Search

In a related study, factor analysis was applied to reduce a selected set of medical school characteristics to their principal dimensions. In this study, the results were then used as input to a series of multivariate cluster analyses that isolated clusters of medical schools that were similar to each other and different from schools in other clusters on the dimensions depicted by the factor analysis. The eight clusters in the final solution each had distinctive profiles on the six factor scores. There were five clusters that consisted completely or predominantly of public schools. Three of these clusters consisted of established schools with varying profiles, while the other two were composed of new and developing schools. Of the remaining three clusters, two were predominantly private schools and one was an equal mix of public and private schools. Each cluster was also described in terms of variables selected from the original data. (Author/MSE)

Multidimensional scaling methods were used to derive interpretable models of medical school similarity with respect to research and graduate medical education intensiveness. On the basis of cluster analysis, private schools seemed to be categorizable into those that are relatively intensive on both research and graduate medical education, those that are not intensive on either, and some that, along with some public schools, are both extensive (in absolute measure) and intensive (in relative measure) on graduate medical programs. These groups are readily apparent in the scaling model, but graduations of difference within groups are also apparent. The public schools showed similar separability along continuous dimensions of difference. The statistical analyses are discussed at length. (Author/MSE)

The basic objective of this research project was to determine the effectiveness of the Marine Corps Task Analysis Program (TA) as it had been conducted by the Office of Manpower Utilization (OMU). Main findings and conclusions are summarized in the eight research areas into which the study was divided: TA observation and interview procedures; task inventory construction; occupational field sample size; computer procedures and data analysis; OMU organization and personnel; orientation, training, and team performance; peace-time task analysis and its relation to war-time conditions; and the use of worker characteristics in classification and assignment. The overall conclusion from the research is that the TA program is a highly valuable asset to the Marine Corps in improving and maintaining effective utilization of its human resources. Appendix A contains an annotated list of 15 technical reports that are end-products of this project, and which contain detailed documentation of all phases of this project. Appendix B contains the study directive used and a distribution list. (Author/TA)

This study investigated (1) relationships between reading comprehension, IQ, and equivalence range (ER) and (2) categorizing styles through multidimensional scaling and varimax rotation analysis. One hundred and six male and 97 female sixth-graders comprised three reading groups according to Metropolitan Achievement Test, Reading subtest, and Otis Quick Scoring Mental Ability Test. Alternate forms of Object-Sorting Test determined categorizing ability. Results show reading comprehension relates to two ER measures for female poor readers; intellectual ability to one ER measure for female good and poor readers and to another ER measure for overall good, overall poor, and male poor readers. Multidimensional scaling and varimax rotation disclose somewhat similar categorizing styles of good and average readers, while both seem different from that of poor readers. (Author)

The Richfield Educational Assessment Program is a districtwide effort to collect specific information about the knowledge, skills, understandings, and attitudes of students in selected areas. The information collected will be used to help Richfield citizens and educators make decisions for program improvement. The program closely parallels the Minnesota Educational Assessment Program, but it allows a local district to analyze the performance of its own students. Before the assessment, teachers set the standards for student performance; Richfield was the first district in the nation to set local standards, or expectation levels, for performance on items used in both the national and state assessments. The 1974-75 program evaluated 9- , 13- , and 17-year-olds in five cognitive areas of mathematics skills: recall and recognition, performing mathematical manipulation, understanding mathematical concepts and processes, problem solving, and analyzing problem situations. (BW)

Occupational analysts using Comprehensive Occupational Data Analysis Programs (CODAP) make subjective decisions at various stages in their analysis of an occupation. The possibility exists that two different analysts could reach different conclusions in analyzing an occupation, and thereby provide divergent guidance to management. Two analysts, working independently, performed job typing using CODAP on occupational data collected by a single administration of a job inventory. Each analyst selected the same size sample, at random but without any cases common, from the total survey. The job types identified by each analyst were compared to determine whether they were significantly different in any respect. Superficial differences appeared between the results of the two analyses, but were largely resolved when job types reported by the first analyst were compared not only on a one-to-one basis with job types reported by the second analyst, but also with combinations of job types reported on the second occasion. The contents of this report are highly technical and will be of primary interest to persons working in the field of occupational analysis, both in terms of content and application of techniques. (Included are 54 statistical tables.) (Author/BP)

This study was a replication of an earlier study by Mazyck (1971) which was designed to analyze characteristics of paraprofessional child care workers, as determined by ratings given on a scale of paraprofessional worker characteristics. The original study is reviewed and the development of the Mazyck Rating Scale for Paraprofessionals, is discussed. The Scale was composed of two categories of characteristics: Personal-Social, and Educational-Biographical-Working Relationships. In the present study, the category Reaction to Stress was added, and the original experiment was replicated with a different and smaller group of child workers and specialists similar to those used in the first study. Findings show that from both studies there are 20 scale items that each of the groups agree on as being significant characteristics of paraprofessionals. The next step is to devise ways of measuring the identified characteristics and to develop instruments useful for persons interested in selection of child care workers. Extensive statistical data are included, as well as a copy of the Rating Scale as used in the second study. (ED)

An empirical assesssment of the space distortion properties of two prototypic hierarchical clustering procedures is given in terms of an occupancy model developed from combinatorics. Using one simple example, the single-link and complete-link clustering strategies now in common use in the behavioral sciences are empirically shown to be space contracting and space dilating, respectively. An extension to an intermediate r-diameter clustering criterion defined in between the complete-link and single-link extremes is discussed briefly along with a technique for preprocessing the original proximity measures to make the results obtained from these two extreme clustering procedures more consistent with the data. (Author)

This study was designed to develop a conceptual scheme and brief reliable measures of parent and teacher involvement and interaction, to be given to kindergarteners' parents and teachers at the time of enrollment and again at the end of kindergarten. The inventories provide a framework for an analysis of (1) characteristics of parents and teachers that influence home-school collaboration; (2) the extent to which different policies, programs, and interventions might influence parent-teacher interaction and involvement; and (3) the relationship of different levels of parent and teacher involvement and interaction to the child's school adjustment and achievement. A search of ERIC documents pertaining to home-school relationship, and a review of research on parent involvement in the child's education in the home and in the school, provided the background for developing the inventories. A small pilot study was conducted with 20 parents of 4- and 5-year-olds at a child development center. Intercorrelations of scale scores were computed for the "a priori" scales for the pilot data and were examined for clusters of scales that might define major dimensions of parent-teacher interaction and parent involvement. These clusters indicate that major dimensions of parent-teacher involvement and interaction can be isolated from this inventory. (CS)

The author has developed a model for transferring and implementing educational innovations from one institution to another. The model consists of an organized set of variables whose precedence relationships were determined by statistical analysis of collected data. A questionnaire was sent to a sample of prospective consumers of educational technology. Using their responses to the questionnaire, several forms of multivariate statistical analysis were used to determine which factors were important to consumers. These factors were then used for the design of a data collection form for innovation developers, and the foundation for the design of a data management system. (Author/CH)

A graph-theoretic paradigm is used to generalize the common measures of categorical clustering in free recall based on the number of observed repetitions. Two graphs are defined: a graph G that characterizes the a priori structure of the item set defined by a researcher, and a graph R that characterizes a subject's protocol. Two indices of clustering, denoted by gamma and omega, are obtained by evaluating the sum of the pairwise products of the weights on the corresponding edges of the two graphs. The gamma statistic is a direct generalization of the commonly used clustering indices and reduces to the number of repetitions whenever G represents a standard categorical decomposition of a stimulus list. The omega statistic, on the other hand, extracts more information from the protocol graph R than does gamma and incorporates a distance measured based on the number of intervening items in a subject's recall sequence. (Author)

Two studies were undertaken to demonstrate the usefulness of partitioning procedures for studying test items. Achievement test items in five content areas of educational measurement were used as stimuli to be sorted by groups of students with varying levels of sophistication with the content, with the hypothesis that sorting by classes with greater sophistication would agree more with simulated target sortings than sortings by classes with less sophistication. These sortings were analyzed using partitioning procedures. Results from both studies indicated that degree of sophistication in measurement was overall a potent variable in the sorting. In addition, several misconceptions among the students concerning the content under study were revealed. It was noted that a moderate number of students enrolled in upper-level measurement courses demonstrated what amounted to errors in knowledge in their sortings. It was concluded that the partitioning procedures were useful for studying how items are perceived by students and for determining how students organize content. (Author)

A complex two-stage sample selection process was used in designing the National Longitudinal Study of the High School Class of 1972. The first-stage sampling frame used in the selection of schools was stratified by the following seven variables: public vs. private control, geographic region, grade 12 enrollment, proximity to institutions of higher education, percentage of minority group enrollment, community income level, and degree of urbanization. Six hundred strata determined the initial sample of 1,200 schools; later, a random sample of 18 seniors per high school was selected. This report considers the effects of stratification, oversampling of schools by percentage of minority group enrollment and community income level, clustering of students within a school and unequal weighting on the variances of the resulting statistics and therefore, on the precision of the sample statistics. Results suggest that school stratification variables reduced the variances of national estimates to twenty percent below what would have been expected with unstratified cluster sampling. Of the five major stratification variables: socioeconomic status, size of school, type of control, geographic region, and proximity to college or university; region is the strongest variable and type of control is the weakest. (Author/MV)

A study was conducted to describe the television viewing habits of grade 6 children in primary schools within the metropolitan area of Melbourne, Australia; to examine the nature of the relationships between factors found to be relevant in explaining television behavior; and to examine the relationship between television behavior and school achievement. Data were collected from a 2-stage cluster sample of 271 grade 6 children using a questionnaire; tests of intelligence, self-esteem, and comprehension in social studies; and a diary of television viewing over a period of seven days. A questionnaire was also completed by the subjects' teachers. Statistical analyses involved simple descriptive statistics, bivariate correlation and multiple regression analyses, and canonical analysis. The hypothesizing and subsequent testing of causal path models of the interrelationships between variables enabled the study to cope more efficiently with the complex interrelationships between variables. Results confirmed the important influence that television has on the lives of children, and the importance of home background variables and personal characteristics in influencing television viewing patterns. (Author/CMV)

This description of the technical details required for using the HIER-GRP computer program, which was developed to group or cluster regression equations in an iterative manner so as to minimize the overall loss of predictive efficiency at each iteration, contains a discussion of the basic algorithm, an outline of the essential steps, specifications of the computer system requirements, descriptions of necessary control cards, and explanations of the program output. Appendices include the mathematical formulas used, some mathematical background helpful for understanding the algorithm, sample output, and a complete source card listing. (Author/RAO)

Although it is commonly assumed that increases in mathematical knowledge and problem-solving skill alter one's understanding of mathematical problems, little research exists to support this assumption. The present study investigated the relationship between mathematical background and problem perception in two experiments. Experiment 1 employed hierarchical clustering analysis to compare the way that experts (nine mathematics professors) sorted 32 mathematics problems typical of college mathematics courses with the way that novices (19 undergraduates) sorted the same problems. The experimenters assigned an a priori mathematical "surface structure" and a mathematical "deep structure" characterizations to each problem. "Surface structure" refers to the items described in the problems themselves. "Deep structure" refers to the mathematical principles necessary for solution. Results indicated that the two groups use different criteria for considering problems to be related. Experiment 2 compared changes in mathematics problem perception of students who took a computer course during the same time period. Training in problem solving resulted in the experimental group's problem perception being more differentiated and more like that of experts. Appendices contain the 32 problems used in the experiments and the mathematics pretest. (Author/RL)

One of the major assumptions of latent trait theory is that the items in a test measure a single dimension. This report describes an investigation of procedures for forming a set of items that meet this assumption. Factor analysis, nonmetric multidimensional scaling, cluster analysis and latent trait analysis were applied to simulated and real test data to determine which technique could best form a unidimensional set of items. Theoretical and empirical evaluations were also made of the effects of guessing on the dimensionality of test data. The results indicated that guessing affected highly discriminating items more so than poorly discriminating items. Of the procedures evaluated for sorting items into unidimensional item sets, principal factor analysis of phi coefficients gave the best results overall. In summary, guessing does have an effect on test data, but the effect is not very large unless items of extreme difficulty are present in the test. Of the procedures evaluated, traditional factor analytic techniques gave the most useful information for sorting test items into homogeneous sets. (Author/GK)

A promising avenue for characterizing individuals lies in an examination of the antecedents of recorded behaviors. Autobiographical information from college students was used as an alternative to traditional personality methodology to demonstrate the scientific utility of biodata. The traits selected for investigation were positive and negative emotionality. Biodata items were clustered rationally, based on their correlations with the traits in question and divided into two classes, i.e., behavioral and developmental. Hypotheses were formed about the relationship between various background variables and emotionality, independent of the clustering process. Results yielded a substantial overlap between the hypotheses and the clusters, providing evidence for the validity of the clusters and for the utility of biodata methodology in personality research. In addition, the biodata substantiated virtually all previous research findings. The findings appear to extend current knowledge of the specificity of the relationships between background and positive and negative emotionality. (Author/JAC)

Organizations are directly influenced by the beliefs and values of the employees. Many of these beliefs deal with the meaning of work and preferences for broad classes of work outcomes. A heterogeneous sample of working people (N=318) completed a questionnaire which examined beliefs about work on nine job-related dimensions that distinguish four beliefs systems. A cluster analysis performed on the nine job-related subscale scores indicated that there were five clusters of people who differed in their beliefs about work. A comparison of the means of the five clusters on the subscales revealed that: (1) Cluster 1 tended to hold Protestant Ethic beliefs; (2) Cluster 2 adhered to the Leisure Ethic; (3) Cluster 3 was neutral relative to the other groups on almost all scales; (4) Cluster 4 endorsed more Marxist beliefs than other clusters; and (5) Cluster 5 held a combination of Protestant Ethic and Humanistic beliefs. For most clusters, beliefs from more than one ethic were held. These data suggest that the theory-based work beliefs scale is a superior instrument for measuring the beliefs of workers. (Author/NRB)

Outcome data on smoking cessation has emphasized that most people have difficulty not in quitting smoking, but in maintaining cessation. An attempt was made to develop a more meaningful typology of relapse-promoting situations using a sample of 183 exsmokers who called a telephone hotline seeking help to stay away from cigarettes. Two higher order clusters emerged from the cluster analysis of their responses: a positive affect situation in which self-indulgence was prominent, and a negative affect situation characterized by high levels of stress and feeling of anxiety and depression. The most homogeneous subtype of exsmoker occurred when the individual was faced with smoking-related stimuli in social drinking situations. The typology of smoking motivations suggests that the enhancements of positive affects and dimunition affects are major motivators for smoking. (Author/JAC)

This study developed, applied, and evaluated a theory-based method of detecting the underlying causes of differential difficulty. The method was applied to two subgroups taking the Scholastic Aptitude Test-Mathematics (SAT-M), 261 visually impaired students taking Braille forms of the test and 1,985 black students at 3 test administrations. It involved: (1) reviewing the literature to identify possible causes of differential item functioning; (2) forming item categories based on those factors; (3) identifying categories that functioned differentially; (4) assessing the functioning of the items composing deviant categories; and (5) relating item and category functioning. Results were compared to a traditional item-level analysis. In both subgroups, the cluster and traditional methods agreed on the overall extent of differential functioning (substantial in the first group, virtually absent in the second). The method would seem to be applied most productively when a small number of hypotheses can be derived from a reasonably strong research base, overlap among cluster structures can be avoided, and results can be supplemented with experimental studies of protocol analysis. (Contains 27 references, 11 tables, and an appendix of category definitions.) (Author/SLD)

This study is an attempt to model the similarities of 84 medical schools with respect to their orientations toward applicants qualified for research and applicants interested in delivering primary care or locating in non-urban settings. These characteristics are defined in 17 institutional variables. The patterns of insti(utional similarity are described by two modeling techniques: cluster analysis and multidimensional scaling. Multidimensional scaling confirmed that there were essentially two dimensions of difference among medical schools with respect to the 17 variables studied. It appears that there are fewer medical schools with preference for research-oriented students with goals to provide health care directly to patients and in geographic areas currently underserved. Cluster analysis identified four groups of schools as characteristically different from one another. Subsequent analysis of data for schools in each group served to identify each group's distinguishing attributes. Research-oriented schools appear to form a distinct class of institutions. Other schools form three classes according to the degree of their preference for students oriented toward primary care service. A simultaneous presentation of the scaling and clustering models provides the most complete picture of medical school similarities with respect to preferences for students who would provide medical services directly through primary care delivery or indirectly through medical research. (SPG)

Many studies which have investigated the differences between decided and undecided subjects have assumed homogeneity of both subsets, but results of these studies do not justify such a assumption. This study attempted to identify, multidimensionally, types of vocationally undecided college students. Data on 11 variables from 276 undecided undergraduate students were entered into a cluster analysis. Data from the cluster analysis suggest that the optimal partioning of the data was obtained with a five-cluster solution. Analyses of variance and a chi-square analysis using variables not included in the clustering process provided evidence of the validity of found clusters. To assess the reliability of the clustering process, the total sample was split into two subsamples and a five-cluster solution was imposed on each of the subsamples. The resulting clusters were compared to the clusters of the total sample and to the findings of a former, similar study. These comparisons provide support for the reliability of three of the five clusters. The emergence of these clusters suggests a multidimensional perspective of vocational decidedness with apparent implications for differential treatment. (Author/NB)

This study of patterns of microcomputer-based instruction in elementary and secondary schools had three objectives: (1) to describe how teachers who were nominated by their peers as "unusually successful" in their microcomputer-based mathematics and science teaching use technology for instruction; (2) to describe how these uses vary as a function of teacher characteristics and other background variables; and (3) to recommend policies for educating teachers in the instructional uses of microcomputers and for devising courseware that serves pedagogical aims. A total of 60 teachers--40 elementary and 20 secondary--in 25 districts and 49 schools in the state of California were interviewed. Schools varied in the number of microcomputers available for instruction (1 to 55, with an average of about 12), and in the resources they provided for microcomputer-based instruction. Based on 16 variables of instructional characteristics underlying microcomputer-based instruction, four homogeneous clusters of teachers were identified: orchestration (N=18), enrichment (N-23), adjunct instruction (N=14), and drill and practice (N=5). Analysis of the data indicates that, although teacher attitudes toward microcomputers were not related to the patterns of microcomputer-based instruction that were identified, all teachers held uniformly positive attitudes. Teacher knowledge about microcomputers was also unrelated to the average amount of coursework taken in mathematics, but the amount of science taken as an undergraduate did reflect positively on instructional use. Recommendations are presented for staff development and the characteristics of teacher-friendly courseware are discussed. Appendices include the interview guides and questionnaires used as well as discussions of methodological considerations and modes of microcomputer-based instruction. A list of references completes the document. (JB)