Assessing Diversity in Academic Library Book Collections: Diversity Audit Principles and Methods

1.1 Diversity Audits in Academic Libraries

As Johnson (2018) and others have noted, the foremost criterion for the evaluation of library collections is not necessarily quality, but utility: the extent to which they serve the purposes for which they are intended. In the university environment, utility is best represented by academic outcomes such as students’ course performance, graduation rates, and postgraduate success (e.g., Stemmer & Mahan, 2016). The impact of the library on academic performance is difficult to isolate and measure, however, so most libraries evaluate their collections through a range of indirect indicators that focus on content, use (including citation impact), and perceptions of utility (Clayton & Gorman, 2006; Johnson, 2018; Kelly, 2021; Saponaro & Evans, 2019). Collection analyses of this type can be used to:

• evaluate whether the library collection provides adequate support for particular academic departments, degree programs, faculty, and student groups; demonstrate to accreditors, donors, and others stakeholders that their needs are being addressed;

• facilitate the effective allocation of resources by identifying areas in which the collection is weaker (or stronger) than required;

• identify candidates for possible cancellation or weeding – information resources that are no longer needed, no longer cost-effective, less suitable than competitors’ products, or unsustainable with regard to cost, maintenance, or technological requirements;

• provide useful information for space planning, collaborative collection development, and other areas of library operations;

• help build greater knowledge of the collection among librarians and faculty in order to improve their effectiveness as selectors or to facilitate the refinement of collection policies and selection criteria;

• understand patrons’ perceptions of the collection and gauge whether those perceptions are consistent with reality; identify the areas in which outreach and marketing are needed.

Unlike other collection evaluations, diversity audits are seldom used to identify resources for potential cancellation. They tend to focus on the first two goals (which deal with support for particular programs and groups) and the last two (which deal with selection methods and public perceptions).

Most diversity audits of book collections reveal the percentage of titles that are associated with particular underrepresented or marginalized groups, allowing librarians to pinpoint the areas in which greater effort is needed. These investigations therefore provide valuable information even when the need to build more diverse collections is fully acknowledged by all stakeholders. Specifically, diversity audits can help us to understand the library collection more fully; evaluate whether the collection is aligned with the demographic characteristics of the student community; determine the extent to which improvement is required as well as the specific areas (topics, groups, etc.) that need to be addressed; establish diversity goals and timelines; measure progress toward those goals; demonstrate to outside audiences that deficiencies in the collection are being addressed; highlight the need for improvement when seeking outside funding, cooperation, or assistance; and help show how improvements in the collection can influence academic outcomes.

While many recent studies have documented academic libraries’ DEI initiatives, the most recent information on diversity-related collection assessment efforts was compiled more than two decades ago. A 2000 survey of 115 liberal art colleges revealed that 76% had undertaken some kind of collection assessment related to diversity (Winston & Li, 2000). However, the survey data do not indicate whether these assessments were formal or informal, whether current data would reveal the same level of commitment, or whether a similar emphasis on collection assessment can be seen at other types of colleges and universities.

1.2 Scope of the Review

This guide reviews the scholarly and professional literature on diversity audits of academic library book collections. (Journal collections are not considered here). It begins with a description of the literature search procedure (Section 1.3), then:

• discusses what is meant by diverse groups and diverse books, and how those concepts can be operationalized for use in a diversity audit (Section 2);

• describes three methods of conducting the diversity audit, their advantages and disadvantages, and their implications (Section 3);

• explains the importance of setting collection diversity targets and of following up on the initial investigation with repeated or complementary analyses (Section 4);

• discusses how the results can be used to build more diverse collections (Section 5);

• presents three fundamental questions that explore the limits of the methods currently in use, highlight underlying issues that librarians may want to consider when planning a diversity audit, and suggest avenues for further research and discussion (Section 6).

1.3 The Literature Search

The primary database searches for this review were conducted in March 2022. Table 1 shows the databases and the search terms that were used. The searches were intentionally broad, favoring recall over precision, on the assumption that it was better to examine and discard non-relevant results than to miss relevant items. The searches were not limited by date. Altogether, the nine database searches yielded 274 results. The combined list of results had just 211 items, however, because some works were found through more than one database.

Each of the 211 items was read and evaluated. The evaluations resulted in a list of 143 relevant items – items intended for inclusion in the literature review. More than 143 bibliographic entries are included here, however, since additional works were identified during the evaluation process (e.g., from the bibliographies of relevant papers) and in a second round of searches (March 2023) that focused on works published in 2022 and 2023. Further additions and exclusions were made during the writing process.

This review is meant to include all important works on diversity audits of academic library book collections. Works that frame diversity in terms of equity or inclusion are included. General discussions of diversity in academic library collections are included to the extent that they provide useful ideas or information about assessment. No restrictions are made with regard to geographic scope, although nearly all the research in this area has focused on US institutions. This review does not discuss diversity with regard to other aspects of librarianship (hiring and retention, staff development, outreach to patrons, bias in subject headings, etc.), other kinds of libraries, or resources other than (non-archival) books and book collections.

It is notable that most of the literature on this topic has appeared in journals, and only a few recent books include useful information. Of the edited volumes that discuss diversity in academic libraries (Dali & Caidi, 2021; Epstein, Smallwood, & Gubnitskaia, 2019; Jones & Murphy, 2019; Lee, Lym, Bryant, Cain, & Schlesinger, 2022), none cover collection development outside the context of special collections and archives. Of the four books that focus on diversity in library collections (Hibner & Kelly, 2023; Hughes-Hassell, 2020; Pattee, 2020; Voels, 2022), all are devoted mainly to public and school libraries. The diversity audit strategies presented by Pattee (2020) are also relevant to academic libraries, however, and one general guide to collection assessment in academic libraries (Kelly, 2021) describes many tools and methods that can be readily applied to diversity audits.

2.1 Defining Diversity

As shown in Table 2, diversity is commonly expressed in terms of demographic or socioeconomic characteristics, cultural traits, self-identification, group membership, physical or mental attributes, and the ways in which individuals are perceived by others. Some individual attributes – in particular, introversion/extraversion, learning styles, and other psychological traits unrelated to disability – are only now beginning to be recognized as aspects of diversity. Others, such as dialect or accent, are generally subsumed under ethnic or country-of-origin categories even though they vary considerably within each group.

All three methods of conducting a diversity audit (Section 3) require that we define the types or aspects of diversity we’re interested in – the groups for which the audit will generate results. These will not be the same for every library. In defining them and deciding which to focus on, we make choices about the aspects of diversity that are important to our institutions and our communities.

Six guidelines may be helpful. First, be specific when identifying or defining groups. Friebel (2019) and Voels (2022), among others, have stressed that labels such as BIPOC (black, indigenous, and people of color) hide important distinctions among groups, and that even widely used designations such as Hispanic encompass a wide range of cultural identities and traditions. Likewise, Alexander (2013) and Matteson (2017) emphasize the need to recognize the diversity of Native American tribes. “The view of Indians as a group is an exogenous perspective that developed with the colonization of the Americas. American Indians tend to identify themselves at the tribal level or at the extended family level” (Alexander, 2013, p. 63).

Second, use terms that are acceptable and meaningful to group members. This can be important when presenting the results to external audiences, especially with regard to terms such as Latinx. Although many scholars have adopted the term as a gender-neutral alternative to Hispanic or Latino, several recent surveys show that it is preferred by fewer than 3% of Hispanic Americans. In fact, 40% find it offensive, generally because it is not a Spanish word and did not originate within the Hispanic community (Torregrosa, 2021). In 2021, a major Latino civil rights organization decided to stop using Latinx due to lack of support for the term outside the academic community (Gamboa, 2021), and even some gender studies scholars have expressed the view that academics “should never impose social identities onto groups that do not self-identify that way” (Ochoa, 2022). (Although it is not strictly synonymous with Latino or Latina, Hispanic is the preferred term among those of Spanish or Spanish-indigenous ancestry in the United States; see Noe-Bustamante, Mora, & Lopez, 2020; Torregrosa, 2021). Likewise, the people first language adopted by many scholars in the field of disability studies has been challenged on the grounds that it reinforces negative perceptions and that it diverts attention from more important issues (Gomes, 2018; Wilder, 2006, pp. 9–11).

Third, consider the cultural and institutional contexts of the diversity audit, since the categories that are meaningful in one setting may be inappropriate in another (Kessler, 2020). For instance, the racial/ethnic designations used in the 2021 Census of England and Wales are very different from those used in the United States. The UK framework has no Hispanic or Latino/Latina category, it provides for multiple mixed-race designations, and it allows for the separate identification of Arab, Bangladeshi, Chinese, Gypsy/Irish Traveller, Indian, Irish, Pakistani, and Roma groups. Within the Black category, the Census also makes a clear distinction between African and Caribbean ancestry (Jensen et al., 2021; U.K. Race Disparity Unit, 2023). Context is also important in an institutional sense. For example, academic libraries might emphasize aspects of diversity that have special relevance to instruction, such as learning styles. They might also focus on groups with distinctive worldviews or on those that students are unlikely to be familiar with through their experiences outside the university.

Fourth, account for intersectionality, the interaction of multiple characteristics or group identities. For instance, the perceptions and experiences of black women are distinctive in ways that cannot be attributed simply to race or gender, but to the interaction of the two. As discussed in Section 3, the methods used to account for intersectionality will vary with the type of diversity audit undertaken. An audit based on library catalog searches might identify the subject headings or keywords that represent African Americans, those that represent women, and count not just the number of books in each category, but the number that can be found in both categories. Likewise, an audit based on the inspection of individual titles might include separate tallies for black men, black women, Hispanic men, Hispanic women, and so on. The need to account for intersectionality in diversity audits was recognized nearly 30 years ago and is perhaps even more important today (Bostic, 1995; Friebel, 2019).

Fifth, be aware that any categorization tends to channel our thinking along culturally approved paths. That is, we tend to think of diversity in terms of the groups that are well established, such as those that correspond to legally protected classes. Other aspects of diversity, such as height, weight, accent, voice pitch and timbre, urban/rural origin, introversion/extraversion, and preferred reasoning style (e.g., deductive or inductive) are thereby de-emphasized regardless of their actual impact. For example, facial dominance – a set of attributes including an angular face, a low forehead, small eyes, and a large chin – is not normally regarded as an aspect of diversity, yet it has a major impact on the career success of men in occupations such as law enforcement and the military (Adams, Mourtgos, Simon, & Lovrich, 2023; Mazur, Mazur, & Keating, 1984; Mueller & Mazur, 1996, 1997). Although we may not be able to incorporate the full range of relevant characteristics into a diversity audit, we should nonetheless acknowledge that there are important aspects of diversity not captured by the categories in general use.

Finally, determine in advance whether the investigation will identify diverse books in terms of characters, authors, topics, perspectives, or a combination of these. Diversity audits of fiction collections often focus on the characters’ or authors’ identities, and biographies are generally not difficult to classify. For other nonfiction works, it may be helpful to distinguish between diverse topics (i.e., books about diversity or about particular groups) and diverse perspectives (i.e., books that are not about diversity but that reflect the distinctive views or perceptions of marginalized groups with regard to topics other than diversity itself). While diverse topics often correspond to particular Library of Congress (LC) subject headings, diverse perspectives generally do not.

The natural sciences are sometimes difficult to evaluate, especially since bibliographies and checklists of diverse books tend to provide better coverage of the social sciences and humanities (Gonzalez, 2023; Kristick, 2020). Moreover, natural science specialists are sometimes excluded from the teams that conduct diversity audits (e.g., McKinzie, 1994), perhaps because the perspectives of underrepresented groups are difficult to discern in the sciences or because scientific findings are felt to be independent of the investigators’ personal characteristics. Kristick (2020) suggests undertaking separate diversity audits in the sciences, perhaps using methods different from those used elsewhere. For example, it may be useful to evaluate diversity of authorship in science, technology, engineering, and math fields even if the library’s more general focus is on topic and perspective rather than authorship.

2.2 Diversity of Authorship

Librarians have long recognized the importance of endogenous perspectives – of the need to represent marginalized groups from the perspectives of the group members themselves. This is especially true in fields such as Native American studies, where many historical accounts have adopted a strongly Eurocentric perspective (Alexander, 2013). The #OwnVoices movement, which emphasizes the contributions of authors and illustrators from marginalized and underrepresented groups, began in 2015 with a Twitter hashtag used to identify books by fiction authors “who openly shared the diverse identity of their main characters” (Lavoie, 2021). The term was later applied to nonfiction as well, and publishers began to include it in their marketing materials. Unfortunately, the designations used to describe #OwnVoices authors have not always been consistent with the authors’ own descriptions of their identities. For that reason, We Need Diverse Books, an organization that promotes diversity in children’s literature, stopped using the term in June 2021, and other organizations have followed suit (Lapointe, 2022; Lavoie, 2021).

Several researchers have described the challenges of evaluating authors’ characteristics on a title-by-title basis (Bird & McPherson-Joseph, 2020; Emerson & Lehman, 2022; Gibney et al., 2021; Wells, Gibney, Paris, & Pfitzer, 2023). Even when the assessment is limited to a few key indicators such as race, gender, and LGBTQ+ identity, a reliance on authors’ self-disclosures – the most reliable source of information – results in missing or incomplete data for many individuals. Other sources of evidence, such as photos, can provide more complete information, but with a higher error rate. Accounting for authors’ characteristics is simpler where children’s picture books are concerned, since scholars can rely on databases such as Diverse BookFinder for that purpose (Bates College, 2023; Fort, 2021).

Although diversity audits do not generally incorporate information on publishers, it may be worthwhile to consider who publishes books as well as who writes or illustrates them. The characteristics of mainstream publishing industry personnel, who are disproportionately white, may result in biases against certain kinds of diverse books (Jiménez & Beckert, 2020). At the same time, independent publishers who focus on diversity are especially likely to produce books that counteract any such biases (Edinger, 2023; Morales, Knowles, & Bourg, 2014; O’Neal, Curé, & Peick, 2020; Phelps, 2021; Ryan, 2022).

3.1 The Catalog Search Method

The catalog search method is straightforward. In its simplest form, it has just four steps:

1. Identify subject headings, call numbers, and/or keyword phrases that correspond to the topics and groups of interest.

2. Using those subject headings, call numbers, and keywords, construct a search string for each topic or group.

3. Search the library’s book catalog using those search strings – one search for each relevant topic or group – with appropriate limits/filters for date, language, item type, and location.

4. Count the number of titles held by the library in each area – the number retrieved by each search.

The catalog search method is often used to generate comparative statistics. That is, the same method can be used with several different libraries (several different catalogs) to see how each library compares with the others.

Instead of simply counting the number of search results, we can also download each set of results to a spreadsheet (i.e., download relevant fields from the set of retrieved records). This allows for several options that would not be feasible otherwise. First, we can clean the data by manually removing titles that should not count toward the library’s holdings of diverse books; we can remove earlier editions of later works, scan for books with titles that don’t seem related to any aspect of diversity, or identify and exclude books that include false or misleading information. (Ideally, the rules for any oversight procedures will be established before the searches are conducted). Second, we can evaluate the degree of overlap among the various sets of search results in order to explore intersectionality and to identify the diversity-related topics that are most often associated with particular groups; we can count the number of books that focus on black women or the number that cover educational discrimination among individuals with disabilities, for instance. Third, we can compile and evaluate descriptive data (author, publisher, price, etc.) and use statistics for the items retrieved by each search.

Although the catalog search method is the most straightforward of the three diversity audit methods, it relies heavily on the subject headings, call number ranges, and keywords used to construct the searches. Users of this method must have confidence in the available search mechanisms. Unfortunately, LC subject headings represent only those concepts that are well established in the literature, and they do not encompass all relevant aspects of diversity. For instance, although the field of disability studies was first recognized by scholars in the 1980s, no corresponding LC subject heading existed until 2001 (Jahnke, Tanaka, & Palazzolo, 2022). Scholars have long recognized the deficiencies of subject headings for the identification of interdisciplinary topics, and a similar problem limits the effectiveness of LC call numbers (Ciszek & Young, 2010; Clarke & Schoonmaker, 2020; Gonzalez, 2023; McKinzie, 1994; Mosby, 1994; Phelps, 2021). Because of these limitations, diversity audits based on catalog searching often require multiple rounds of search strategy refinement. The best approach may be to conduct relatively broad (inclusive) searches, then to manually remove any titles that are not actually related to diversity.

3.2 The Checklist Method

The checklist method has five basic steps:

1. Identify bibliographies or other lists of books that correspond to each topic or group of interest.

2. Evaluate those lists to determine which of them best represent the works that ought to be included in the library collection.

3. For each topic or group of interest, compile a checklist of books from the source(s) identified in step 2.

4. Search the library’s book catalog by title (and by author, etc., if necessary) to determine which of the books on each checklist can be found within the collection. This can be done manually, title by title, or through automated means such as matching the ISBNs on the checklist to those in the library catalog.

5. For each topic or group, count how many of the checklist titles that are held by the library – how many were retrieved by the title searches.

Like the catalog search method, the checklist method can be used to generate comparative data for multiple institutions. The lists of diverse books held by the library can also be used in the compilation of descriptive data and use statistics. In a notable example, Monroe-Gulick and Morris (2023) downloaded acquisition information for each of the checklist titles held by their library, then examined the mechanisms by which the books had been acquired.

The checklists used in diversity audits can be compiled from a range of sources:

• lists of the books that have won, or been nominated for, diversity-related awards;

• selective bibliographies or book lists compiled by subject experts, professional organizations, interest groups, or library book vendors (e.g., EBSCO, 2022);

• review sources such as Booklist and Choice magazines;

• searches for diverse books in vendor’s databases, databases of recommended books (e.g., Resources for College Libraries), or large-scale book databases such as Amazon.com;

• searches of databases that are devoted entirely to diversity-related books, such as Diverse Bookfinder (Bates College, 2023);

• in-house sources such as course syllabi and reading lists;

• consultations with faculty at the home institution or elsewhere.

Just as the catalog search method requires the use of professional judgment in the development of search strategies, the checklist method requires judgment in the identification of diverse topics or groups, the selection of information sources (particular awards, bibliographies, databases, etc.), and the decision to include all or just some of the works listed in each source. The audit team must therefore have confidence in their information sources and in their ability to use those sources effectively.

Although diversity-related book awards and specialized bibliographies are the most commonly used sources for checklists, vendors’ databases such as GOBI may also be helpful. GOBI’s search capabilities provide for the generation of book lists based on subject headings, call numbers, and more than 100 interdisciplinary descriptors assigned by the GOBI subject specialists. About 30 of the descriptors represent diversity-related topics or groups, including both widely recognized areas of interest (black studies, women’s studies, etc.) and aspects of diversity that are less often considered, such as Appalachian studies. Alerts, notification plans, and approval plans can also be constructed using these same fields.

As discussed later (Section 5.1), a major disadvantage of vendors’ databases is that they do not include the full range of publishers. Many books related to diversity are published by smaller or specialized presses that vendors such as GOBI simply don’t cover (Bostic, 1995; Edinger, 2023; O’Neal et al., 2020; Phelps, 2021; Ryan, 2022; Stone, 2020). However, an advantage of these databases is that they include many good books that have not received national awards or critical acclaim. An overreliance on award winners and other “best books” may limit the range of topics and perspectives represented in the collection. It may also result in higher selection standards for diverse books than for non-diverse books.

Several tools allow for the comparison of library holdings against the checklists compiled by vendors and other organizations. For instance, the DEI Analysis Tool of collectionHQ can be used to identify holdings that fall into any of 12 categories representing various facets of race, ethnicity, LGBTQ+ identity, disability, health status, religion, and other attributes (Baker & Taylor, 2023; Bridgeall Libraries, 2023). Although intended mainly for public libraries, collectionHQ may be useful for academic libraries as well (Fuller-Gregory, 2022).

Librarians seeking to evaluate the diversity of their children’s and young adult (YA) collections have at least three products to choose from: Diverse BookFinder (Bates College, 2023), the Titlewave Diversity Analysis Tool (Follett School Solutions, 2023), and the TeachingBooks (2023) Collection Analysis Toolkit. Diverse BookFinder covers only picture books but is probably the most widely used tool of this type, perhaps because it can be used with either the checklist method or the book inspection method (See Section 3.3.3).

3.3 The Book Inspection Method

With the checklist method, the audit team compiles a list of diverse books, then determines how many of them are held by the library. In contrast, the book inspection method starts with a list of the library’s holdings. The team must then subjectively determine whether each title qualifies as a diverse book – whether it falls into any of the diversity-related categories that have been established in advance (Pattee, 2020; Powe, 2021). In nearly all cases, this method involves the inspection of the books themselves, although it can also be conducted using substitutes such as Choice cards or publishers’ descriptions. Like the catalog search method, the book inspection method lends itself to the subsequent compilation of descriptive data and use statistics. It is not well suited to comparative analyses, however, since those would require the large-scale inspection of the individual titles held by each institution.

The book inspection method relies on professional judgment in at least two respects – first in the delineation and description of the various categories of diversity-related books, then in the assignment of each individual book to any relevant categories. The process does not require the use of subject headings, call numbers, search strategies, or checklists, but it does require knowledge, expertise, and a common understanding of standards and procedures on the part of the audit team members (Jahnke et al., 2022).

The diversity categories used in book inspection audits often correspond to the aspects of diversity shown in Table 2. However, the categories need to represent specific groups (e.g., black, Hispanic, indigenous) rather than general concepts such as race or ethnicity. The six guidelines presented in Section 2.1 may be helpful when delineating diversity categories. Each category should be clearly defined in terms that everyone on the audit team understands, and the team should establish (in advance) a systematic method for working out any ambiguities or concerns that come to light as the evaluation process continues.

4.1 Setting Collection Diversity Targets

As explained in Section 1.1, diversity audits can help us determine the extent to which change is required, identify specific areas for improvement, set goals, and measure progress toward our goals. Ideally, efforts to diversify our collections would be pursued with specific targets in mind (e.g., “X% of the books in the HM–HX range should discuss topics or perspectives related to Hispanics in North America”). However, no recent study presents a strong, coherent basis for the establishment of specific benchmarks or targets. Instead, most authors seem to implicitly adopt one or more of three goals:

1. Improve the representation of diverse groups over time.

2. Ensure that the book collection covers underrepresented or marginalized groups in accordance with their representation within the student population or the broader community. (See Section 6.1).

3. Ensure that the library’s coverage of diverse groups is at least as good as that of its peer institutions. Products such as GreenGlass (OCLC, 2023a) and WorldShare Collection Evaluation (OCLC, 2023b), which compare the holdings of multiple libraries, may be useful for this purpose.

The second of these goals provides the clearest basis for setting targets and for knowing when the goal has been met (e.g., when the percentage of books representing group A matches the percentage of group-A individuals within the community). Even then, however, the implications of the diversity audit results are not always straightforward. For instance, Salem (2022) found that blacks were substantially overrepresented (by this measure) within the library’s picture book collection at San Diego State University. One-third of the characters in the picture books were black, compared to just 8% of the student population and 6% of the San Diego County public school population. (In contrast, Hispanic characters were seriously underrepresented). The full implications of the discrepancy in representation are not clear – obviously, the library will not stop collecting picture books with black characters – but this example does show the importance of setting targets and, more broadly, of understanding the goals of the library’s DEI efforts with regard to collection development.

4.2 Additional Analyses

Because the goal of collection-building efforts is usually to show improvement over time, the diversity audit cannot be a one-time exercise. Fort (2021) recommends an annual or biennial audit, while other authors recommend an audit of the current collection followed by ongoing audits of newly acquired books as they are ordered or received (Bird & McPherson-Joseph, 2020; Jahnke et al., 2022; Jensen, 2018).

Investigations into the use of diverse books may also be helpful. These studies can potentially account for print circulation, e-book downloads, the use of books in ILL transactions, or even citation impact. The simplest approach is to gather use data when completing the main part of the diversity audit – to download circulation statistics when compiling lists of search results (catalog search method), determining which checklist titles are held by the library (checklist method), or creating a shelflist of library holdings (book inspection method). So far, only two diversity audit studies have incorporated data on book use. Mosby’s (1994) checklist study includes a brief analyses of circulation data for 350 books related to African American studies. Her analysis is limited in scope, however, as it was intended mainly to identify print volumes in need of replacement due to heavy use. More recently, Gonzalez (2023) supplemented her catalog search study with information on the percentage of books in each of 11 diversity-related call number ranges that had circulated at least once. However, neither study indicates whether diverse books are checked out more or less frequently than other books in the same subject areas.

Qualitative analyses can be useful as complements to conventional diversity audits. Ciszek and Young (2010) discuss the use of interviews and focus groups to elicit patrons’ opinions regarding the representation of underrepresented and marginalized groups. Kester (2022) found faculty input especially helpful in her evaluation of the children’s and YA collections at the University of Florida; she identified the courses that made use of diverse books, examined the course syllabi, and interviewed the faculty. One advantage of qualitative methods is that they can be readily tailored to specific groups of patrons such as students and faculty from countries other than the United States. One disadvantage, however, is that nearly all qualitative methods focus on individuals’ perceptions, which are often biased in both systematic and non-systematic ways (Applegate, 1993; Bergen & Labonté, 2020; Cyr, 2016; Walters, 2003).

5.1 Book Vendors and Selection Tools

As mentioned in Section 3.2, most academic book vendors’ databases allow for the identification of diverse books based on their subject headings, call numbers, and vendor-supplied descriptors. For instance, the GOBI bibliographers assign nearly 30 descriptors or tags that correspond to various aspects of diversity. Selectors can therefore identify and evaluate diverse books as part of their routine activities – as a task undertaken within GOBI on a monthly basis, for instance.

Moreover, the same identification mechanisms used to search vendors’ databases can be incorporated into notification or approval plans. Many plans allow libraries to specify broader coverage in diversity-related areas than elsewhere. For instance, an approval plan might be limited to basic essential and research essential books, overall, but provide for the delivery of basic recommended and research recommended books in cases where one or more diversity-related descriptors have been assigned. Notification and approval plans may also include award-winning books that do not otherwise meet the library’s selection criteria. GOBI tracks nearly 1,000 book awards, and many of them are related to diversity (EBSCO, 2021). Librarians may also want to ensure that their approval plans include all the relevant publishers of diverse books, to the extent that their vendors cover them.

Unfortunately, a reliance on mainstream publishers, vendors, and selection tools is unlikely to provide comprehensive coverage of diversity-related books. The problem begins with the publishers themselves. As Jiménez and Beckert (2020) have pointed out, US publishing industry personnel are disproportionately white. Just 5% are black, and just 6% are Hispanic. It is not clear whether the demographic composition of the industry has a strong impact on the books that are published, however. If that were the case, we might expect to see more books that present female and queer perspectives, since 74% of publishing industry personnel are female and nearly 20% identify as LGBTQ+. Morales et al. (2014, p. 446) offer a more convincing explanation for the underrepresentation of authors from marginalized groups – that systemic economic and cultural biases, which influence mainstream publishers but are not specific to the publishing industry, “affect access to the resources necessary for a scholar to publish her work and to have that work marketed and recognized as authoritative.” Although these biases appear to be widespread, independent publishers that focus on diversity are working to change the situation. In fact, one of the most effective ways to build a diverse collection is to expand the range of publishers from which books are acquired (Bostic, 1995; Edinger, 2023; Morales et al., 2014; O’Neal et al., 2020; Phelps, 2021; Ryan, 2022).

The underrepresentation of marginalized groups can be seen not just in the titles published by mainstream publishers, but in the book reviews that selectors often rely on. For instance, the 1988 edition of Books for College Libraries (American Library Association, 1988) included just 41% of the 395 titles mentioned in four women’s studies bibliographies (Delaney-Lehman, 1994). More recent evidence shows that whites and men are overrepresented among book reviewers, and that books by white authors are especially likely to be reviewed (Gay, 2012; Franklin, 2011; Morales et al., 2014). Subjective reports support these same conclusions (Bird & McPherson-Joseph, 2020; Matteson, 2017).

Perhaps the biggest problem is that many diverse books are simply not listed in the databases of the largest academic book vendors. This limits selectors’ awareness of these titles and also makes it difficult to acquire them through standard acquisition procedures. Several authors have noted that vendors such as GOBI provide poor coverage of smaller or specialized presses (Bostic, 1995; Edinger, 2023; O’Neal et al., 2020; Ryan, 2022; Stone, 2020), and three studies make the case very clearly. Kristick (2020) found that only 9 of 22 major diversity-related book awards could be incorporated into GOBI approval plans. Likewise, Phelps (2021) discovered that GOBI offered books from only 4 of the 420 black studies publishers she identified, and that the database included relatively few of the books listed under 20 diversity-related subject headings in WorldCat. Finally, Monroe-Gulick and Morris (2023) examined the methods by which diverse books were acquired, or could have been acquired, at the University of Kansas. They found that a high proportion of the award-winning books that the library failed to acquire were from publishers not covered by GOBI’s approval and notification plans.

5.2 New Selection Strategies

Most of the methods used to improve the library collection in response to a diversity audit can be grouped into two categories: new selection strategies and new initiatives. With regard to selection strategies, we can:

• search vendors’ databases for diverse titles on a regular basis, as part of our routine selection activities;

• reconfigure our notification and approval plans to include more diversity-related titles, perhaps by arranging for special coverage of award-winning books and those with diversity-related descriptors;

• identify publishers and imprints that focus on diverse books (e.g., Edinger, 2023);

• increase the range of publishers from which we receive catalogs, e-mail announcements, and other selection tools;

• seek out book reviews and finding aids that focus on diverse topics and perspectives (e.g., EmbraceRace, 2023; We Need Diverse Books, 2023);

• develop mechanisms that provide for the straightforward acquisition of books from alternative publishers and vendors;

• participate in local and regional events that focus on books and media for particular underrepresented groups.

Many of the strategies that help increase selectors’ awareness of diverse books can also lead to better relationships with underrepresented communities. Bird and McPherson-Joseph (2020) have emphasized the importance of seeking out minority-owned or minority-serving bookstores, publishers, conferences, book lists, and newsletters.

Although these activities can be undertaken quietly, there are advantages to publicly asserting the library’s commitment to diverse collections. Serebnick and Quinn (1995) present compelling evidence that among public libraries, the existence of a formal collection policy is itself associated with greater diversity of opinion on key issues such as abortion, capital punishment, disarmament, and euthanasia. A similar relationship may hold with regard to underrepresented and marginalized groups. Likewise, libraries can demonstrate their support for greater diversity by documenting the selection strategies they have adopted in pursuit of that goal. Although most academic libraries’ policy documents include no mention of collection diversity or representativeness (Anaya & Maxey-Harris, 2017; Jamison, 2021; Maxey-Harris & Anaya, 2010; Mestre, 2011; Young, 2006), there are some notable exceptions. Of the collection diversity statements currently available online, the most comprehensive are those of the Massachusetts Institute of Technology (2017), the University of Maryland (2019), and the University of San Francisco (2021).

5.3 New Initiatives

One-time retrospective purchasing projects, which fill gaps in the collection by supplying the most important titles published in previous years, are among the initiatives most often pursued in response to diversity audits. Even modest investments can produce significant improvements. Lake Superior State University initially held only 11% of the works on their checklist of diverse titles, but a small investment – just $1,000 – was enough to increase that proportion to 30% (Delaney-Lehman, 1996). At the University of Florida, intensive efforts to upgrade the children’s picture book collection were enough to increase the number of diverse titles from 290 to 554 over a 6-month period (Kester, 2022). While out-of-print titles have not always been easy to find (Pettingill & Morgan, 1996), web sites such as Alibris and Abebooks have done much to alleviate the problem. In fact, these books are often inexpensive, at least before transaction and processing costs are considered.

Another method of diversifying the collection is to acquire print standing orders or e-book packages designed specifically to provide greater representation of diverse groups. Several vendors and publishers offer products of this type, which can focus on diversity in general or on particular themes such as gender studies and immigrant studies (e.g., Booksource, 2023; EBSCO, 2023a,b; Gale, 2023; Junior Library Guild, 2023; ProQuest, 2023). When considering these collections, it is important to keep in mind that they may not be representative of the broader literature in terms of topics, perspectives, or authorship, and that vendors’ conceptions of diversity may not be fully aligned with those of particular libraries. In particular, some vendors’ diversity collections seem to include many titles that are not about particular groups or perspectives, but about diversity itself or the push for diversity. They may be unlikely to cover black men’s perspectives on the justice system, for instance, but more likely to focus on the ways in which companies can meet their legal obligations with regard to DEI. E-book packages pose a number of additional, format-related, challenges (Walters, 2013, 2014), including one that is specific to diversity collections: some key titles are available only in print, since the publishers that focus on diversity are sometimes smaller, based outside the United States, or less fully integrated into the scholarly communication system (Jahnke et al., 2022). Ideally, we would evaluate each standing order or e-book collection by applying the same criteria used for a diversity audit, such as the percentage of titles that fall within relevant call number ranges, the percentage that can be found on diversity checklists, or the percentage that can be classified as diverse books based on a review of their content.

Some libraries have been successful in obtaining special funding for diverse books, and donors or external agencies may be eager to fund special collections that center on particular groups or particular aspects of diversity. Even without new revenue, the establishment of special funds – i.e., the diversion of money from other library budgets – can help support diversity efforts and allow for the tracking of diversity-related acquisitions and expenditures (McKinzie, 1994; Pettingill & Morgan, 1996). Moreover, new special collections established through the regular library budget may attract external funding once they have gained public attention (Proctor, 2020).

Collection-related DEI initiatives should involve faculty outside the library, especially at institutions where those faculty have book selection responsibilities. At Dickinson College, the librarians worked with professors in a special summer program to increase the library’s holdings of black studies titles (McKinzie, 1994). The participants independently compiled checklists of diverse books, then conducted a diversity audit and attempted to acquire the checklisted books not already held by the library. McKinzie’s description of the project suggests that its success can be attributed to several factors:

• recognition that the library’s usual selection practices, directly linked to academic departments and programs, were not enough to build a strong, multidisciplinary black studies collection;

• a well-defined summer project, separate from the regular activities of the librarians and faculty, with stipends for participating faculty;

• the involvement of individuals from a wide range of disciplines, and the independent compilation of checklists that reflected the unique perspectives of each field;

• the designation of a special fund for project-related acquisitions even though the library received no additional funding.

Another collaborative project, at the University of Florida, began with the identification of the courses for which diverse children’s books were most likely to be needed. The librarians examined course syllabi, interviewed instructors, and recruited particular faculty to take the lead in the selection of new titles (Kester, 2022).

6.1 Representing Our Communities or Expanding their Horizons?

One fundamental question is whether we should build collections that match the characteristics of our communities (i.e., help patrons “see themselves” in the collection) or intentionally seek out topics and perspectives that are less familiar to patrons (i.e., help students expand their horizons through exposure to ideas they might not have encountered outside the academic environment). The first approach has been adopted, explicitly or implicitly, in a wide range of diversity audit studies. (See, for example, Fort, 2021; Gibney et al., 2021; Lawton & Frank, 2022; Matteson, 2017; Morales et al., 2014; Powe, 2021; Salem, 2022; Wells et al., 2023; Williams, 2022). Advocates of this approach stress that diverse collections help students understand that their own experiences, attitudes, and perspectives are valid, and that the library is there for them. Students who see themselves in the collection may be more comfortable using the library and more likely to take advantage of the full range of collections and services available to them.

There are two problems with this approach, however. First, the goal of helping students understand that their own perspectives are valid is based on the assumption that individual students’ perspectives will match those of the groups to which they belong. The see themselves approach validates not individuals’ backgrounds or views, but the norms of particular subpopulations that have been recognized by the broader community, such as Asian Americans or older adults. It reinforces the idea that individuals need not conform to the norms of the majority, but it also suggests that individuals should conform to the norms of some recognized group. Young black men who see themselves in the music of Céline Dion may not be well served by this approach. That is, an emphasis on particular groups as surrogates for individuals may itself marginalize divergent views and reduce students’ acceptance of the more complex or subtle forms of intersectionality (“What if I don’t like hip hop? Is it o.k. to feel that way?”)

A second problem is that strict adherence to representation-based standards can actually work against the goals of many DEI initiatives. For example, a librarian quoted by Matteson (2017, p. 26) argues that “If I have 80% students of color, then my library should have 80% characters of color.” This standard is unreasonable in at least two respects. First, it will tend to bring in too few books whenever a particular group is poorly represented within the local population. A particular university may have no Native American students at all, but a collection without any books on indigenous groups cannot possibly be considered satisfactory. Second, the universe of published books may include relatively few books with characters representing a particular group (e.g., Pacific Islanders). In that case, the only way to match the proportion of Pacific Islander students at the University of Hawai’i is to either (a) add every available book with Pacific Islander characters, including low-quality titles that would never be considered otherwise or (b) limit the overall size of the collection (i.e., the number of books without Pacific Islander characters) so that the percentage criterion can be met.

The alternative approach to building diverse collections – helping students expand their horizons through exposure to ideas they might not have encountered otherwise – has been explicitly recognized by only a few authors (Cavanaugh, 1995; Naidoo, 2014). Nonetheless, it seems to support one of the overall aims of higher education: to build respect for opinions, outlooks, ways of reasoning, sources of evidence, and ways of life other than one’s own. Within this framework, it is especially important to collect books that cover those aspects of diversity (or those marginalized groups) that are not well represented within the student population. After all, students unfamiliar with Cuban immigrants may benefit more from reading about their experiences than students who interact with Cuban immigrants every day. Likewise, students in Pennsylvania or Ohio will probably be aware of Amish communities, have at least some knowledge of them, and know where to find more information about them, while students in Arizona may be completely unfamiliar with the Amish or their unique way of life. Arguably, the Arizona students have more to gain from exposure to books about Amish culture.

Conceptually, these two approaches – representing the community and expanding students’ horizons – are not complementary at all, since they lead to completely different conclusions about libraries’ collection development priorities. In practice, however, it is possible to adopt one approach while tempering it with elements of the other. We might choose representation-based standards, for instance, but add a caveat that books about small or unusual groups will be eligible for inclusion regardless of the groups’ representation within the student population. Likewise, we might focus on broadening students’ exposure to unfamiliar ideas while still acknowledging the need to provide a specified level of coverage of particular groups within the university community.

6.2 Societal or Educational Goals?

Another key question is whether our collection-building efforts should focus on broad societal goals or on specific educational objectives. This question addresses the potential conflict between the two commitments mentioned in Section 1: facilitating social change and promoting educational success.

Lawrence (2021) has argued that libraries have an obligation to work toward positive social change, and that the educational mission of the library can be regarded as one part of a broader effort to promote social justice. In fact, he defines diverse works of fiction based not just on their content or their authorship, but on the extent to which they aid in that effort: “Diverse books are those systematically devalorized works we must make an ameliorative effort to promote in order to advance informational justice for oppressed persons in particular” (p. 189).

In contrast, the educational approach is based on the acknowledgment that libraries, like their parent organizations, are almost always devoted to the education of particular individuals in particular ways. That is, they are democratically or privately supported to meet specific educational objectives – and held accountable for doing so. Universities succeed in this more focused role by broadening the perspectives, enhancing the capabilities, and improving the career and life prospects of their students. The institutional commitment to DEI can therefore be expressed in terms of educational outcomes such as the academic achievement of students from underrepresented and marginalized groups.

Societal and educational goals are often aligned, and the strength of a university’s commitment to social change does not always enter into the collection development decisions of librarians and faculty. Nonetheless, there may be situations in which the two perspectives are not fully compatible. For instance, consider a scenario in which the Black Student Union of a small, minority-serving college requests a substantial number of self-help books that link personal development to “the creation of the social and political conditions under which that development can take place” (McKeen, 2002, p. 409). Assume that the books are engaging and perhaps even inspirational but that they do not meet the library’s usual selection criteria. That is, they do not directly support the students’ educational achievement or the faculty’s research. A library with an overriding commitment to social justice might acquire the books anyway, to promote awareness, foster social change, demonstrate support for the black community, and provide a platform for authors who may be unfairly excluded from the mainstream media (Morales et al., 2014). In contrast, a library that prioritizes educational outcomes might evaluate the requested titles based solely on the extent to which they influence students’ academic success – i.e., the extent to which they are likely to promote engagement with the library and encourage students to take advantage of other library resources that will influence their course performance, graduation rates, and career prospects.

In a well-funded library, the distinction between these two approaches may not be important. When resources are limited, however, we may need to choose between the self-help book requested by the Black Student Union and the text that will help students learn statistical concepts through intuitive rather than formal methods. We therefore need to evaluate the relative importance of societal and educational goals within the context of our universities’ missions. Although nearly all universities have expressed a strong commitment to DEI, the form of that commitment can vary substantially based on each university’s size, student characteristics, teaching or research emphasis, academic or career focus, institutional history, and external relationships.

6.3 Representation or Diversity?

Diverse books is a misleading phrase for at least two reasons. First, we are almost always interested in the extent to which diverse topics, perspectives, opinions, and groups are represented in the collection as a whole – not within individual titles. In most cases, the objective is not to collect individual works that present multiple perspectives, such as CQ Researcher or Opposing Viewpoints, but to build a collection that reflects diverse perspectives through its components (Jahnke et al., 2022). Second, even when we accept that our real interest lies in collections rather than individual titles, a fundamental question remains: Are we concerned primarily with representation or with diversity?

The literature on diversity in library collections deals almost exclusively with representation (Lawrence, 2020, 2021; O’Neal et al., 2020). In this context, diverse books are those that present the voices, perspectives, or experiences of underrepresented or marginalized groups. This emphasis on representation within the library collection may be an outgrowth of the longstanding concern for representation within libraries more generally – among library staff, for instance. As Winston (2014, p. 85) has noted, libraries’ efforts to support diversity have been measured

largely on the basis of representation of members of various segments of the population as a percentage of those in the organization or profession…. To a large extent, the diversity goals in the profession have been related primarily to underrepresentation, and less so to diversifying collections and enhancing services.

For many purposes, this emphasis on representation is appropriate. At the same time, we should recognize that diversity can also be considered in another way, as the extent to which widely varying viewpoints or groups are included. The key concept here is variation, in a quasi-statistical sense. For instance, a collection that provides outstanding coverage of black and Hispanic groups, but only those particular groups, might score high in terms of representativeness (since diverse groups account for most of the collection) but low in terms of variation (since only two groups are represented). In contrast, a collection with high variation would include books associated with a wide range of groups, perhaps with a specified minimum number of books in each category. The goal is not just to collect a relatively high number of books that represent diverse groups, but to maximize the degree of variation within the collection.

This alternative conceptualization of diversity has not yet been applied to a diversity audit. However, a similar approach was adopted by Garland (1981, 1982), who pointed out that for any categorical variable, such as academic subfield, the diversity of a collection can be expressed in terms of (a) the number of categories represented in the collection and (b) the extent to which each category is represented. Garland’s diversity index has a high value when the categories with the fewest items have relatively many items – when the books in the collection are distributed evenly among the various categories. In contrast, it has a low value when the categories with the fewest items have relatively few items. Garland introduced her diversity measure in the context of topic or subject, but the same principle can be extended to the racial/ethnic categories used to classify authors, fictional characters, or cultural themes, or to any other aspect of diversity that can be expressed as a set of categories. The advantage of this approach is that efforts to maximize the representation of any particular category (or group) are valued to the extent that they increase the representation of the least-represented groups. Once a group is well represented, additional increases will have a less positive impact, or even a negative impact, on the diversity score. Garland’s diversity index does have two important limitations, however. First, the various categories must be mutually exclusive. This poses problems if we want to account for intersectionality – for the interaction effects associated with particular combinations of characteristics. Second, Garland’s diversity index rewards equal representation rather than some other level of representation that might be more appropriate. Nonetheless, the underlying principle remains valid, and it is possible to develop measures that account for preferred outcomes other than equal representation.

A related measure, Simpson’s index of diversity (Simpson, 1949), has been used to evaluate diversity of opinion within book collections that cover topics such as abortion, capital punishment, disarmament, and euthanasia (Serebnick & Quinn, 1995). Methods based on indicators such as these hold promise for a more nuanced approach to the assessment of diversity – one that accounts for both representation and variation.