A test of the function predicting the book circulation in public libraries by multiple linear regression analysis

In many paper regression models of book circulation have been discussed as a use study in public libraries. But it is insufficient in Japan to study such regression models. This paper examines the validity of regression models of book circulation in public libraries of Osaka and Toyama, Japan. In our regression models dependent variable is book circulation per capita in each community and independent variables are library collection per capita, a number of annual acquisitions per capita, number of libraries in each community, proportion of professional occupation, daytime population and so on. Multiple linear regression analysis with a backward elimination method is used to four data sets; Osaka and Toyama in each 1980 and 1985. As a result regression models fit well in three data sets except Toyama in 1985 (R² of these three models are 0.79-0.87). These results suggest that the regression model is applicable for predicting book circulation of public libraries.