Simplicity in Path Analysis

Abstract

Path analysis is too concerned with the observable and not concerned enough with the conceptual. Many path analyses are rendered trivial, and overly complicated by this emphasis. A method proposed by Hope (1971) within the framework of path analysis is shown to add its own further confusion, and to be invalid. The same data is analysed by two (differing) approaches which are less concerned with the observed and more concerned with what underlies the observed. Basic deductions from both methods (one is regression-based, the other factor analysis-based) are the same. One, the data armed by Hope contains much redundancy in conceptual terms: two, the underlying process is very simple: `A person's level of education is predicted (caused?) by his mental ability and type of parental home environment, with ability being a more important factor. Most systematic variation in a person's occupational prestige is "caused" by the person's level of education'.