Abstract
The relationship between waste arisings and various socio-economic factors has been researched for more than 30 years. This exercise attempts to explain local differences in waste arisings, and is also useful in forecasting the amount of waste, and for assessing the efficacy of waste management policies under different socio-economic conditions. However, in some cases, researchers have presented results that appear to contradict each other. It has been pointed out that there is a lack of consistency in waste statistics and that careless choice of independent variables can adversely influence the results of multiple regression. These factors probably cause the instability in analysis results. In this presentation it will be indicated that results contrary to each other can be generated from the same dataset, due to the variation in statistical methods being employed. This will be demonstrated by examining the effect of average income of the residents on waste arisings. When conducting multiple regression analysis using data aggregated at the district (shi-cho-son) level, the problem of heteroskedasticity is prominent. Weighted Least Squares method is used to combat this problem, but depending on the weight variable, the coefficient for income can turn out to be either positive or negative.
