College and Research Libraries Ranking and Evaluating the ARL Library Map Collections Charles A. Seavey This study measures and analyzes the map collections of the Association of Research Libraries (ARL) using data from 1984 and 1988. The nature of the cartographic format means that the size of a map collection can, within limits, be viewed as a measure of information content. Problems with the collected map data are noted. Size of collections and growth rates are computed and elements of change noted. A composite index, based on size and growth figures, is developed and used to rank the ARL map collections. Cartographic collection ranks are compared to ARL library index ranks and found to have a weak correlation. Directions for future research are suggested. • ilmer L. Hall has suggested that "Maps are often the stepchil- dren of libraries, receiving some attention at christening, but neglected henceforth." 1 'none sense this statement is true. The map does not fit on conventional library shelving, does not convey information in textual format, does not conveniently fit into any cataloging code (present or past), is rarely treated as a topic in library schools, and often appears to be a vexa- tious problem to the harried library ad- ministrator. Yet the map is an invaluable medium of communication, often im- parting in a single glance information that would require hours of reading2 Re- gardless of the problems, academic li- braries have long engaged in collecting and organizing maps, and today a collec- tion of materials in the cartographic for- · mat is common in many academic libraries.3 This article tests the hypothesis that map collections are stepchildren within Association of Research Library (ARL) institutions. Of course, no precise defini- tion of Hall's characterization of map collections as "stepchildren" is possible. However, for the purposes of this inves- tigation, it is assumed that if cartographic collections are stepchildren, then the ARL libraries will not expend the same amount of effort on collecting cartographic mate- rials as they do on collecting other for- mats. The assumption leads to the statement: If the ARL libraries place equal emphasis on collecting both carto- graphic material and noncartographic material, then rank-orders for both types of collections should be the same. In general, academic libraries have done little in the way of measuring or evaluating, in any quantitative sense, their collections of cartographic materi- als (here defined as maps, aerial photog- raphy, and remote sensing imagery). The ARL collects data on book, serials, and microform holdings but does not gather data for maps or other cartographic for- mats. Hence neither researchers nor li- brary administrators have much in the way of data, let alone agreed-upon norms on which to base possible methods Charles A. Seavey is Assistant Professor at the Graduate Library School at the University of Arizona, Tucson, AZ 85719 . 31 32 College & Research Libraries of evaluation for map collections. While the literature includes descrip- tions of individual map collections, or portions thereof, few studies attempt an overall description or analysis of more than one collection. Stanley D. Stevens surveyed twenty-six predominantly ac- ademic map collections in the western United States. His analysis dealt largely with personnel issues, although it did include some general observations on collections.4 David A. Cobb provided the first attempt to analyze the upper eche- lons of map collections in this country.5 His analysis was descriptive in nature and limited to the largest map collec- tions in various categories. The second edition of the Guide to U.S. Map Resources contains a similar analysis.6 Beyond these first analyses no significant at- tempts to compare map collections exist. The present study measures ARL map collections. This study assumes that size of collection is a strong indicator of the information content of the collection. A long-standing, almost unspoken, as- sumption in libraries is that a collection with more information content is of higher quality than one with lower infor- mation content. This assumption needs further explication. The notion that collection size equals quality is implicit in the ARL library index and is discussed in the data section of this article. Widespread anecdotal ev- idence indicates that most librarians equate collection size with collection quality. Susan A. Cady flatly states, "The quality of a research library is still measured primarily by the size of its holdings" (italics in the original).7 The equation of size with quality has come under a great deal of scrutiny in recent years. In A Planning Process for Public Libraries and in Output Measures for Public Libraries, the Public Library As- sociation (PLA) has specifically rejected the notion.8 The Association of College and Research Libraries (ACRL), in Mea- suring Academic Library Performance, has produced a system conceptually similar, in many ways, to that of the Public Lf- brary Association.9 The Association of Research Libraries, however, still bases January 1992 its measures and ran kings largely on size of the collections being measured.10 In the context of large research librar- ies, the argument can be made that size of book and serial collections equals in- formation content and quality. Undeni- ably, larger collections contain more information than smaller ones. While the ARL measures, when compared with the PLA and ACRL methods, are simplistic in terms of library interaction with its user community, they do, in a fashion, address the notion of what a research library is all about. In evaluating a collection of carto- graphic materials, the argument that size of collection equals information content is even stronger than a similar argument in- volving books and serials. Cartographic items are unique, or nearly so, in their information content. The size of most car- tographic collections is largely determined by their holdings of large-scale topo- graphic maps. It requires, for instance, 57,401 individual maps to cover the con- tiguous 48 states of the United States in the standard series of U.S. Geological Survey (USGS) 1:24,000 topographic quadrangles. USGS has map series at 1:50,000, 1:100,000, 1:250,000, 1:500,000, and 1:1,000,000 that either will, or do, cover the entire United States. Virtually all other nations have mapping projects employing a similar, or greater, range of scales. Such topographic maps are often used as base maps for thematic maps, such as soil or geologic maps, which are available at various levels of scale, or generaliza tion. 11 Each topographic map is a unique item. If the user desires a large-scale topo- graphic map of an area in northwest Albu- querque, New Mexico, there is only one choice.12 No alternative interpretations of the same body of information exist. A map, unlike subjects presented in book or article form, does not have a Marxist, or deconstructionist, or feminist alterna- tive explanation. Maps can differ in scale or time of situation, but those features are also items of unique information con- tent. While thematic, or subject, maps will occasionally have alternative expla- nations, these are in the minority. The great bulk of cartographic items (topo- graphic maps, or thematic maps based on topographic quadrangles) in a given collection are, therefore, unique and complete additions to the information content of the collection. Other informa- tion formats present overlapping amounts of information content. The unique contri- bution of an individual book to the knowl- edge base is considerably less than the contribution of an individual map. Therefore, the claim that larger map collections contain, in absolute terms and assuming little or no duplication, more information content than do smaller ones is valid. A study that ranks cartographic collections according to variables associ- ated withsizeandgrowthfigurescan provide &JJ.rea:mparative data about cartographic infonnationcontentwithin the ARL libraries. Within the context of the ARL libraries, if one accepts the idea that higher information content equals a higher-quality collection, then quality can be measured. THE DATA The ARL, as noted above, does not collect data on cartographic holdings. However, the Map and Geography Round Table (MAGERT) of the Ameri- can Library Association (ALA) has pro- duced two editions of the Guide to U.S. Map Resources, in 1986 and 1990Y The data presented are for calendar years 1984 and 1989, respectively. Preparations for the first-edition of the Guide started in 1983 and involved a large group of expert map librarians.14 The data-collection instru- ment went through several drafts.15 As data were collected, they were subject to review from a group of nineteen regional editors, chosen for their knowledge of their respective areas, and a final review by the compiler of the Guide. This pro- cess was repeated for the second edition. While no data-collection process is per- fect, the data are sufficiently reliable for the variables and methodology em- ployed in this study. Where weaknesses exist, they are in the original data re- ported by the various map collections. For example, size of collection (sheet count) is often expressed in figures end- ing in four or five zeroes (e.g., 40,000 or ARL Library Map Collections 33 200,000) which implies less-than-precise knowledge on the part of the reporting collection. THE POPULATION The cartographic holdings of eighty- eight ARL libraries were considered. The A~L population is actually higher, but the Guide does not report data on Cana- dian institutions. Further, some Ameri- can institutions reported no data, and two collections (Howard and George- town) are so atypical that they were re- moved from the study.16 The eighty-eight ARL libraries in this study have among them 169 map collections. All collections associated with a given institution were counted on the assumption that they were all open for use by the libraries' public. Cartographic collections are often sepa- rated. Typically there might be collections in the main library, a geology or science collection, and a hiStorical collection. VARIABLES COLLECTED The following variables were extracted from the two editions of the Guide. 1. Total sheet count. Traditionally, map collections have been counted by the number of individual map sheets held in the collection. While numerous maps are parts of sets, or series, individual sheets form separate bibliographic and infor- mation-bearing entities. Counting individual sheets does not address issues of multiple-copy teaching sets (which are in the collection at Louisiana State University, for in- stance) or the geographic areas covered by the collection. 2. Total count of aerial photographs. Aerial photography is both the basis for much map making and heavily used as a supplemental information source in map collections. A consid- erable amount of time can pass be- tween editions of topographic maps for rural and lightly populated areas. Aerial photography is often available from various sources much more fre- quently. While some overlap exists among adjacent aerial photographs, this duplication is necessary for 34 College & Research Libraries January 1992 TABLEt RANK, BY SIZE OF MAP COLLECTION, 1989 Top Twenty University No. of Maps UCLA 700,780 Harvard 520,000 Indiana University 510,575 U.C. Berkeley 489,985 University of Illinois, Urbana 444,774 Louisiana State University 402,200 Yale 385,500 U.C. Santa Barbara 375,000 University of Florida 373,854 University of Tennessee 359,199 Pennsylvania State University 336,000 University of Chicago 330,100 University of Georgia 326,850 University of Texas, Austin 323,163 University of Minnesota 319,123 Princeton University 290,914 Kansas University 281,048 University of Michigan 242,900 University of Washington 232,426 University of Oregon 230,000 Mean= 177,303 STD = 132,548 detailed interpretation of the infor- mation presented. Each photograph, therefore, is counted as a separate information-bearing item, although the bibliographic picture is consider- ably more confused. 3. Total count of remote sensing im- ages. Data on holdings in remote sensing imagery were collected from the 1990 edition of the Guide only. The earlier edition had not collected data on remote sensing imagery. Remote sensing imagery is just starting to become available to the ARL libraries and represents a new field of cartographic informa- tion. The current convention follows the model established in counting aerial photography: each image is a separate information-bearing item. Bottom Twenty University No. of Maps Iowa State University 91,033 Washington University (St. Louis) 89,112 Temple University 88,450 University ofPittsburgh 86,457 Mass. Institute of Technology 83,004 Brown University 75,000 University of Cincinnati 74,930 Colorado State University 34,799 Case Western Reserve University 32,500 U.C. Riverside 30,000 North Carolina State University 24,230 Wayne State University 22,000 Rice University 21,000 Miami University 17,621 Tulane University 15,000 Washington State University 15,000 Boston University 10,500 University of Southern California 8,500 Rutgers 3,150 U.C. Irvine 660 DATA MANIPULATION The data were entered onto creation sheets and then transferred to the Quattro Pro spreadsheet program, which was used for all further data analysis. An initial printout was reviewed and various anomalies identified. Where anomalies existed, correspondence was initiated to ascertain the cause of the problem. Prob- lems encountered included: 1. Decrease in the size of the collec- tion. All institutions showing a de- creasewererontacted for explanations. The decreases, in general, may be attributed to two causes. First, mul- tiple collections were consolidated between the two editions of the Guide, and the institution eliminated duplicate copies. Boston University, ARL Library Map Collections 35 TABLE2 RANK, BY INCREASE IN MAP HOLDINGS, 1984-89 Top Twenty University Growth University of Texas, Austin 116,163 University of Tennessee 101,753 UCLA 94,233 University of lllinois, Urbana 83,044 University of Maryland 80,500 U.C. Santa Barbara 75,000 University of Alabama 63,800 U.C. Berkeley 56,415 University of Chicago 55,100 Duke University 52,900 University of Arizona 50,437 Emory University 50,000 TexasA&M 48,792 University of Connecticut 47,000 U.C. San Diego 45,000 Pennsylvania State University 41,000 S. Illinois University 40,253 Cornell University 38,400 Kansas University 37,716 Universi!Y_ of Florida 37,469 Mean = 18,766 SID= 32,500 for instance, consolidated three collections into two, eliminating duplication and decreasing their total sheet count. Second, the insti- tution counted their collection and discovered it to be somewhat smaller than the previously esti- mated figure. 17 2. Growth rates that appeared to be far beyond normal expectations. Some of the growth can be explained by better data collection. The 1990 Guide lists more collections for the University of Texas at Austin than the 1986 edition. Part of the Uni- versity of Tennessee's increase de- rives from consolidating the old geography department collection into the new main library there. Bottom Twenty University Growth University of Southern California 1,000 Rutgers University 650 U.C. Irvine 60 Columbia University 0 University of Oklahoma 0 Vanderbilt University 0 Boston University -1,500 University of Colorado -5,201 SUNY Albany -8,000 Virginia Polytechnic -9,900 University of Virginia -11,750 University of Michigan -17,100 Case Western Reserve -17,500 Wayne State University -20,000 Ohio State University -22,461 Washington State University -25,000 University of Cincinnati -35,070 University of Oklahoma -54,820 Princeton University -66,086 Louisiana State Universi!Y_ -99,300 DESCRIPTIVE RESULTS After gathering the supplemental infor- mation and correcting the spreadsheet, in- stitutions were ranked according to the following variables. 1. Total map sheet count, 1991 2. Absolute increase in map sheet count, 1984--1989 3. Percentage increase in map sheet count, 1984--1989 4. Total cartographic holdings (maps, aerial photos, remote sensing im- ages), 1989 5. Absolute increase in cartographic holdings, 1984-1989 6. Percentage increase in carto- graphic holdings, 1985-1989 7. A final, composite ranking, ex- 36 College & Research Libraries January 1992 TABLE3 RANK, BY PERCENTAGE INCREASE IN MAP HOLDINGS, 1984-89 Top Twenty University %Growth University of Miami 1570 U.C. Riverside 233 N. Carolina State 193 University of Maryland 111 Emory University 100 Duke University 85 TexasA&M 59 University of Texas, Austin 56 University of Connecticut 47 University of Alabama 46 Temple University 43 University of Tennessee 39 Rice University 36 University of Hawaii 35 Arizona State University 33 University of New Mexico 31 University of Arizona 30 University of Utah 29 U.C. San Diego 29 Rutgers University 26 plained below, composed of vari- ables 1-6 Absolute increase is an indicator of the willingness, and ability, of the library to acquire new material. Collections that score high in these categories are work- ing to acquire new material (informa- tion) and to keep current as new maps and aerial photographs are produced. Percentage increases tend to reward the smaller collections, since they are starting from a smaller base. Collections that score high in these categories are being more active in acquisition. It can be argued that smaller collections are aware of a small information base and are working on improving that base. Table 1 presents the top and bottom twenty collections in terms of holdings of maps in 1989. Table 1 does not present the same rankings in the second edition of the Bottom Twenty University %Growth University of Oregon 4 University ofWlSCOnsin, Madison 3 Brown University 1 Harvard University 0 Columbia University 0 University of Oklahoma 0 Vanderbilt University 0 University of Virginia -5 University of Michigan -6 Virginia Polytechnic Institute -7 Boston University -12 Colorado State University -13 Princeton University -18 Ohio State University -18 Louisiana State University -19 University of Cincinnati -31 Oklahoma State University -33 Case Western Reserve -35 Wayne State University -47 Washington State Universi!_y -62 Guide. Because of additional requests for information made for this article, more complete data are presented here than in the Guide. Table 2 presents the top and bottom twenty collections by absolute increase in map holdings, 1984-1989. As mentioned above, the increases in holdings for both the universities of Ten- nessee and Texas may be more related to data-collection issues than actual in- creases in size. Table 3 presents the top and bottom twenty collections by per- centage increase in map holdings, 1984- 1989. The University of Miami has an ex- tremely small collection (17,621 sheets), and the University of California, River- side, is only 30,000 sheets. After those two, larger collections appear in the top twenty growth collections. Interestingly, some of the ' large collections, such as ARL Library Map Collections 37 TABLE4 RANK, BY TOTAL CARTOGRAPHIC HOLDINGS, 1989 Top Twenty Universit~ Holdings U.C. Santa Barbara 4,075,000 UCLA 839,471 U.C. Berkeley 660,437 University of Oregon 628,000 Cornell University 595,400 University of Illinois, Urbana 592,614 University of Florida 548,716 University of Georgia 547,250 Indiana University 521,425 Harvard University 520,000 University of Minnesota 487,929 University of Hawaii 478,259 Louisiana State University 472,225 Yale University 385,503 Kansas University 376,698 University of Tennessee 359,199 University of WISconsin, Madison 344,324 Washington University (St. Louis) 339,112 University of Chicago 339,100 Pennsylvania State University 337,336 Mean = 252,098 STD = 439,190 Harvard and Michigan, are static or ac- tually shrinking. Particular attention has been paid to the size and growth of the map portions of the collections. The map remains the primary information carrier and the largest proportion of most cartographic collections. Tables 4 and 5 are based on total cartographic holdings. Here the ef- fect of collecting aerial photos and re- mote sensing imagery is apparent. The more technologically advanced collec- tions rank higher in these listings. The University of California, Santa Bar- bara, has a large map collection, but the large difference between it and second- ranked UCLA results directly from U.C., Santa Barbara's large (2.5 million) aerial photo collection and its 1.2 million re- mote sensing images. Washington Uni- Bottom Twenty University Holdings Vanderbilt University 100,000 Ohio State University 96,000 Temple Uni.versity 88,450 University of Pittsburgh 86,457 Mass. Institute of Technology 83,004 Brown University 75,000 University of Cincinnati 74,930 Colorado State University 34,899 U.C. Riverside 32,500 Case Western Reserve 32,500 N. Carolina State University 24,230 Wayne State University 22,000 Rice University 21,400 University of Miami 17,621 Tulane University 15,000 Washington State University 15,000 Boston University 11,050 University of Southern California 8,500 Rutgers University 3,170 U.C. Irvine 1,180 versity in St. Louis is a bottom-twenty map collection, but the library there has acquired 250,000 remote sensing images to bring its collection into the top twenty in terms of total cartographic holdings. The earlier comments about the growth rates of U.C., Santa Barbara and Washing- ton University apply in table 6 as well. As noted previously, the University of Miami has a very small collection, but some of the other large-growth collections are far from small. U.C., Santa Barbara's ranking is particularly impressive. Rank by percentage increase highlights the collections that are inactive or simply not collecting or reporting data. OVERALL RANKING The overall ranking was computed by assigning a rank to eacn institution for 38 College & Research Libraries January 1992 TABLES RANK, BY INCREASE IN CARTOGRAPHIC HOLDINGS, 1984-89 Top Twenty University Growth U.C. Santa Barbara 3,475,000 Cornell University 434,400 University of Hawaii 368,659 Washington University (St. Louis) 255,896 UCLA 222,500 Purdue University 179,500 Kansas University 133,166 University of Alabama 131,400 University of Texas, Austin 120,167 U.C. Berkeley 114,376 University of Tennessee 101,752 University of Illinois, Urbana 86,570 University of Maryland 80,490 University of Oregon 73,823 University of Georgia 65,150 University of Connecticut 62,999 University of Chicago 54,600 Duke University 52,950 Emory University 49,999 Universi!Y of Arizona 49,661 Mean = 73,945 STD = 366,410 each of the six variables presented above. A score was then assigned based on the reciprocal of that rank, using 100 as a base. In other words, if a collection ranked first on a given variable, it re- ceived 99 points. If the collection ranked 30th, it received 70 points (100-30), and so forth. The overall ranking was com- puted by adding the six variable scores and dividing by six. A perfect score would have been 99.18 The numerical scores obtained by this process are statis- tically meaningless. They serve only to present a rank order of the collections and hence are not reported. Any ranking system will have its crit- ics. This one attempts to reward collec- tions that have a large information content (i.e., large holdings), are active in acquiring new holdings, and are ac- Bottom Twenty University Growth Rutgers University 669 U.C. Irvine 579 Columbia University -1 University of Oklahoma -1 Vanderbilt University -1 Boston University -1,450 Colorado State University -5,102 SUNY Albany -8,001 University of Virginia -11,750 University of Michigan -16,500 Case Western Reserve -17,501 Brigham Young University -18,914 Wayne State University -20,020 Ohio State University -22,462 Washington State University -25,030 University of Cincinnati -37,570 Va. Polytechnic Institute -43,400 Oklahoma State University -56,738 Princeton University -66,086 Louisiana State Universi!Y -89,275 tive in acqu1rmg technologically ad- vanced cartographic formats. The rank- ings are based on the theory that size equals information content and, per- haps, quality. The rankings do not tell us anything about usage, effectiveness of collection development, or other vari- ables explicitly addressed by the PLA and ACRL systems mentioned above. TESTING THE HYPOTHESIS Readers familiar with the ARL com- posite rankings will note that the rank- ings presented in table 7 differ considerably from those published by the ARL. Some very highly ranked ARL collections fair poorly in this study. The University of Wisconsin, Madison, for instance, which ranks 13th in the ARL, ranks 58th in this study. The top-rated collection in this ARL Library Map Collections 39 TABLE6 RANK, BY PERCENTAGE INCREASE IN CARTOGRAPHIC HOLDINGS, 1984-89 Top Twenty University %Growth University of Miami 1,568 U.C. Santa Barbara 579 University of Hawaii 336 Washington University 307 Cornell University 269 U.C. Riverside 256 North Carolina State 193 Purdue University 119 University of Maryland 111 Emory University 99 U.C. Irvine 96 University of Alabama 94 Duke University 85 University of Connecticut 62 TexasA&M 58 University of Texas, Austin 57 Kansas University 54 Temple University 43 Arizona State University 40 University of Tennessee 39 .. -o means growth was negative, but less than 1%. study, the University of California, Santa Barbara, is ranked 46th by ARL. This study assumes that size of collec- tion is a strong indicator of the information content of the collection. The basic issue being investigated in this study is whether cartographic mate- rials are stepchildren in ARL libraries. The method employed is to develop a ranking of ARL cartographic materials collections that is similar conceptually to the familiar ARL rankings largely based on size and growth rate of book and serial holdings. If cartographic collec- tions receive the same emphasis in col- lection development enjoyed by book and serial collections, the ARL rankings Bottom Twenty University Brown University Univ. of Wis., Madison Harvard University ·columbia University ·university of Oklahoma ·vanderbilt University University of Virginia University of Michigan Brigham Young University Boston University Colorado State University Louisiana State University Princeton University Ohio State University Oklahoma State University Va. Polytechnic Institute University of Cincinnati Case Western Reserve Wayne State University Wa. State University %Growth 1 -0 -0 -0 -5 -6 -8 -11 -12 -15 -18 -18 -25 -26 -33 -35 -47 -62 and the ranking produced by this study should be closely related. The statistical technique selected to measure the relationship between the two rankings is Spearman's rank order coefficient, or rho. Spearman's rho (rs) is commonly used to compare matched- pair rankings. 19 Spearman's rho requires that each set of ranks be ordinal-level data, so the ARL library index list was matched to the eighty-eight institutions considered in this study. That is, they were ranked 1-88, rather than being ranked within the total ARL population. Spearman's rho produces a correla- tion coefficient that measures the strength of the relationship between the two rankings. Rho can vary from -1 to + 1, where 1 expresses a perfect negative or positive correlation. One statistician has suggested the following guidelines 40 College & Research Libraries January 1992 TABLE7 RANK, BASED ON COMPOSITE SIZE/GROWTH RANKINGS, 1989 Rank University Rank University 1 U.C. Santa Barbara 39 University of Kentucky 2 University of Texas, Austin 40 University of Nebraska 3 UCLA 41 University of Massachusetts, Amherst 4 University of Tennessee 42 Dartmouth University 5 Cornell University 43 Yale University 6 University of Illinois, Urbana 44 U.C. Davis 7 University of Alabama 45 University of Miami 8 U.C. Berkeley 46 Michigan State University 9 University of Hawaii 47 Northwestern University 10 Kansas University 48 North Carolina State 11 University of Chicago 49 Harvard University 12 University of Maryland 50 Florida State University 13 University of Arizona 51 Johns Hopkins University 14 Purdue University 52 North Carolina, Chapel Hill 15 U.C. San Diego 53 University of Missouri 16 University of Connecticut 54 Brigham Young University 17 Southern Illinois University 55 Iowa State University 18 University of Georgia 56 University of Pennsylvania 19 Pennsylvania State University 57 University of Wisconsin, Madison 20 TexasA&M 58 SUNY Buffalo 21 Duke University 59 University of Michigan 22 University of Florida 60 University of Delaware 23 Arizona State University 61 Rice University 24 University of Iowa 62 Syracuse University 25 Stanford University 63 SUNY Stony Brook 26 Emory University 64 Notre Dame 27 University of New Mexico 65 University of Pittsburgh 28 University of Utah 66 Rutgers University 29 University of Washington 67 Louisiana State University 30 Washington University (St. Louis) 68 University of Colorado 31 University of Minnesota 69 Columbia University 32 Indiana University 70 University of Virginia 33 University of South Carolina 71 Princeton University 34 University of Oregon 72 Tulane University 35 Temple University 73 U.C. Irvine 36 U.C. Riverside 74 Massachusetts Institute of Technology 37 Georgia Tech 75 University of Oklahoma 38 Kent State University 76 University of Southern California ARL Library Map Collections 41 TABLE7 RANK, BASED ON COMPOSITE SIZE/GROWTH RANKINGS, 1989 (cont.) Rank University 77 Brown University 78 Vanderbilt University 79 Oklahoma State University 80 Virginia Polytechnic Institute 81 Colorado State University 82 Ohio State University for interpreting values between 0 and 1: r value Interpretation <.20 Slight, almost negligible relationship . 20-.40 Low correlation, definite, but small relationship .40-. 70 Moderate correlation; substantial relationship .70-.90 High correlation; marked relationship .90-1.0 Very high correlation; very de- pendable relationship 20 The hypothesis being tested involves the strength of the relationship between the two rankings. Statistical significance is not an issue in this situation because sampling error is not involved. The ques- tion is: Is the observed rs strong enough to dispute the idea that cartographic materi- als are stepchildren in the matter of size and growth of the collection? The ARL rankings and the rankings in this study were found .to correlate at rs = 0.31. This means that there is a positive relationship between the two ranking systems. The strength of the relation- ship, however, falls in the low, or weak category in the scheme noted above. 21 An observed rs of 0.31 suggests that the re- lationship between the ARL rankings and those obtained in this study is low. This suggests that little consensus about the importance of the cartographic for- mat exists within the ARL libraries. Some of the libraries at the top of the ARL list-Texas, UCLA, Cornell, and the University of Illinois, Urbana, for exam- ple-seem to place cartographic materi- Rank University 83 Boston University 84 Case Western Reserve 85 University of Cincinnati 86 Wayne State University 87 Washington State University als on approximately the same level of importance as other formats. Other top ARL collections-Yale, Harvard, and Columbia being the notable examples- do not seem to value cartographic mate- rial as much as other formats . SUMMARY AND CONCLUSIONS The map collections of the ARL librar- ies have been measured and ranked ac- cording to a composite index based on size and growth figures for both maps and other cartographic formats. The rankings reflect not only absolute collec- tion size, but also the rate of increase in all cartographic formats. The rankings should provide a useful measure for comparisons among ARL libraries inter- ested in one measure of their collections and also provide possible benchmark data for rton-ARL libraries to consider. Given the low correlation between ARL rankings and rankings in this study, it seems that Wilmer Hall's comment about map collections as stepchildren is as correct now as it was in 1925. It is hoped that this study will prompt ARL chief collection development offi- cers to discuss the cartographic format. Clearly, at a number of institutions the relationship between effort expended in collection development for books and serials and that for cartographic materi- als is insignificant. Equally clear from this study is that individuals in charge of map collections at ARL institutions have work to do in improving the nature of the data they are reporting. As was noted above, such basic elements as collection size appear to be estimates in all too many cases. Other data elements need similar atten- 42 College & Research Libraries tion. Reporting on number of staff was so inconsistent that the editor of the Guide considered eliminating the cate- gory from the published version. 22 The ARL libraries fare no better than any others in this category. The collection index is limited concep- tually in that it does not address issues of user interaction with the collection or the relationship of the cartographic col- lection with the rest of the library or parent institution. Because of limits im- January 1992 posed by institutional reporting of data, more sophisticated measures, such as those suggested in Output Measures or Measuring Academic Library Performance, cannot be generated for all ARL map collections at this time.23 Future studies drawn from data in the two editions of the Guide will attempt to develop mea- sures related to user-collection interac- tion, facilities, and collection-institution interaction for the subset of ARL librar- ies with sufficiently complete data. REFERENCES AND NOTES 1. Wilmer L. Hall, "A Classification for Maps," Library fournal50:257-59 (Mar. 15, 1925). 2. Mark Monmonier and George A. Schnell, Map Appreciation (Englewood Cliffs, N.J.: Prentice Hall, 1988), p.3-5. 3. For instance, Map Collections in the United States and Canada (New York: Special Libraries Association, 1978) listed 345 map collections in academic institutions. Both editions of the Guide to U.S. Map Resources contain considerably more academic map collections than that. Of special significance to the spread of map collections has been the inclusion of U.S. Geological Survey and other federal mapping agency products into the U.S. Government Printing Office depository system in the mid-1980s. 4. Stanley D. Stevens, "Map Librarians hip-Today and Tomorrow," Drexel Library Quar- terly 9:3-14 (Oct. 1973). 5. David A. Cobb, "Map Librarianship in the U.S.: An Overview," Wilson Library Bulletin 60:14-16 (Oct. 1985). The Wilson Library Bulletin piece was largely based on Cobb's "Introduction," Guide to U.S. Map Resources, 1st ed. (Chicago: American Library Assn., 1986), p.ix-xvi. 6. David A. Cobb, ed., "Introduction," Guide to U.S. Map Resources, 2d ed. (Chicago: American Library Assn., 1990), p.ix-xvii. 7. Susan A. Cady, "The Electronic Revolution in Libraries: Microfilm Deja Vu?" College & Research Libraries 51:374-86 (July 1990). 8. Vernon C. Palmour and others, A Planning Process for Public Libraries (Chicago: Amer- ican Library Assn., 1980); Douglas L. Zweizig and Eleanor Jo Rodger, Output Measures for Public Libraries (Chicago: American Library Assn., 1982). 9. Nancy A. Van House, Beth T. Weil, and Charles R. McClure, Measuring Academic Library Performance: A Practical Approach (Chicago: American Library Assn., 1990). 10. The ARL index is based on the following variables: number of volumes held, number of volumes added during the previous fiscal year, number of current serials, total expenditures, and size of staff. For a brief description of the ARL index, see ARL, ARL Statistics, 1982-83 (Washington, D.C.: ARL, 1984) p.27-28. For a complete description see Kendon Stubbs, The ARL Library Index and Quantitative Relationships in the ARL (Washington, D.C.: ARL, 1980). ARL composite rankings used in this study are drawn from The Chronicle of Higher Education, "Almanac" supplement, 37:27 (Sept. 5, 1990). 11. Charles A. Seavey, "Collection Development for Government Map Collections," Gov- ernment Publications Review, 8A:17-29 (1981). 12. U.S. Geological Survey, Los Griegos, New Mexico, 1:24,000 scale topographic map (Reston, Va.: The Survey, 1960, photorevision, 1972). 13. David A. Cobb, Guide (1986, 1990). 14. Cobb, Guide (1986), p.x. 15. The instrument is included as an appendix in both editions of the Guide. 16. Both collections have fewer than 1,000 maps. However, they are both in the Washing- ton, D.C., area and have access to the largest concentration of cartographic holdings in the country: the collections of the Library of Congress and the National Archives. ARL Library Map Collections 43 17. An ARL collection under the author's care went from an estimated size of 120,000 sheets to an actual count of 85,061. This is not an uncommon phenomenon, which certainly indicates that the ARL libraries, and probably all collections, need to establish better record-keeping procedures than currently employed. 18. The ARL index used to produce overall rankings based on the variables noted in note 10 is considerably more complex in construction than the method employed here. The ARL has had years of experience in collecting and refining the reliability of its data. In this researcher's judgment the data presented in the Guide are simply not reliable enough to be subjected to the rigorous statistical manipulations employed by the ARL. 19. Richard C. Sprinthall, Basic Statistical Analysis, 2d ed. (Englewood Cliffs, N .J.: Prentice Hall, 1987), p.200-202. 20. Joy Paul Guilford, Fundamental Statistics in Psychology and Education, 3d ed. (New York: McGraw-Hill, 1956), p.145. 21. Interpretation of correlation tests is subject to considerable debate. Obtaining a statis- tically significant correlation with Spearman's rho, or the more rigorous Pearson's r, is largely a function of the size of N and need only be employed when samples, rather than populations, are being tested. The interpretation of the strength of relationship presented here is a conservative one. Guilford's scheme is usable, but problematic, particularly in the middle interval. Given the data employed in this study and general approach to correlational statistics, this researcher would not describe correlations as "strong" until values above rs = .75 are obtained. Others are free to interpret the obtained value as they choose. 22. The 1990 edition of the Guide, in both the published version and a prepublication version in the possession of this author, carries statements to the effect that staff data were not included because of reliability problems (see page xv, published version). The published version does report staffing levels for most collections. 23. For instance, many institutions either do not circulate cartographic materials or fail to collect (or report) such data if they do.