Econometric Analysis Suggests Possible Crowding Out of Public Libraries by Book Superstores among Middle Income Families in the 1990s

Objective – To determine the effect of large bookstores (defined as those having 20 or more employees) on household library use. Design – Econometric analysis using cross-sectional data sets. Setting – The United States of America. Subjects – People in over 55,000 households across the U.S.A. Methods – Data from 3 1996 studies were examined using logit and multinomial logit estimation procedures: the National Center for Education Statistics’ National Household Education Survey (NHES) and Public Library Survey (PLS), and the U.S. Census Bureau’s County Business Patterns (CBP). The county level results of the NHES telephone survey were merged with the county level data from the PLS and the CBP. Additionally, data on Internet use at the state level from the Statistical Abstract of the United States were incorporated into the data set. A logit regression model was used to estimate probability of library use based on several independent variables, evaluated at the mean. Main results – In general, Hemmeter found that "with regard to the impact of large bookstores on household library use, large bookstores do not appear to have an effect on overall library use among the general population” (613). While no significant changes in general library use were found among high and low income households where more large bookstores were present, nor in the population taken as a whole, middle income households (between $25,000 and $50,000 in annual income) showed notable declines in library use in these situations. These effects were strongest in the areas of borrowing (200% less likely) and recreational purposes (161%), but were also present in work-related use and job searching. Hemmeter also writes that “poorer households use the library more often for job search purposes. The probability of library use for recreation, work, and consumer information increases as income increases. This effect diminishes as households get richer” (611). Finally, home ownership was also correlated with higher library use. Households with children were more than 20% more likely to use the library (610). Their use of the library for school-related purposes, general borrowing, program activities, and so on was not affected by the presence of book superstores. White families with children were somewhat less likely to use the library, while families with higher earning and education levels were more likely to use the library. Library use also increased with the number of children in the family. Shorter distances to the nearest branch and a higher proportion of AV materials were also predictive of higher library use. Educational level was another important factor, with those having less than high school completion being significantly less likely to use the library than those with higher levels of educational attainment. Conclusions – The notable decline in public library use among middle income households where more large bookstores are present is seen as an important threat to libraries, as it may result in a decline in general support and support for funding among an important voting block. More current data are needed in this area. In addition to the type of information examined in this study, the author recommends the inclusion of information on funding, support for library referenda, and library quality as they relate to the presence of large bookstores.