“The problem of the unused book”…still

First, I should explain my absence…I’ve become temporarily disabled by a broken foot.  To those who know my alternate life, no, it was not from playing hockey (although if it was, at least I’d have been having fun breaking it).  Simply missing a step on the stairway in our house led to the loss of my hockey season, spending upwards or $1500 (after insurance), and missing at least two weeks of work.  Now, back to our regular broadcast.

I am trying to write a literature about the analysis of book circulation data for an article that will focus on the distribution of book circulations.  When I write literature reviews, I like to try to find temporal trends in research on whatever topic it is – I think it enables me to put the research into context.  In my literature searching, I came across this article:

Problem of the unused book.(1911). Library Journal (1876)., 36, 428-429.

So, even as far back as 1911, librarians were concerned about this problem.  I am constantly amazed at how little in our field has really changed.  Not even the solutions are much different – display “dying” books to regenerate interest, and put “dead” books into storage.

This article is intriguing because it includes a breakdown by year last used.  I’ve attempted to supplement the analysis, reversing the years since last year used to provide a different perspective.

Last Used	Num. Titles	Cumulative Total	% of Not Used	Cum. % of Not Used	% of All Titles	Cum. % of All Titles
1908	3585	3585	26.8%	26.8%	5.59%	5.59%
1907	2184	5769	16.3%	43.1%	3.40%	8.99%
1906	1439	7208	10.8%	53.9%	2.24%	11.23%
1905	1107	8315	8.3%	62.1%	1.73%	12.96%
1904	781	9096	5.8%	68.0%	1.22%	14.18%
1903	626	9722	4.7%	72.6%	0.98%	15.15%
1902	480	10202	3.6%	76.2%	0.75%	15.90%
1901	474	10676	3.5%	79.8%	0.74%	16.64%
1900	403	11079	3.0%	82.8%	0.63%	17.27%
1899	359	11438	2.7%	85.5%	0.56%	17.83%
1898	234	11672	1.7%	87.2%	0.36%	18.19%
1897	237	11909	1.8%	89.0%	0.37%	18.56%
1896	218	12127	1.6%	90.6%	0.34%	18.90%
1895	125	12252	0.9%	91.5%	0.19%	19.10%
1894	185	12437	1.4%	92.9%	0.29%	19.38%
1893	77	12514	0.6%	93.5%	0.12%	19.50%
1892	60	12574	0.4%	94.0%	0.09%	19.60%
1891	45	12619	0.3%	94.3%	0.07%	19.67%
1890	37	12656	0.3%	94.6%	0.06%	19.73%
1889	25	12681	0.2%	94.8%	0.04%	19.76%
1888	22	12703	0.2%	94.9%	0.03%	19.80%
1887	13	12716	0.1%	95.0%	0.02%	19.82%
1886	51	12767	0.4%	95.4%	0.08%	19.90%
1885	2	12769	0.0%	95.4%	0.00%	19.90%
Never	614	13383	4.6%	100.0%	0.96%	20.86%
Total	13,383		4.0%		20.86%

So…21% (13,383 titles) of the collection (n=64,162) had not been used in the last 2 years.  It is evident that the bulk of the unused titles are more recent, suggesting that it takes time for books to become noticed and used by patrons.  The famous (infamous?) “Pittsburgh Study” tracked usage of books added the collection over five years.  They found that 60% of these titles had not circulated over that same time period (Bulick, 1976).  Based on the data above, the problem of time-in-collection is apparent.  However, other research has clearly shown a decrease in the probability of use over time (Chen, 1976; Jain, et al. 1969).

What is not stated in the article above, however, is their weeding policies.  Were they removing books that had not been used and were deemed undesirable to keep?  The tone of the article suggests that they were not, that this was a novel assessment, indeed removing items entirely from the collection that were still in good shape was not even an option.

Interestingly, they even broke down usage by broad Dewey call number ranges:

Class	Subject	Vols. In Library	Vols. not used in last two years	% Not Used
000	Polygraphy	3057	691	22.60%
100	Philosophy	1162	250	21.51%
200	Religion	3023	693	22.92%
300	Sociology	4333	1226	28.29%
400	Philology	1226	95	7.75%
500	Natural Science	4785	911	19.04%
600	Useful Arts	2667	543	20.36%
700	Fine Arts	2236	347	15.52%
800	Literature	6447	1394	21.62%
910-919	Travel	5226	940	17.99%
920	Biography	5311	1899	35.76%
Other 900’s	History	6690	1568	23.44%

Their explanation for the high rate of disuse for biographies was their policy to collect titles “written and published from a sense of duty, as pious memorials, rather than for any real literary or historical reason…”.  Generally, there appears to be a direct relationship between collection size and percent not used.  Here’s the scatterplot of these two variables, with the regression formula for the linear trendline:

Direct relationship of collection size & disuse

The effect (0.05) is modest, indicating that for every increase of titles in the collection, the percentage of disuse increases .05%, but the fit (R2) is also modest, explaining only about 23% of the variation.  The correlation coefficient (R, not listed here) is 0.48, which is moderately strong.  So it is clear that the size of the collection could have some effect on the usage, in that the fewer items there are, the lower rate of disuse.  This conclusion should not be extended to the end of reducing ones collections.  After all, if this were true, then you would not be seeing the explosion of choices available on retail shelves.  There is more going on here than a simple cause-effect relationship.

Well, my tangent has come to an end.  I am the more wiser to know that the library data has, indeed, been analyzed to explore such problems and derive solutions.  It would be nice to see the data on the effects of their solutions.

References

Bulick, S. (1976). Use of library materials in terms of age. Journal of the American Society for Information Science, 27, 175-178.
Chen, C. (1976). Applications of operations research models to libraries : A case study of the use of monographs in the francis A. countway library of medicine, harvard university. Cambridge, Mass.: MIT Press.
Jain, A. K., Leimkuhler, F. F., & Anderson, V. L. (1969). A statistical model of book use and its application to the book storage problem. Journal of the American Statistical Association, 64(328), 1211-1224. Retrieved from http://www.jstor.org/stable/2286062.

NOTE: I’ve reproduced the article below for my readers’ reference.  This article was published before 1925, and thus is in the public domain.