Statistical and Substantive Implications of the Use of Stepwise Regression to Order Predictors of Leisure Behavior

The correlation coefficient and problems of inference in recreation research

“…Blalock (1972) was one of the first to popularize the notion that inferences about the relative strength of a relationship between two variables, in two different samples, was frequently inappropriate if the conclusions were based on a comparison of the correlation coefficients. Christensen and Yoesting (1976) later noted this problem in leisure research. The correlation coefficient.…”

“…As a consequence, r xy would vary across samples for substantively irrelevant reasons. Comparing r xy 's across different studies would have no substantive meaning and would lead to erroneous conclusions about the nature of the relationship between variables x and y (Christensen and Yoesting, 1976). McCuen (1974) discussed correlations in recreation demand studies to illustrate a second way that correlation analysis could lead to misleading inferences.…”

The importance of urban milieu in predicting recreation participation: The case of day hiking

“…Although there has been much debate concerning the problems of choosing the "best" predictors in a multiple regression equation, it has generally been recognized that the order of step wise entry is a simple and efficient method of identifying such predictors (Crandell 1976;Christensen and Yosting 1976;McCuen 1974), and Table 3 shows the best five predictors for each milieu. As one might expect, given the previous results, there are predictors that are common to at least two of the groups; however, there are also substantial differences.…”