In a letter published in Library Journal in 1908 U.K. librarian James Duff Brown was puzzled by comparisons of British and American public library statistics. He noticed that the number of registered borrowers per capita among prominent U.S. libraries seemed inordinately high when compared with British libraries. 1907 statistics for East Orange, NJ indicated that 50% of its residents were registered patrons, an extreme percentage that Brown observed would be impossible in the U.K. because “70 per cent [of U.K. residents] are too old, too young, too illiterate, too high-class, or in jails and workhouses, safe from the temptations of public libraries.”1
In a Library Journal issue later that year American librarian Frances Rathbone clarified the East Orange, NJ registration figures and suggested a modified statistic that might help in comparisons with U.K. data. Still, that statistic put East Orange registered borrowers at 33% of the town’s residents, causing Rathbone to write:
It is an unusually large portion of the population, but our conditions are unusual. East Orange..[has] street after street of thrifty and well-to-do residences. There is scarcely a person to whom the book is not part of his life…It is surrounded by other cities, not open country, and whatever persons are in jails or workhouses are looked after by the county elsewhere.”2
More than a hundred years later unusual libraries with unusual data are still evident in U.S. public library statistics. The chart below (borrowed from my April 2009 post) shows how national library statistics are spread unevenly—or, in statistical jargon, have “skewed distributions.”3 Most libraries’ statistics fall within a given range while a few libraries’ data are especially high.
Yet, as Frances Rathbone explained, statistics that appear unusual are not necessarily wrong. There can be quite legitimate reasons for high figures. A good example is how libraries serving substantial numbers of patrons outside their legal service areas end up with very high per capita statistics.
On a continuum of library evaluation tools, library ratings are among the simplest and most rudimentary. Second-guessing, changing, or eliminating unusually high (or low) library statistics is inappropriate when calculating library ratings. Are a given library’s operations superb (or substandard)? Is the community an especially fertile (or infertile) environment for library services? Is the library highly attuned (or insensitive) to community information needs? Have data collection methods exaggerated (or under-reported) the data? Only intensive surveys and onsite audits of U.S. libraries would lead to the information needed to answer these questions.
1 Library Journal, Vol 33, No 10, Oct. 1908, p. 395.
2 Library Journal, Vol 33, No 12, Dec. 1908, p. 506.
3 This is true for traditional measures like visits and circulation and also for newer ones like electronic usage.