Spurious Correlation in Estimating Recreation Demand Functions

Publisher Website
Google Scholar
Abstract
Correlation and regression analyses are used extensively in recreational analysis. The correlation coefficient is frequently used as an indication of the validity of a linear relationship between variables. Recreation demand functions are often derived using empirical prediction equations that relate visitation per capita to either variables representing the supply of recreation facilities or variables expressing the cost of travel. But analyses involving ratios of variables, such as visitation per capita or water acreage per capita, should not be extrapolated to inferences concerning the individual variables, such as visitation. A correlation coefficient derived using ratios of variables may not indicate the ability of predictor variables to explain variation in a component variable of the ratio. In fact, such correlation may be spuriously high or low and inferences about the individual variables highly misleading. The concept of ratio correlation is discussed and examples presented to demonstrate potential ambiguity in recreational analysis.1