Recent Library Practitioner Research: A Methodological Analysis and Critique Submitted by Dr. Charles R. Hildreth and Ms. Selena Aytac Palmer School of Library and Information Science Long Island University/C.W. Post Campus 720 Northern Blvd. Brookville, NY 11548 516-299-2178 (Send all electronic correspondence to hildreth@liu.edu) 1-(2/15/2016) ABSTRACT Recent Library Practitioner Research: A Methodological Analysis and Critique Every few years Library and Information Science (LIS) scholars report on the state of research in librarianship in general and practitioner research, in particular. These reviews, some more extensive in journal coverage than others, focus on topics addressed, methodologies used, and quality of the research. At the January 2003 ALISE meeting, we presented the findings of our review of the 1998-2002 published librarianship research [1]. We decided to update that analytical review for two reasons: to look at more recent literature (2003-2005), and to use refined methodological criteria to evaluate these studies. From a purposive sample of 23 LIS journals we have mined 401 research articles. 206 articles were randomly selected for this in-depth analysis. A checklist of 35 factors was applied in our analysis of these research articles. These project and report factors include authorship, topic, location and setting, type of research, data collection methods, type of data analysis, statistical techniques, and components and quality of the report itself. The data collected were statistically analyzed to produce summary data and to provide an opportunity to discover any significant associations among the many variables. Such analysis supports comparisons of practitioner and academic scholar research. The descriptive data has enabled us to document the status quo in recent practitioner research. These findings are used to explore recent patterns or trends in library practitioner research and they provide a basis for comparisons with earlier reported findings and assessments. 2-(2/15/2016) Recent Library Practitioner Research: A Methodological Analysis and Critique INTRODUCTION Every few years Library and Information Science (LIS) scholars report on the state of research in librarianship in general, and practitioner research, in particular. These reviews, some more extensive in journal coverage than others, typically focus on topics addressed, methodologies used, and quality of the research. At the January 2003 ALISE meeting, we presented the findings of our review of the 1998-2002 published library research, with special focus on research conducted by library practitioners [1]. We decided to update that analytical review for two reasons: to look at more recent literature (2003-2005), and to use refined methodological criteria to evaluate these studies. Consistent with the aims of most earlier reviews, we designed a study that would permit us to discover and document the status quo of recently published library research. In the review reported here we focus primarily on research methodology issues and the quality of the published reports of both practitioner and academic scholar research. By “academic scholar” we mean those individuals who are members of teaching and research faculties at colleges and universities, typically in graduate degree programs in library or information science. Topics of research were also included in this review and analysis. Topics found in the research literature were coded and documented using a concise taxonomy of broad categories. In a future study we plan to employ a more detailed, finely-grained classification of research topics that will more accurately represent the number and diversity of research topics we encountered in this review of the research literature. The data recorded for more than 30 characteristics of the 2003-2005 published research studies in our sample has enabled us to identify patterns and trends in research approaches, strategies, and use of a variety of methodologies. Our findings permit comparisons with earlier studies and assessments of the accuracy of past predictions regarding the direction of future library research. As the title of this document indicates, we were most interested in documenting and analyzing recent library practitioner research. For purposes of comparative 3-(2/15/2016) analysis we also looked at a representative sample of research conducted and published by academic scholars during the same period, 2003-2005. LITERATURE REVIEW If there is a common theme in past reviews of the library and information science (LIS) research literature it is this: library-related research is substandard, although developments over the last few decades of the twentieth century indicate that improvements are continually being made. As summarized by Powell and Connaway, “Those who have assessed the previous research of librarians have been of a consensus that the quantity and quality have left something to be desired.”[2] No doubt mindful of Hernon and Schwartz’s “call for a higher standard for the conduct of research and the publication of research results,”[3] Powell and Connaway, focusing on recent library research in the first chapter of their 2004 textbook, Basic Research Methods for Librarians, present a credible case that there has been marked improvement in library research in recent years. The areas of improvement they highlight include better conceptualizations of topics and questions to be investigated, and more rigorous application of a variety of research methods and statistical analysis techniques. The research “standard of quality” referenced in reviews like these reflect the reviewer’s views on what constitutes sound “scientific inquiry.” It is not always clear if a research orientation is primarily positivist or post-positivist. One’s standard of “scientific inquiry” may be no more than the elementary, conventional version of the five-step “scientific method” in which one proceeds from formulation of problem, through testing, to results. Hernon and Schwartz more felicitously describe the research process as “an inquiry process that includes components for reflective inquiry, research design, methodology, data collection and analysis, and the communication of findings.”[4] Unlike Hernon and Schwartz, reviewers of library research rarely reveal their own philosophical or methodological orientations and biases. They may be positivists or post-positivist-constructivists, experimentalists or ethnographers. They may prefer a qualitative methodological approach over quantitative methods, or vice versa. These orientations and preferences may well influence a reviewer’s assessment of “good” research and the type of library research area most in need of improvement. We think it is safe to say that critical reviews published before the 1980s reflect a positivist point of view, with its preference for quantitative research methods. 4-(2/15/2016) In his 1967 invited editorial, Ennis, senior researcher at the National Opinion Research Center, observes that “library research is non-cumulative, fragmentary, generally weak, and relentlessly oriented to immediate practice….Most of these studies have been critically reviewed recently and the conclusions of the review are that there is little hard and useful knowledge gained. Samples were too small or poorly drawn, study designs were underconceptualized, measuring instruments and techniques were primitive.”[5] Commenting on the library literature of the 1970s, Grotzinger bemoans the lack of progress in library research and fears that librarians will continue to be “not only reluctant but resistant to the achievement of an understanding of the nature of scientific inquiry.”[6] Grotzinger attributes this resistance to the failure of LIS educators and their curricula to instill in library professionals a knowledge-based respect for the true “meaning of research” and scientific inquiry. The author makes it clear that competence in quantitative methods and statistics is the knowledge to be gained and applied to library research questions. In their 1980 textbook, Research Methods in Librarianship: Techniques and Interpretations, Busha and Harter state that “Unfortunately, a large population of librarianship’s research has been uneven in quality and demonstratively weak methodologically; therefore, too much of it has only approximated scientific inquiry.”[7] The authors’ positivist view of “scientific inquiry” is made clear immediately after this passage. Experimental research is the “key instrumentation of science,” to be used for the testing of hypotheses. Furthermore, establishing the validity and reliability of measurement instruments is of paramount importance to these authors. Beginning in the 1990s, reviewers of library research demonstrate a sympathetic understanding of non-positivist, qualitative research approaches. The methodologies of the social sciences have begun to exert a greater influence on LIS scholars than the longdominant quantitative methodologies of the natural sciences. This appreciation of the diverse methodologies applied insightfully and successfully in the social sciences can be seen in their assessments of library research. This influence is evident in the construction of more inclusive taxonomies of research methods, and in new and revised standards of “good” research. 5-(2/15/2016) Discussions of good research now emphasize the unequaled benefits of qualitative research and methodological “triangulation,” the complementary use of more than a single type of method. The emphasis on “quality” research shifts somewhat from maximizing the validity and reliability of data collection procedures, to personal immersion in the context of study, and insightful analysis of that context and the verbal or textual data extracted from that context. From the early 1990s to the present day, reviewers of library-related research increasingly place their analyses in the context of the social sciences and the qualitative methodologies that have gained favor in those disciplines. In her 1991 review of the quantity and quality of LIS research (“more but not better”), Van House turns to research in the social sciences to shed light on the deficiencies in LIS research, and for guidance on ways to improve this research, with special focus on research methodologies. She asks of LIS research, “How good is it, according to the standards of social science research?”[8] She concludes that library-related research “suffers from serious, but not fatal, methodological and content problems.” The causes of this state of affairs are best understood by looking at the quality of research in the social sciences and how that quality is achieved and evaluated by social science researchers. Oddly, when Van House discusses reasons for the methodological shortcomings of LIS research, she looks inwardly to the profession and discipline of library science. The “culprits” identified by Van House include the pressure for academic library practitioners to “publish or perish,” and the inadequate training in research methods provided by the graduate library schools. She recommends that practitioners not conduct research unless they are well-trained and can address significant problems of interest to a wider audience in related disciplines. Until this comes about, Van House recommends that practitioner research be published primarily in “practitioner-oriented journals,” not scholarly journals. An enthusiastic proponent of qualitative research, Raya Fidel documents the growing interest in and increased use of qualitative methods in information retrieval research.[9] Her review of the LIS research literature is confined to research in information retrieval, but she pleads her case for the expanded use of qualitative methods in all areas of LIS research. She offers three explanations for the increased use of qualitative methods: “the failure of quantitative methods to produce what was expected of them; the move toward a user-centered approach; and the 6-(2/15/2016) growing interest in qualitative methods in other areas in the social sciences.”[10] Fidel clearly believes the path to improved, more relevant library research goes through the methodological territory of the social sciences. In his examination of recent trends in LIS research at the end of the twentieth century, Powell tells us he “reviewed the research literature of the social sciences, including library and information science.”[11] He reports on the many research methods identified and “deemed to be relevant to library and information science research.” Special attention is given to qualitative methods used in the social sciences. In successive editions of his well-respected textbook on library research methods, Powell has consistently documented the shortcomings of library research. In this “Trends” paper, Powell encourages researchers to expand their methodological repertoire to include qualitative methods used in the social sciences. His optimism about improvements in future library research is accompanied by his prediction that the use of qualitative methods and “triangulated” research designs will increase. In 2002, McKechnie and her colleagues conducted a methodological analysis of human information behavior research published between the years 1993 and 2000.[12] The aim of this review was to document recent methodological practices and identify any methods trends. This is one of the few reviews of LIS research to consider separately research by practitioners and research by academic scholars (generally, members of LIS graduate teaching faculties). The authors discovered the unfortunate fact that less than four percent of first authors of research reports in this area were practitioner researchers. McKechnie et al. report that the majority of the human information behavior studies were primarily qualitative and used two or more research methods. Interviews were identified as the method of choice, followed by survey questionnaires, observation, and content analysis. Wishing to focus on practitioner researchers in academic libraries, Watson-Boone examines 24 practitioner-authored research articles published in the Journal of Academic Librarianship between the years of 1985 and 1995.[13] After an illuminating context-based discussion of practitioners’ diverse motives for conducting research, she applies a six-method taxonomy to document methods used in their research. “Survey research” was the method of choice in one- 7-(2/15/2016) half of the studies (12), followed by “action research” (5, 20.8%). Only one experiment was found, and one study was classified as “evaluative.” Watson-Boone’s review suffers not only from its small sample of practitioner research, but also from its short, confusing list of research methods. She equates types of research with specific data collection techniques. Applied research, for example, may well use survey research methods for data collection. In short, she confuses research orientations or motive-based research strategies and approaches with specific data collection methods (e.g., questionnaires, interviews, observations, etc.). In our study we have attempted to avoid this confusion by making clear distinctions between methodological orientation, research types and strategies on the one hand, and data collection and data analysis methods, on the other hand. Inspired by the emergence of the “evidence-based librarianship movement” (EBL), Koufogiannakis and her associates conducted an extensive review of 807 research articles mined from 91 LIS journals for the year 2001. Their content analysis of these articles focused on topics researched, methods used, authorship characteristics, journal source, and indexing and abstracting services. The authors also attempted to “identify what if any correlation exists between the research method used and the subject (‘domain’) of research.”[14] After providing an extensive literature review of previous LIS research study reviews, Koufogiannakis et al. report that few of these studies have attempted to demonstrate correlations between variables studied, especially correlations between subject areas researched and research methods employed. In pursuit of this objective, Koufogiannakis et al. create a six-item subject domain taxonomy and a 15-item methods (“study type”) taxonomy. Their descriptive data analysis revealed that the subject domain most researched was “information access and retrieval” (38.9%), followed by “collections” (24%), and “management” (16.7%). Research methods identified in the 15-item methods classification ranged from the general type (“descriptive”) to the specific (computer logging and analysis). The highest proportion of the research articles were classified as “descriptive” studies, followed in frequency by “comparative” studies. Warning that their classification of research methods may need further refinement, the authors display counts of articles by subject domain that reflect the use of one or another study type. Thus, we are presented with suggested correlations between subjects and study types. Unfortunately, there is no attempt to use 8-(2/15/2016) statistical techniques such as Chi Square tests for independence to determine if any of these associations between subject domains and study type (method) are statistically significant. In many ways our current study updates the review of the 1998-2002 library practitioner research literature presented by Hildreth.[15] That methodological study looked at a random sample of 100 research articles selected from 23 LIS journals. The purposive sample of journals from which these articles were mined was designed to include a majority of practitioner journals. A formal checklist was used to document both study characteristics (e.g., topic, research type, methods, statistics, etc.) and “quality” report characteristics (e.g., presence of clear research question, literature review, evidence-based discussion of findings, discussion of study limitations, etc.). The descriptive analysis revealed that survey questionnaires topped the list of most-used data collection methods, and that many studies used two or more data collection methods, suggesting an increase in methodological pluralism in LIS research. Parting from earlier reviews, Hildreth attempted to distinguish between overarching “research approaches” based on researcher orientation, motivation, and study aim or strategy (e.g., descriptive, exploratory), and data collection methods or data analysis techniques. These separate aspects of research study design, methods, and analyses have not been sufficiently distinguished and coded in earlier review studies. Applying these distinctions, Hildreth reported that only a low percentage of the studies could be classified as explorative in aim (25%). Only 29 percent of the studies used qualitative analysis in whole or part. Few of the studies employed correlational or inferential statistical analyses. Hildreth did not use statistical analysis in his 2003 study to determine the presence of significant relationships between variables such as topic and method, or authorship (practitioner or academic scholar) and research type. In the sections that follow we report on our review of the 2003-2005 LIS research literature. The findings include discovery of statistically significant correlations between key variables of interest. RESEARCH DESIGN AND METHOD A purposive sample of 23 LIS journals was selected for this analysis of the recent research literature in librarianship (see Appendix A). This selection of journals is weighted toward practitioner research. Several key journals that are primarily publishing venues for academic 9-(2/15/2016) scholars are included in this sample for purposes of comparison. Every issue of these 23 journals published between 2003-2005 was examined by the authors of this report to identify all of the articles we deemed to be reports of original empirical research studies. The criteria we applied in this selection of research articles is based on Hernon’s definition of research as “an inquiry process that has as its aim, the discovery or creation of knowledge, or theory building; testing, confirmation, revision, refutation of knowledge and theory; and/or investigation of a problem for local decision making.”[16] Hernon lists five activities that comprise such research activity: reflective activity, including identification of a problem and research questions; adoption of appropriate procedures in the research design and choice of methodologies; collection of data; data analysis; and presentation of findings and recommendations for future study. We judged a published article a good candidate for our review if it reported on original empirical research, research that addressed one or more specific research questions, systematically collected and analyzed relevant data, and presented and discussed the databased findings of the research. Thus, excluded from this review were literature reviews, methodology papers, historical studies, and analytical and opinion pieces. This stringent identification and selection process yielded 401 research articles. 206 of these articles were randomly selected for in-depth analysis. Appendix A displays the number of articles initially selected from each of the 23 journals, and, in parentheses, the number of articles from each journal included in the randomized sample. Each of the 401 candidate articles was given a unique number identifier, and, journal by journal, a random number generator program was used to select approximately one-half the articles for inclusion in our review and analysis sample. From a set of ten general variables (e.g., topic, setting, etc., See Appendix B), a coded checklist of 35 factors was developed, tested, and refined to support analysis and documentation of the characteristics of each of the 206 research articles. The final checklist of analysis factors emerged inductively from the actual literature. The study variables were divided into two broad categories: study variables, and report variables. The study variables focused on authorship, topic, setting, and methodological factors. The report variables focused 10-(2/15/2016) on the report itself, its components, organization, and overall quality of composition. The definitions of each of these variables and how they were applied in the article analyses will be discussed in the presentation of findings which follows. The 35-item article checklists were used independently by each of the authors of this paper. There were few differences in our separate analyses, and these differences were easily reconciled upon further reading, rereading, and fine-tuning of our factors. The data recorded on each article checklist were coded and entered into a database for statistical analysis. This analysis produced summary descriptive data on key methodological factors and provided a basis for analysis of significant correlations between key variables, including comparisons of practitioner and academic scholar research. (Hereafter, academic scholar research or authorship rubrics will be shortened to “academic,” with “scholar” assumed, and this will be differentiated from library “practitioner.”) Research questions addressed in this study include: 1. With regard to the key methodological variables, what do the data show? 2. What is the quality of recent research study reports? 3. In terms of methodology, has there been any change in patterns of use? 4. Have the predictions of earlier reviewers regarding methodological trends been confirmed? 5. How do practitioner and academic researchers compare on methods used and report quality? 6. What, if any, methodological aspect of this research literature most distinguishes practitioner research and academic research? 7. What significant categorical associations might emerge from these data (e.g., authorship and type or method of research)? FINDINGS Research-oriented Journals Appendix A identifies those journals that have yielded the greatest number of research articles for this study. Six of the 23 journals account for more than one-half of the research articles initially reviewed for inclusion in the study. They are led by the Journal of Academic Librarianship (69), the Journal of Documentation (48), and Library & Information Science 11-(2/15/2016) Research (33). On a per issue basis, the leading publishers of research articles among this group of 23 journals are the Canadian Journal of Information & Library Science, the Journal of Academic Librarianship, Journal of Documentation, Library & Information Science Research, Library Resources and Technical Services, and Research Strategies. Findings on the authorship variable will be presented next, but we have ascertained that six journals in our purposive sample are predominantly academic scholar in authorship, and ten journals are predominantly practitioner in authorship. Taken together, 82.4% of research articles selected from the academic journals were authored only by academic scholars (i.e., non-practitioners). Of the ten journals classified as practitioner journals, practitioners account for 74% of the articles authored only by practitioners. Contrary to our expectations, academic scholar journals did not yield more of the research articles for this study than the practitioner journals. Three of the top publishers of research articles are predominantly practitioner journals, two are predominantly academic scholar journals, and one has a nearly equal amount of practitioner-authored articles, academic scholar articles, and mixed authorship articles (at least one practitioner and one academic). Additional findings from a comparative analysis of practitioner journals and academic journals will be published in a separate paper. Authorship Three factors related to the research article authorship variable were examined: number of authors per article; relative frequencies of articles having only academic authors, those having only practitioner authors, and those with mixed authorship; and frequency of each type of author by journal. While the range of authors per article was 1-7, the mean number of authors per article was 1.86, the median, 2. There was very little mixed authorship found in this sample of 206 research articles (20, 9.71%). As indicated previously, six of the 23 journals are predominantly academic scholar in authorship, and ten are predominantly practitioner in authorship. None of the 23 journals yielded a significant amount of mixed authorship articles. Topic Areas The primary focus of this study is on research methodology, not topic coverage. We did use a 14-item list of topics to record the topics of research in each article (See Table 1, included after the appendices). This proved difficult for a number of reasons. Some articles were 12-(2/15/2016) encoded as having three or more topics, and it was not always clear which topic was primary, which secondary, and so forth. Our taxonomy needs to be expanded and refined to represent the many topics being addressed in library research. In light of these shortcomings, we collapsed our 14-item list to four categories (library operations and services, use and user studies, LIS education, and miscellaneous others). This short list of topic areas is used for preliminary analysis of associations of method variables with topic areas researched (See Table 2). The most surprising topical finding is the small amount of research having anything to do with LIS education (9, 2.39%). Practitioners more frequently investigate library operations, collections and services, and academics more frequently conduct use and user studies (this difference is statistically significant, even when including mixed authorship: =17.454, df=6, p=0.008, See Table 3). Both are investigating many additional topics. Study Location and Setting The study “location” in each article was coded as “specific site” or “site-independent.” The former refers to “in-house” studies and the latter was used to record studies that investigated phenomena of interest at more than a single institutional site and studies that investigated phenomena that are truly site-independent, such as Web search engine use or performance. These site-independent studies account for 59.71% of the total (See Table 4). Academic scholars conduct significantly more site-independent studies than practitioners (=5.570, df=1, p=0.019). The study domain variable “setting” was encoded as university, public library, special library, and other. The majority of the studies were coded as “university” (115, 55.83%). A significant number of the studies (55, 26.7%) were not library domain studies (See Table 5). We were intrigued to discover that academic scholars conducted most of the research activity in non-university library settings. On the other hand, practitioners conducted nearly two-thirds of the research set in the university environment (73 cases, 63.5%). Adding studies with mixed authorship raises this total to 81 cases (70.4%). Methodology We now turn to the primary focus of this study, methodology. Five variables were identified for this analysis: type of research by aim or strategy, sampling procedure and sample drawn, data collection method, data analysis approach, and statistical analysis employed. 13-(2/15/2016) Type of Research Several classifications of research types have become familiar to readers of the research literature. The best known is the classification of research as “basic or pure,” “applied,” and “action research.” A simpler approach distinguishes “primary” research from “secondary” research. More recently there has been the thought-provoking debate about the virtues of “qualitative” research and “quantitative” research. We will use this distinction in our characterization of the data analysis approaches used in research studies. We find most useful and discriminating the classification of research types described by Schutt in his book titled, Investigating the Social World: The Process and Practice of Research [17]. Schutt describes four types of research, based on the overarching aim, purpose, and intentions of the researcher evident in the conceptualization and design of the research. The four types of research are descriptive, exploratory, explanatory, and evaluative: Descriptive research aims to discover and document the “facts,” the status quo. The data acquired is usually quantitative data, such as counts, percentages and ratios, data from which means and medians can be calculated. Systematic observations on a well-defined and measurable variable and survey questionnaires are typical methods used. Exploratory research aims to “find out what’s going on” in a specific situation by immersion in that situation, prior to and perhaps in place of taking any measurements of designated variables. Seeing “reality” through their eyes, seeking meaning as understood by the subjects of the investigation. Less-structured than the other types of research, the pertinent data is typically qualitative data, such as textual or verbal data. Methods include participant observation, open-ended interviews and focus groups. Exploratory research usually employs qualitative research methods, and qualitative methods may also be used in other types of studies. Explanatory research, often in the past considered the paramount type of research, explanatory research aims to answer the “Why?” question by discovering and establishing (confirming) the actual cause of some phenomenon or state of affairs. This research typically tests hypotheses through the application of rigorous protocols for observing and measuring interactions between variables. The data is quantitative and is analyzed using inferential 14-(2/15/2016) statistics. Experiments, clinical trials, and even descriptive surveys are used in explanatory research. Evaluative research is a kind of “action” research that brings science and scientific methods to the evaluation of specific programs, policies, or techniques (e.g., outreach programs, web sites). This evaluation may be “formative” (used as a program is being implemented) or “summative” (used after a program or policy has been in place for a period of time, to assess its impact and outcomes). Both quantitative and qualitative research and data may be used in such studies. Each study in our review was coded as one or more of these four types of research. Table 6 shows the number of studies in each category and separately reports the number of studies characterized as reflecting in aim two or more research approaches (78, 37.9%). Most of the studies are descriptive in aim and strategy. One-third of the studies are exploratory in aim or are included in studies having multiple aims reflected in multiple research types in the design of the research. Few studies are explanatory studies (12.62%). We anticipated a larger percentage of evaluative studies, especially by practitioners, as they are the ones closest to programs and policies that often beg for objective assessments of impacts and outcomes. Practitioners conducted only slightly more evaluative research than academics. Practitioners conducted a larger percentage of the descriptive research than academics. And this difference is significant (=4.836, df=1, p=0.028). Academics conducted significantly more exploratory research than practitioners (=3.881, df=1, p=0.049). There was too little explanatory research to determine significant relationships between authors of each type. The same must be said for evaluative research. 37.9 percent of the studies reflect two or more research approaches. The authorship of these multi-purpose studies is nearly equally practitioner and academic. Samples and Data Collection Methods In addition to recording methods used for data collection, we looked carefully at the descriptions of study participants and samples used in these studies. If samples were used we tried to determine if the sample was a randomized, probability sample, or a non-probability 15-(2/15/2016) sample (e.g., purposive, convenience, quota, etc.). Most authors described the participants and samples used in their studies. From these descriptions we were able to determine that only 17.7% (31) of the studies that used samples used randomized sampling procedures to obtain probability samples. There was little difference between academics and practitioners on the use of randomized samples (18.31% and 16.5%). Eleven data collection methods were identified in these studies (See Table 7) Many of the studies employed more than a single data collection method, for example, survey questionnaires and focus group interviews (multiple methods frequency = 84, 40.8%). Practitioner researchers appear to prefer survey questionnaires, observational methods, and, to a lesser degree, content analysis. Academic researchers use survey questionnaires in onequarter of their studies. More frequently than practitioners, academics use content analysis, structured and semi-structured interviews, and bibliometric methods. Academic researchers employed multiple data collection methods significantly more often than practitioners (=5.183, df=1, p=0.023). Data Analysis Method, Quantitative or Qualitative The research studies in our sample were coded as quantitative, qualitative, or “triangulated” (used both methods) with regard to the type of data analysis used. Quantitative analysis is used when the data collected is categorical (counts) or some other form of numerical data. Qualitative analysis is used when the data collected is verbal, pictorial, or textual. Following Swygart-Hobaugh [18], we coded a study as quantitative if it used descriptive statistics (frequency counts, percentages, means, etc.) or inferential statistics in tests of significance (chi square tests, regression analysis, ANOVA, etc). We coded studies as qualitative if they employed, for example, participant observation, in-depth interviews, focus groups, or content analysis. These methods typically yield verbal and textual data. Table 8 displays the frequencies for each type of data analysis. Unmixed, qualitative-only studies make up only 16.99 percent of the studies. However, an additional 29.61 percent of the studies included some qualitative data analysis, and this brings the total percentage of studies employing qualitative analysis to nearly 47 percent. Academic researchers conduct more qualitative research than practitioners, but the difference is not statistically significant. 16-(2/15/2016) Statistical Analysis Techniques Used Statistical analyses employed in these studies were encoded as descriptive, correlational, inferential (including regression analysis), or some combination of these techniques. Table 9 displays the frequencies of these types of statistical analyses in the studies reviewed. Studies having no statistical analyses were excluded from these counts. The frequent use of descriptive statistics was anticipated. The low level of chi square and correlation coefficient analysis was not anticipated. The data for conducting such analyses was often present. Academic researchers did conduct more of this type of statistical analysis (=9.558, df=1, p=0.002), and their studies more frequently employed two or more types of statistical analysis (=9.180, df=1, p=0.003). Research Report Variables Each of the published research reports in this study were analyzed and evaluated for content and presentation. Ten factors were identified for this purpose (See Table 10). With the exception of “Visual Data Presentation,” these are binary (Yes/No) observations. The factor was present or not. To avoid subjective assessments in the initial analysis, we did not make judgements as to quality of description. One possible exception to this could be the first factor, “Is the report well-organized with appropriate sections easily identified?” The quality of the presentations varied greatly. If anything, we erred in the direction of being too liberal in our assessments of content presentation. The data seem to support this. Almost all reports were judged “Yes” on the first six report factors. Some research questions and aims were present but poorly stated, and many literature reviews were token attempts to cite something relevant. In some cases, descriptions of samples and data collection procedures were not sufficient to support replication of the studies. However, if the factor was described, it was marked as “Yes” for present. A future study could use a more refined assessment scale that would be applied to published reports by a panel of expert judges. This would improve the validity of such assessment attempts. Having said this, our humble opinion is that research reports in librarianship have improved considerably in recent years. Improvements have been made in content, organization, and readability. Our data do indicate there is room for improvement in report organization on the part of practitioners. There has been noticeable 17-(2/15/2016) improvement in the discussions of data-supported findings. Especially evident is the increased use of visual data formats such as tables, figures, and charts. It is a rare thing to encounter visual displays of data 20-30 years ago. We turn now to three report factors we group together in a category we call the “Humility Factor.” This characteristic is sadly absent in a majority of research reports. As Table 10 indicates, not even one-third of the reports addressed the critical issues of validity and reliability. Furthermore, nearly two-thirds of the published reports made no mention of study limitations or shortcomings. Recommendations for needed, pertinent research on the topic investigated are found in only slightly more than one-half of the reports. Comparing practitioners and academics on the Humility Factor, we found no difference. Editors and reviewers must share some of the responsibility for this. DISCUSSION From our sample of 206 articles from 23 LIS journals, it is clear that there is a considerable amount of systematic, empirical research being conducted by practitioners. 97 (47.1%) of the articles are practitioner-only in authorship, 89 (43.2%) are by academics and 20 (9.71%) have mixed authorship. The data indicate significant collaboration among colleagues in the same profession, but collaboration between academics and practitioners is rare. Only 9.71% of the studies had mixed-type authorship. This low level of collaboration is also reflected at the journal level. Journals tend to be predominantly practitioner or predominantly academic. Perhaps the recommendations of Van House, that practitioners publish in practitioner journals, have borne fruit. Measured by having a large majority of authors of one type or another, six journals can be classified as academic in authorship and ten journals can be classified as practitioner in authorship. We have indicated these journal types in Appendix A. The six top journals that produce the greatest number of research articles are one-half practitioner and one-half academic. These findings are consistent with the findings of the Schoegl and Petschnig study of 38 LIS journals. Based on a survey of editors, they found that authorship distributes nearly equally between academics (‘scientists’) and practitioners. They conclude that “a clear distinction can be made between journals with predominantly publishing scholars and those with a higher share of writing practitioners.”[19] Our study 18-(2/15/2016) found no difference between academic journals and practitioner journals in the percentage of published articles having mixed authorship. Many and diverse topics are addressed in these research studies. It seems no economical taxonomy of research topics can adequately document the variety of topics being addressed. A list of unique topics addressed in our sample studies would exceed 25 in number. Our “other” topics category is well-populated. New topics appear (e.g., Web searching or usability, chat reference, digital repositories), and interest in earlier topics wanes (library history, integrated library systems). When topics are collapsed into the four general areas, library operations and services, use and user studies, LIS education, and other, the data show a significant difference in topic areas researched by practitioners and academics. Practitioners more frequently investigate aspects of library operations, collections and services. Academics more frequently investigate use and user problems and issues. This difference is significant (=10.835, df=1 , p=0.001). With regard to location and setting of research, academics conduct most of the public library-related research and practitioners conduct nearly all of the academic library-related research. We believe we have contributed to the literature on library research methods by making distinctions between research type or approach (based on intention, purpose or strategy), data collection methods, and types of data analysis. Studies may be exploratory in aim, but include both qualitative and quantitative data collection and analysis methods. Studies may be descriptive in aim, yet include qualitative elements. We believe the type of research to pursue depends as much on the intention and purpose of the researcher as on the nature of the data considered relevant for this pursuit. It is often said that the data should dictate the method to be used. We say that the type of research the researcher wishes to pursue to satisfy a particular research objective (exploratory, descriptive, explanatory, evaluative) as seen by that researcher, should determine the choice of data collection method and type of data analyses. In effect, we are saying the researcher is primary. We have found that most exploratory research is conducted by academics. Exploring reasons why practitioners conduct little exploratory research is a much-needed research study. 19-(2/15/2016) When samples were described in these studies, we found two disturbing trends. Only 31 (17.7%) reported the use of randomized, probability samples. Without such samples, no claims can be made about the representativeness of the sample or application of findings to the larger population. Thus, most library-related research has no external validity and the findings can only be viewed as preliminary and tentative. Follow-up studies that address these limitations are simply not found in the literature. Academics do not grade better on this factor than practitioners. In addition, reported survey response rates were very low (seldom more than 30%), and rarely was there any analysis of the constituents of the non-response group and how this might affect the value of the sample findings. Join these findings with the infrequent discussions of validity issues and study limitations, and we have identified a major area of concern and needed improvement in library-related studies. Survey questionnaires and interviews continue to be the most used methods for data collection (38.3%), followed in frequency by content analysis (20.4%) and semi-structured interviews, including focus groups (10.2%). A variety of methods were used in the remaining studies. Practitioners employed proportionally more survey questionnaires and observational techniques than academic researchers. The latter employed proportionally more survey interviews, semi-structured interviews, and content analyses than practitioners. However, none of these differences were statistically significant. Less than one-half of the studies employed two or more data collection techniques (40.8%). Academic researchers used multiple methods significantly more often than practitioners (= 5.183, df=1, p=0.023). We characterized the type of data analysis used in the studies as quantitative, qualitative, or both. Only 17 percent of the studies were solely qualitative, but an additional 30 percent combined qualitative and quantitative methods. Thus, nearly one-half of the studies had a qualitative component. As for the use of statistical techniques, academics employ correlational and combined techniques far more often than practitioners (= 11.28, df=2, p=0.004 and = 12.11, df=2, p=0.003).). With regard to the quality of published reports, compared to 10-20 years earlier, there has been undeniable progress. Journal editors and reviewers deserve a large amount of the credit 20-(2/15/2016) for these improvements in the published accounts. The majority of the reports are detailed, comprehensive, and well-organized. The greater use of tables and graphic displays is welcome. The news is not all good. Discussions of validity and reliability, study limitations, and future research are unacceptably low. More attention must be given to these factors by researchers, reviewers, and editors. CONCLUSION We have identified significant differences between practitioner research and academic research. What most distinguishes these researchers and their research studies? As it turns out, not a great deal. There is little difference in the quality and organization of their published reports. With regard to topics investigated, practitioners conduct more library-specific studies and academics conduct more use and user studies. This topical gap is not wide, and there is cross-over, if little collaboration between academics and practitioners. A closer look at the frequencies for “information seeking” and “database searching” combined reveals that academics account for 59.1 percent of theses studies, practitioners account for 27.3 percent, and the remainder (13.64 percent) have mixed authorship. Academics appear to be more sophisticated in the use of methods of data collection and analysis, but this gap may be closing. At this time, academics conduct more qualitative studies than practitioners, and more often use multiple methods of data collection and statistical analysis. Nothing in our review of the 2003-2005 research literature brings into question earlier predictions by Fidel, Powell, and others. Our review confirms that qualitative research methods are being used to investigate a variety of research topics. However, the use of qualitative methods may have leveled off, and the low use of qualitative methods by practitioners should be a concern of LIS educators and researchers. Whatever type of library or information center they are employed in, practitioners are closest to the “lebenswelt,” the world as perceived and engaged in by those who we so often study. Practitioners, to a greater or lesser degree, are participants in this world, yet participant observation and other qualitative methods have a low use in practitioner studies. Do we know why this is the case? Can it be attributed to lack of time, or lack of training in qualitative methods? Would there be more qualitative research conducted if there were more collaboration between academics and 21-(2/15/2016) practitioners? How can such collaboration be encouraged and supported? Who should take the lead? It is precisely this kind of collaboration that is encouraged by Simon and Schlichting in their report on a successful collaboration between public librarians and university professors in the design and administration of a user survey.[20] This case is also interesting because the professors who contributed research expertise to the project were from journalism and sociology departments, not LIS faculties. We suspect that every published researcher suffers from the “nagging doubts syndrome” at the close of a study. Have we missed anything? Did we have a good sample for our purposes? Are our calculations accurate? Should we have approached this question in an entirely different manner? Such reflections are an essential component of the research adventure. We would like to see more evidence of such reflections in the library research literature. We are aware of the limitations of our study. The selection of journals for our sample frame was purposive, as our aim was to include several top academic journals and a broad selection of practitioner journals. Another set of journals might well yield a different distribution of research articles. Furthermore, our 51.4 percent sample of articles (206 of 401), although selected randomly, might not be representative of the whole group. We believe there is a high probability that they are representative of the research published in these 23 journals during this period. We do have doubts about our list of data collection methods. This group emerged inductively from the reports, but our named categories may benefit from greater delineation. This might be an appropriate project for a Delphi study by methods experts. The problem is similar but more complex with regard to research topic naming and classification. This is a slippery, ever-changing terrain. LIS researchers are branching out, topically speaking, into many new domains. We refer this task to the knowledge organization experts in our profession. In the meantime we think it is best to document the topics as they appear in the literature. Several areas for future research have been mentioned in this report. We will not attempt to prioritize them, but a few tug hardest on our coat-tails. We would like to see more qualitative, exploratory research on practitioner research (or lack thereof) to investigate barriers to qualitative research in this domain. We would like to see more qualitative, exploratory 22-(2/15/2016) research on the low level of collaboration between academic scholars and library practitioners. We know that it is low, but we do not know why it is low and how it might be increased. Are there meaningful comparisons to be discovered in the health sciences and professions? Additional problem areas that could benefit from qualitative research need to be identified. Lastly, would not it be interesting to learn of any developments in LIS graduate curricula in response to study findings like those reported here? REFERENCES 1. Hildreth, Charles R. "How Are They Going About It? A Comparison of Research Methods Used by LIS Academic and Practitioner Researchers." Peer-reviewed and juried paper presented on January 22, 2003 at annual meeting of the Association for Library and Information Science Education (ALISE), January 21-24, 2003, Philadelphia, PA. (Available at: http://myweb.cwpost.liu.edu/childret/practitioner-res.ppt) 2. Powell, R. R. & Connaway, L.S. (2004). Basic research methods for librarians. 4th ed. Libraries Unlimited, Westport, Connecticut, and London. 3. 3. Hernon, P. & Schwartz, C. (1999). Editorial: LIS Research – Multiple Stakeholders. Library & Information Science Research. 21(4), 423-427. 4. Hernon, P. & Schwartz, 423. 5. Ennis, P. H. (1967). Commitment to Research. Wilson Library Bulletin, 41, 899-901. 6. Grotzinger, L. (1981). Methodology of Library Science Inquiry – Past and Present. In C.Busha (Ed.), A Library Science Research Reader and Bibliographic Guide (pp.38-50). Littleton, CO: Libraries Unlimited. 7. Busha, C. H. & Harter, S. P. (1980). Research Methods in Librarianship: Techniques Interpretation. Academic Press, Inc. & 8. Van House, N. A. (1991). Assessing the Quantity, Quality, and Impact of LIS Research. Library & Information Science Research, Norwood, NJ: Ablex, 85-100. 9. Fidel, R. (1993). Qualitative Methods in Information Retrieval Research. Library & Information Science Research. 15, 219-247. 10. Fidel, R., (1993) 233. 11. Powell, R. R. (1999). Recent Trends in Research: A Methodological Essay. Library & Information Science Research, 21(1), 91-119. 23-(2/15/2016) 12. McKechnie, L. E.F. & Baker, L. & Greenwood, M. & Julien, H. (2002). Research Method Trends in Human Information Behaviour Literature. Paper presented at: ISIC-2002: Information Seeking in Context. The Fourth International Conference on Information Needs, Seeking and Use in Different Contexts. Lisbon, Portugal, September 11 – 13. 13. Watson-Boone, R. (2000). Academic librarians as practitioner-researchers. The Journal of Academic Librarianship. 26(2), 85-93. 14. Koufogiannakis, D. & Slater, L. & Crumley, E. (2004). A content analysis of librarianship research. Journal of Information Science. 30(3), 227-239. 15. Hildreth, C. (2003). 16. Hernon, P. (1991). The Elusive Nature of Research in LIS. In C. McClure and P. Hernon (Eds.), Library and Information Science Research: Perspectives and Strategies for Improvement (pp.3-14). Norwood, NJ: Ablex. 17. Schutt, R. (2006). Investigating the Social World: The Process and Practice of Research, 5th ed. (13-17). Sage Publications. 18. Swygart-Hobaugh, A. J. (2004). A Citation Analysis of the Quantitative/Qualitative Methods Debate’s Reflection in Sociology Research: Implications for Library Collection Development. Library Collections, Acquisitions, & Technical Services, 28, 180-195. 19. Schloegl, C. & Petschnig, W. (2005). Library and Information Science Journals: An Editor Survey. Library Collections, Acquisitions, & Technical Services, 29, 4-32. 20. Simon, J. & Schlichting, K. (2003). The College Connection: Using Academic Support to Conduct Public Library Surveys. Public Libraries, November/December, 375-378. 24-(2/15/2016) APPENDIX A: Journal Titles for 2003-2005 Study of Research Methods: TITLE ARTICLES ARTICLES MINED SELECTED Behavioral & Social Sciences Librarian @ 7 4 Canadian Journal of Information & Library Science @ 18 9 Cataloging & Classification Quarterly 16 8 Collection Management 6 3 College & Research Libraries 4 2 Information Technology and Libraries % 21 11 Journal of Academic Librarianship % 69 35 Journal of Documentation @ 48 24 Journal of Librarianship & Information Science 21 11 Journal of Library Administration % 10 5 Library & Information Science Research @ 33 17 Library Collections, Acquisitions & Tech. Services % 18 9 Library Resources and Technical Services % 23 12 Library Trends @ 16 8 Public Libraries 10 5 Public Library Quarterly @ 12 6 Public Services Quarterly % 3 3 Reference & User Services Quarterly 17 9 Reference Librarian 8 4 Reference Services Review % 12 6 8 Research Strategies % 4 Serials Review % 10 5 Technical Services Quarterly % 11 6 TOTALS 401 206 Journals in bold type had two or more research articles per issue. @ - Predominantly academic journals % - Predominantly practitioner journals 25-(2/15/2016) APPENDIX B: Study and Report Variables STUDY VARIABLES 1. Journal Title 2. Article Year of Publication 3. Number of Authors 4. Practitioner or Academic Author 5. Topic-1 6. Topic-2 7. Location of Study (specific site or site-independent) 8. Setting/Context of Research (Academic, Public, Special Lib., or Misc.) Research Approach: 9. Descriptive 10. Explanatory 11. Exploratory 12. Evaluative 13. Multiple/Combined Approaches Sample & Research Methods: 14. Sample Type 15. Method-1 16. Method-2 17. Method-3 18. Multiple Methods Used (2 or more) Data Analysis Type: 19. Quantitative 20. Qualitative 21. Both Statistical Analysis Employed: 22. Descriptive 23. Correlation (variable associations) 24. Inferential (significance tests, linear regression) 25. Multiple Analyses Used 26. Not Applicable to this study REPORT VARIABLES 27. Well-organized? 28. Research Question/Problem/Aim Stated at Beginning? 29. Literature Review? 30. Data Collection Explained? 31. Data-supported Presentation of Results? 32. Visual Presentation of Data Findings? 33. Validity/Reliability Issues Addressed? 34. Limitations of Study Mentioned? 35. Future Research Recommendations Discussed? 26-(2/15/2016) Table 1. Research Topics TOPIC Admin/Mgmt/Ops Systems/IT Collections Services Users (needs, preferences, behavior) Info. Seeking/use Database searching/IR Catalogs/indexes, etc Publishing Citation Analysis Cooperation/Consortia LIS Education User Education/Info. Literacy Other Frequency 36 (9.55) 33 (8.75) 40 (10.61) 45 (11.94) 49 (12.00) 42 (11.14) 24 (6.37) 26 (6.90) 17 (4.51) 14 (3.71) 4 (1.10) 9 (2.39) 24 (6.37) 14 (3.71) Table 2. Topic Categories CATEGORY Frequency Library Ops. & Services 158 (41.91) Use & User Studies 139 (36.87) LIS Education 9 (2.39) Other 71 (18.83) 27-(2/15/2016) Table 3. Contingency Table: Topic Category by Author Type FREQUENCY| TOT PCT | ROW PCT |Lib. Use/ LIS COL PCT |Ops. |User |Educ.|Other| -----------------------------------| 13| 19| 0| 5| Mixed AU | 3.4| 5.0| .0| 1.3| | 35.1| 51.4| .0| 13.5| | 8.2| 13.7| .0| 7.0| -----------------------------------| 54| 69| 6| 35| Academic | 14.3| 18.3| 1.6| 9.3| TOTAL 37 9.8 164 43.5 | 32.9| 42.1| 3.7| 21.3| | 34.2| 49.6| 66.7| 49.3| -----------------------------------| 91| 51| 3| 31| 176 Practitioner| 24.1| 13.5| .8| 8.2| 46.7 | 51.7| 29.0| 1.7| 17.6| | 57.6| 36.7| 33.3| 43.7| -----------------------------------TOTAL 158 139 9 71 377 41.9 36.9 2.4 18.8 100.0 Statistic DF Value p-value ------------------------------------------Chi-Square 6 17.454 0.008 Table 4. Research Location Type Frequencies Specific Site Site-independent 83 (40.29) 123 (59.71) Table 5. Research Setting Type Frequencies University Public Library Special Library Misc. other 115 28 8 55 (55.83) (13.59) (3.880) (26.70) 28-(2/15/2016) Table 6. Research Type Frequencies Descriptive Explanatory Exploratory Evaluative Multiple 158 26 68 36 78 (76.70) (12.62) (33.01) (17.48) (37.86) Table 7. Data Collection Method Frequencies By Authorship Data Collection Method Survey Questionnaire Survey Interview Observation Semi-structured Interview Experiment Computer Log Analysis Bibliometric/Citation Anal. Content Analysis Case Study Delphi Others TOTAL *Row Percentages Mixed Authorship 12(12.6)* 4(16.0) 0(0.0) 4(12.5) 4(21.1) 4(21.1) 1(5.6) 4(6.3) 1(10.0) 0(0.0) 0(0.0) 34 Academic Author(s) 35(36.8) 14(56.0) 7(38.9) 16(50.0) 8(42.1) 8(42.1) 11(61.1) 35(54.7) 4(40.0) 0(0.0) 6(46.2) 144 Practioner Authors 48(50.5) 7(28.0) 11(61.1) 12(37.5) 7(36.8) 7(36.8) 6(33.3) 25(39.1) 5(50.0) 1(100.0) 7(53.8) 136 TOTAL 95 25 18 32 19 19 18 64 10 1 13 314 Table 8. Type of Data Analysis Frequencies Quantitative Qualitative Combined 110 (53.40) 35 (16.99) 61 (29.61) Table 9. Statistical Analysis Frequencies Descriptive Correlational Inferential Combination Not Applicable 171 44 32 57 36 (83.01) (21.36) (15.53) (27.67) (17.48) 29-(2/15/2016) Table 10. Study Report Variable Frequencies Variable Well-organized Research Quest./Aim Literature Review Participants/Sample Data Collection Data-supported Findings Validity or Reliability Study Limitations Future Research Visual Data – Tabular Visual Data – Graphic Visual Data – Both Visual Data – None Visual Data – NA Yes 162(78.64) 183(88.83) 172(83.50) 183(88.83) 199(96.60) 191(92.72) 56(27.18) 72(34.95) 108(52.43) 92(44.66) 23(11.17) 48(23.30) 16( 7.77) 27(13.11) No 44(21.36) 23(11.17) 34(16.50) 23(11.17) 7( 3.40) 15( 7.28) 150(72.82) 134(65.05) 98(47.57) 30-(2/15/2016)