Search

This final report summarizes past research and suggests new approaches to the problem of estimating long-term individual constants using path analysis. The general objective of the research was to detect and measure the likelihood that one variable, x, measured at time, t, has a causal influence on another variable, y, measured at a subsequent time t + k (k being the measurement interval). Work on this problem continues under a grant from the National Science Foundation. (CK)

The objective of this study was to review and integrate the various methodologies used in the study of individual growth (especially academic growth). This was accomplished by means of Joreskog's general model for the analysis of covariance structures, i.e., each of the disparate methodologies available from the literature was shown to be a special case of Joreskog's general model. Two general considerations enter into the study of growth and its determinants: (a) making provision for errors of measurement, and (b) constructing a model which relates growth to its determinants in a causally meaningful way. Errors of measurement typically involve questions about the reliability and/or validity of measures, i.e., only indirect measures of the desired variable (construct) are available. Multiple measures of each construct would appear necessary to deal with measurement errors in a quantitative manner. For this purpose the multitrait/multimethod approach devised by Campbell and Fiske (1959) is a useful approach since in principle it allows for correlated errors of measurement. Because the Campbell-Fiske approach does not specify the exact relationships between observed variables and constructs, a factor analytic formulation of their approach was used in order to summarize various approaches to measurement error. The constructs, which represent the growth variable and its determinants, were then interrelated in terms of a linear structural (causal) model. The implications of this model, which itself is a special case of Joreskog's general model, were considered. (Author)

Conducted as part of the Vocational Development Study project, this study utilized Path Analysis to develop, evaluate, and modify a model for predicting the stability of early expressed occupational choice of secondary students. Analysis of data from 550 students indicated that none of the 21 student characteristics investigated served as a denominator to separate those students who will express the same occupational choice throughout the secondary years from those who will not. The best indicator of stability of occupational choice throughout high school appeared to be stability of occupational choice from ninth to tenth grades. Of those students who chose the same occupation in both ninth and tenth grades, two thirds of them continued to choose the same occupation at twelfth grade. Several other student characteristics which increase the chances of stability of occupational choice include success in terms of grade point average, enrollment in a vocational curriculum, and vocational maturity. The student's socioeconomic background, sex, or level of occupational choice do not have an effect on stability of occupational choice. (Author/SB)

Since the development of causal path analysis by Wright, both the standardized and unstandardized procedures in path coefficient estimates have been advocated in the related literature, but the superiority of one technique over the other has not been demonstrated empirically or mathematically. In this study, the sampling properties of both coefficient estimates in a chain prediction model are compared under known conditions by Monte Carlo methods at systematically selected parameter points which include sample size and Var(E-sub-one, E-sub-two). The implications, interpretation, and problems of both methods are discussed with reference to the evaluation of learning hierarchies. (Author)

A general method for estimating the unknown coefficients in a set of linear structural equations is described. In its most general form the method allows for both errors in equations (residuals, disturbances) and errors in variables (errors of measurement, observational errors) and yields estimates of the residual variance-covariance matrix and the measurement error variances as well as estimates of the unknown coefficients in the structural equations, provided all these parameters are identified. Two special cases of this general method are discussed separately. One is when there are errors in equations but no errors in variables. The other is when there are errors in variables but no errors in equations. The methods are applied and illustrated using artificial, economic and psychological data. (Author)

This paper reports on attempts by the author to construct a theoretical framework of adult education participation using a theory development process and the corresponding multivariate statistical techniques. Two problems are identified: the lack of theoretical framework in studying problems, and the limiting of statistical analysis to univariate techniques. The process of theory development is divided into: (a) definition of concepts; (b) establishment of variable relationships; (c) specification of causal ordering of variables; (d) identification or determination of relationships among variables within the same causal system; and (e) estimation of change. The use of regression analysis and models in the final phase is discussed, as is also functional path analysis to determine whether the proposed set of interpretations is consistent throughout. The final results will be presented in a doctoral thesis. (author/pt)

THIS REPORT PRESENTS A DISCUSSION OF 2 TECHNIQUES WHICH CAN BE USED TO REPRESENT AND INTERPRET MULTIVARIATE STATISTICAL SYSTEMS WHEN IT IS FELT THAT THERE ARE CAUSAL RELATIONS BETWEEN SOME OF THE VARIABLES. THE BASIC TECHNIQUE IS PATH ANALYSIS AND THE OTHER IS ITS EXTENSION THROUGH THE USE OF RECURSIVE SYSTEMS OF EQUATIONS. THE ANALYSIS IS RESTRICTED IN APPLICATION TO RELATIONSHIPS BETWEEN INTERVAL-MEASURABLE VARIABLES THAT ARE LINEAR, ADDITIVE, AND ASYMMETRIC. TO MAKE A PATH ANALYSIS, THE VARIABLES IN THE SYSTEM ARE CLASSIFIED AS EITHER EXOGENOUS, THAT IS, HAVING THEIR VALUES DETERMINED BY FACTORS OUTSIDE THE SYSTEM, OR ENDOGENOUS, THAT IS, HAVING THEIR VALUES DETERMINED BY FACTORS REPRESENTED BY VARIABLES WITHIN THE SYSTEM. BASED ON THIS ANALYSIS, A SET OF REGRESSION EQUATIONS REPRESENTING THESE RELATIONS IS FORMED. THIS SET IS TERMED THE PATH MODEL, AND GRAPHIC CONVENTIONS ARE GIVEN FOR DIAGRAMING IT. THE COEFFICIENTS IN THE EQUATIONS ARE SIMILAR TO THE CORRELATION COEFFICIENTS OCCURRING IN ORDINARY LEAST-SQUARES REGRESSION EQUATIONS. THE ADVANTAGE OF THE PATH ANALYSIS APPROACH IS THAT IT ENABLES THE EXPERIMENTER TO UTILIZE ALL THE INFORMATION AT HIS DISPOSAL, PARTICULARLY THAT CONCERNING CAUSAL RELATIONS BETWEEN VARIABLES. THE TECHNIQUE IS ILLUSTRATED WITH APPLICATIONS TO BIVARIATE AND MULTIVARIATE SYSTEMS HAVING SINGLE AND MULTIPLE STAGES OF CAUSAL INFLUENCE. SOME EXAMPLES DRAWN FROM ACTUAL RESEARCH PROJECTS ARE INCLUDED. (DR)

THIS STUDY ATTEMPTS TO DEMONSTRATE THAT PATH ANALYSIS IS A VALUABLE TOOL FOR INTERPRETING CORRELATIONS IN A CAUSAL SENSE. PATH ANALYSIS IS APPLIED TO A NONEXPERIMENTAL, PANEL SURVEY IN AN EFFORT TO DETERMINE WHETHER THE MORE SELECTIVE OR LESS SELECTIVE COLLEGES HAD A DIFFERENTIAL IMPACT ON THE EDUCATIONAL PLANS OF THEIR STUDENTS. THE PROBLEM IS TO INTERPRET THREE CORRELATIONS--SELECTIVITY WITH CHANGES IN EDUCATIONAL PLANS, SELECTIVITY WITH COLLEGE GRADES, AND COLLEGE GRADES WITH CHANGES IN EDUCATIONAL PLANS. DATA FOR THE STUDY WERE OBTAINED FROM 127,125 ENTERING 1961 FRESHMEN IN 248 FOUR-YEAR COLLEGES AND UNIVERSITIES. THE GENERAL PROCEDURE WAS TO CONSTRUCT SIX EQUATIONS USING SEVEN VARIABLES--FATHER'S EDUCATION, NATIONAL MERIT SCHOLARSHIP TEST SCORE, HIGH SCHOOL GRADE AVERAGE, FRESHMAN EDUCATIONAL PLANS, SELECTIVITY OF COLLEGE ATTENDED, FRESHMAN YEAR COLLEGE GRADES, AND SOPHOMORE EDUCATIONAL PLANS. THE RESULTS SUGGEST THAT CHANGES IN EDUCATIONAL PLANS ARE A POSITIVE FUNCTION OF THE DEGREE TO WHICH A STUDENT'S ACADEMIC PERFORMANCE DIFFERS FROM THAT PREDICTED FROM HIS BACKGROUND AND THE COLLEGE HE ATTENDS AND THAT THE DIRECT INFLUENCE OF COLLEGE SELECTIVITY ON EDUCATIONAL PLANS APPEARS TO BE SMALL OR NONEXISTENT. IT IS CONCLUDED THAT IT IS EXTREMELY DIFFICULT TO PUT THEORIES ABOUT COLLEGE ENVIRONMENTS INTO TESTABLE FORM. (HW)

Author presented a critical analysis of the conclusions drawn by Jencks in his book. (RK)

Author concluded that Jencks' book was about the inequality of pay in the United States and its relationship to family background as well as educational opportunity. (Author/RK)

This paper presents the finding of an investigation for determining course achievement utilizing the statistical technique of path analysis. (Author)

The Werts-Linn procedure for dealing with categorical errors of measurement in "Comments on Boyle's 'Path Analysis and Ordinal Data'" in The American Journal of Sociology, volume 76, number 6, May 1971, is shown to be inappropriate to the problem of ordered categories. (For related document, see TM 002 301.) (DB)

Paper proposes a set of techniques for measuring the importance" of each independent variable in a multivariate causal law (i.e., a law showing the combined effect of several independent variables on a single dependent variable). (Authors)

Focus of this paper is on the logic of the model rather than the estimation of parameters or testing the fit of the model to data. (Authors/MB)

This study of the relationship of perceived role to satisfaction with a job presents four plausible models based on the variables role accuracy compliance, performance evaluation, and satisfaction. These models are evaluated by path analysis. Compliance and performance evaluation are shown to be important variables mediating the relationship between role accuracy and satisfaction. (Author/JH)

Focuses on the analytic and methodological base for Jencks' finding that the correlation between income and educational variables is very low; apparently the Jencks group considered path analysis a powerful inferential technique, while ignoring the limitations placed on it by a highly restrictive series of assumptions. (Author/JM)

The author of Inequality'' attempts to clear up some misconceptions about the contents of his book and rebuts arguments concerning the use of path analysis. (Author/KM)

A detailed review of the methodological shortcomings in Jencks' Inequality'' suggests that the policy recommendations implied in the book are not strongly supported. (Author/KM)