Thematic analysis and "obsolescence"

References

Cite

Share

Abstract

Kuch’s [ l ] most interesting discussion of the role of themata in information science uses my contributions to the literature on “obsolescence” as a study example. He is only partly right in attributing my M2 attitude to a “thema” as “a belief deeply and often unquestioningly held.” I entered the library world relatively late in life in 1963, having trained as a biologist in the 1930s before much sophisticated observational apparatus was available and before it was possible for a degree course to omit systematics and anatomy. In those fields, as well as behavioural studies, observational descriptions are paramount, and although subsequent work may refine the data these are rarely overthrown. I was aware too of the rediscovery in 1900 of Mendel’s fundamental observations published in 1866. Huxley [2] has shown how descriptions of the appearance of contracting muscle fibres were correct in the 1890s, but wrong in the textbooks of the 1930s and 1940s, a fact which he attributed to bright young men employing electrical and chemical methods and distrusting discoveries made by the microscope. There was therefore nothing in my training or experience to justify rejecting scientific evidence or ideas just because they were old. The true thema to which I operate by “inherent belief unquestioningly held” is the need to get down to and challenge the underlying assumptions in any theory, to check them against the observational and experimental evidence, and to check that the experimental and deductive stages are valid and logical. I did not find that the concepts of “obsolescence” and “exponential decay in value” stood up to such critical examination. From the observation that fewer old books are used it is fallacious to deduce that each older book is less frequently used. The fallacy arises from the failure to appreciate that every observed count of use is not a pure number but a ratio; that is to say it is a count of the number of uses from a group of items. If the two (or more) groups from which the counts are taken are of different sizes (as all age groups are), then the counts cannot be compared until all have been reduced to a common denominator, i.e., expressed as uses per book, per hundred books, per meter of shelf, or whatever, and it clarifies the position to describe such uses-per-x as “densities.” (Ten uses of one book are not equivalent to 10 uses of 100 books.) Failure to recognise this really represents a low level of numeracy and it is a saddening reflection on our profession and on our journals’ referees that a review of the literature on obsolescence [3] revealed that most of the studies of library use or citations had failed even to consider whether the groups for which the use was counted were the same size or not. There has been little improvement since 1974. Similarly the evidence for exponential decay depends almost entirely on data from which the effects of growth (usually exponential) and updating have not been removed. Even when they have, fits to an exponential relation have been accepted as good enough without checks that linear, or other, relations do not fit as well or better. The old fallacy that correlation or a fitted regression implies causality remains overactive. When I have compared linear and exponential fits, the difference has usually been well within experimental error. I have not yet seen enough cases of exponential regressions fitting densities of basic use s ig nificantly better than linear to go beyond saying that the existence of a universal relation and its nature remain unproven. I am grateful to Kuch for making me think about my motivation: I hope that other workers will profit similarly from his article. J u A. Sandison 93 Ridgmount Gardens London WCIE 7AZ