGHT DIMENSIONS OF THE JOB (RELATIONS WITH ASSOCIATES, RELATIONS WITH EMPLOYER, INTEREST IN JOB, JOB INFORMATION, WORK CONDITION, GOALS, PHYSICAL AND MENTAL EXERTION, AND FINANCES), WAS USED TO COLLECT INFORMATION ON THE NATURE OF COUNSELOR JOB SATISFACTION AND TO UNDERSTAND THE RELATIONSHIPS OF SUCH SATISFACTIONS WITH OTHER CRITERIA OF PERFORMANCE. A CLUSTER ANALYSIS PERFORMED ON THE CORRELATION MATRICES OF MALE AND FEMALE COUNSELORS RESPECTIVELY FOUND THE JOB SATISFACTION OF MALE AND FEMALE COUNSELORS TO BE SIMILAR. HOWEVER, THE ASSOCIATION OF SATISFACTION WITH OTHER CRITERIA OF PERFORMANCE WAS DEPENDENT TO SOME EXTENT ON THE SEX OF THE COUNSELOR. THIS DOCUMENT WAS PUBLISHED IN "THE CRITERIA PROBLEM IN REHABILITATION COUNSELING" AS CHAPTER VI (PP. 42-47), APPENDIX D (PP. 76-82), AND REFERENCES (PP. 123-129). (PS)

THIS RESEARCH CONCERNS THE RELATIONSHIPS AMONG REHABILITATION COUNSELOR PERFORMANCE CRITERIA CURRENTLY BEING USED OR READILY AVAILABLE TO STATE VOCATIONAL REHABILITATION AGENCIES. THE 143 COUNSELORS STUDIED CAME FROM MIDDLE-SIZED AGENCIES IN SIX STATES AND, SINCE COWORKER RATINGS WERE REQUIRED, THE SAMPLE WAS LIMITED TO COUNSELORS WORKING WITH TWO OTHER COUNSELORS AND A SUPERVISOR. THE FOLLOWING PERFORMANCE MEASURES AND RATING PROCEDURES--THE COWORKER RATING BLANK, SUPERVISOR RATINGS, EFFICIENCY RATINGS, JOB SATISFACTION INVENTORY, AVERAGE CASELOAD, AVERAGE NUMBER OF CLOSURES, AND CASELOAD VELOCITY INDEX--ARE DESCRIBED AND RELIABILITY DATA GIVEN. THE EIGHT VARIABLES WHICH RESULTED WERE CORRELATED AND A CLUSTER ANALYSIS OF THE MEASURES WAS MADE. THE TWO RESULTING CLUSTERS WERE"CASE MANAGEMENT" AND "PERFORMANCE RATING." THIS DOCUMENT WAS PUBLISHED IN "THE CRITERIA PROBLEM IN REHABILITATION COUNSELING" AS CHAPTER IV (PP. 28-34), APPENDIX C (PP. 73-82), AND REFERENCES (PP. 123-129). (PS)