College and Research Libraries To the Editor: In the January 1972 CRL Brigitte Ken- ney of Drexel University volunteers to send me a list of "money-saving automated library operations." This implies that she has studied and approved detailed cost analyses of these operations. Let her pre- sent just one such analysis in these pages so we can see exactly what she takes to be valid cost analyses. Frankly, I think she is talking through her professorial hat. Ellsworth Mason Director of Libraries University of Colorado, Boul.der, Colorado To the Editor: Three cheers for Roger Horn's "Think Big: The Evolution of Bureaucracy" ( CRL, Jan. 1972). It is a devastatingly accurate exposure of the "numbers game" as it is currently played in the research libraries and their would-be imitators. There will un- questionably be howls of protest from the establishment. The truth hurts. Paul DuBois Assistant Director Kent State University Libraries Kent, Ohio To the Editor: I suspect CRL's January editorial, "Li- braries for Decision-Makers"-with which I agree wholeheartedly-was prompted in part by Herbert M. E. Pastan' s depressing letter published in the same issue. The point is that usually the library has not demonstrated that it is essential to the op- 408 I Letters eration of a business organization or an ed- ucational institution. As a librarian, I do~'t like to admit it, but it seems that, in general, libraries and li- brarians are not as important as we'd like them to be. When money is tight, · library budgets-which typically amount to less than 5 percent of total institutional expend- itures-are often among the first to be cut. Currently, the principal mode of teaching! learning, i.e., the lecture-textbook metl1od, makes minimal use of academic libraries. If space is needed for additional classrooms, library reading areas are frequently appro- priated by the decision makers. And when library buildings are being planned, campus librarians are not always consulted. Perusing L. Carroll DeWeese's article on status and professionalism in the same issue of CRL has caused me to reexamine my ideas. While I have been an ardent support- er of faculty status for librarians-and con- tinue to be-I now feel that we have gen- erally been "placing the cart before the horse." It is unrealistic for most librarians to expect higher standing for themselves while the library is not of greater impor- tance on campus. Only when library service is commonly considered the equal of teach- ing will academic librarians throughout the country achieve the status they are seeking. To be judged a high-priority concern, the library must prove that it is vital to the ex- istence of the institution-that an organiza- tion will fail without one. Campus leaders must be shown that academic libraries can supply them with the information they re- quire-even before the need for informa- tion is fully perceived-at a reasonable cost. Through such current-awareness ser- vices as selective dissemination of informa- tion, librarians can bring new material to the attention of decision-makers before the desire for the specific items arises. We must be more successful as a profes- sion in reaching the campus ]eaders. As Pastan puts it, "We certainly won't sell the case for librarianship talking to each other at ALA meetings." Leonard Grundt Director Nassau Community College Library Garden City, New York To the Editor: It was with some consternation that I read your little editorial in the March 1972 issue. I suppose at this stage in life I should not be surprised at the continued examples of administrative and budgetary theft that are so often at hand. I do not know how else to describe the failure to use fees for their implied purpose. You ask for reader re- sponse. I would be most unhappy if CRL was to cease or be curtailed. While I would like CRL News to continue, I would not miss it to the same degree. I would think that in this day and age when we are faced with so many separate and separated library organizations to which we feel we should belong, but from whose number most of us must choose a few, that the central authorities of ALA will not increase the temptation to not renew membership in ALA by removing from our hands CRL. I think the issue is that specific, and should be brought to the attention of the ALA authorities in as vigorous a man- ner as is possible. R. G. Bracewell Emmanuel College Librarian and Library Coordinator Toronto School of Theology Toronto, Canada Ed. note: Mr. Bracewell's letter appears to refer to the editorial in the March 1971 issue of CRL, rather than the issue indi- cated. To the Editor: I enjoyed reading Mr. Mcinnis' article (May 1972, p. 190-198) describing his use of regression techniques to investigate de- Letters I 409 terminants of library size; it seems that li- brary science can only profit from the intro- duction of tools that have proven them- selves so valuable for the other social sci- ences. Because of this, however, it is impor- tant that these techniques be presented as clearly and accurately as possible. Without commenting on substantive issues, such as the propriety of using a predictive equation based on the nation's finest research li- braries as a norm for college libraries in general, I do wish to clarify one technical point that recurs in Mr. Mcinnis' article and that may be confusing to readers. Mr. Mcinnis, referring to an earlier treat- ment of this problem by Reichard and Orsagh, states that their conclusions are ef- fectively invalidated since the "size vari- ables are patently not (statistically) inde- pendent of each other." This statement and others about the problem of multicolli~eari­ ty, can be misleading. Linear regression equations of the form y=a + bx + cx2, for example, often appear in the literature, al- though the "size variables" x and x2 are "patently not independent of each other." It is true that if the variables on the right side are statistically uncorrelated, the analy- ses may be simpler, but there is nothing in the theory that prevents one from using correlated variables. If two of these vari- ables are perfectly correlated in the sense that one completely determines the other, an investigator will not get incorrect results, but will rather find he gets no solution at all; that is, he won't be able to solve the equations. If the variables are strongly, but not perfectly, correlated he will get a solu- tion, but will find that the confidence inter- vals (or better, confidence ellipsoids) for the coefficients are larger than he might otherwise have expected. These statements are not the same, however, as saying that the conclusions reached are invalid. Mr. Mcinnis introduced the discussion of correlated size variables in an attempt to explain the negative coefficients of the Reichard and Orsagh paper. One can ob- tain an alternative explanation of these neg- ative regression coefficients by carefully considering what a regression coefficient is. A regression coefficient indicates how much the dependent variable will increase if a particular dependent variable increases while the other dependent variables remain 410 I College & Research Libraries • September 1972 constant. For the case at hand, if faculty size is fixed, an increase in the size of the student body probably represents a reduc- tion in the quality of the college; the "worse" a school, the fewer books its library will own, and hence the negative coeffi- cient. Thus student body may, in this in- stance, be more a measure of quality than size. This effect is eliminated in Mr. Mcin- nis' treatment since his sample includes only excellent schools. I would finally like to note that the equa- tion given is a linear approximation, in an extreme region, of what is most likely a nonlinear equation. This should be consid- ered if one wishes to apply this equation to schools of small or medium size. The negative constant term may be explained by this observation. Abraham Bookstein Professor Graduate Library School University of Chicago Chicago, Illinois