Half a Century of NVSQ: Thematic Stability Across Years and Editors

Introduction

Some academic fields have a long tradition of disciplinary development and their journals are the focus of their respective scholars. Others are more interdisciplinary and in the process of assuming a clear identity and focus. The study of the nonprofit and voluntary sector in all its forms is an evolving one. As we note below, starting from a common interest of a few scholars in a diverse set of disciplines, it has become a focal field itself. The study of the nonprofit and voluntary sector has both American and international associations of scholars that publish in many journals. There are a growing number of academic programs awarding either certificates or degrees in nonprofit management, leadership, and studies.

The field produces numerous academic articles. Ma and Konrath (2018) reported that they identified 12,016 bibliographic records from 19 journals published between 1925 and 2015. ¹ Brass et al. (2018) and Brudney and Durden (1993) demonstrated that over time the interest in certain topics can grow or decrease. Even within governance, for example, Brass et al. (2018) found shifts among six subcategories of governance over time. Brudney and Durden (1993) found that in the 20 first years of Nonprofit and Voluntary Sector Quarterly (NVSQ), articles addressing motivation, recognition, and retention of volunteers were the dominant topic of that era, though their share in the last four studied years went down. They also reported a dramatic increase in studies reporting fund/resource raising.

Indeed, much knowledge has been produced in this emerging academic field. In this article, we are concerned with the stability versus changes in the topics covered in the field. We used NVSQ as our case study, including the precursor of this journal with different names. NVSQ is the longest running academic journal devoted to all aspects of nonprofit studies and has the highest impact factor among nonprofit sector journals. We wonder if, in the 48 years since its inception, topics covered in the journals changed over time or the focus stayed relatively stable.

In the next section, we review the development and scope of scholarship related to nonprofit studies. This is followed by documenting the journal’s history and establishing three eras that will serve as the basis for the methodology of the article. In the methods section, we review our methods and explain our use of structural topic modeling (STM; Roberts et al., 2014), a family of topic modeling recently developed and applied in the fields of natural language processes (NLP) and machine learning (Blei, 2012; Blei et al., 2003). Next, we present our findings resulting from the STM analyses. In the summary and conclusions section, we discuss the few topics that saw change over time and note that overall it is a slow change amid wide range continuity.

The Nonprofit Sector and Its Scholarship

Peter Dobkin Hall (2006) suggested that nonprofit organizations and voluntary activities existed for generations but only matured into a recognized sector in the mid-1970s. In the United States, this period was marked by the Filer Commission and the formation of the Association of Voluntary Action Scholars (AVAS), the precursor to the Association for Research on Nonprofit Organizations and Voluntary Action (ARNOVA). NVSQ, along with its precursors (JVAR: Journal of Voluntary Action Research), is the journal for these key professional organizations.

To some extent even now, but more so in the first 20 years after the formation of AVAS, most scholars in the field were anchored in traditional academic disciplines with very few nonprofit scholars or departments (Katz, 1999). Jackson et al. (2014, p. 798) suggested that “The field of nonprofit and philanthropic studies crosses multiple boundaries, making its coherence less obvious and less definable.” Not surprisingly, Brass et al. (2018) found that most articles on nongovernmental organizations (NGOs) between 1980 and 2014 were published in interdisciplinary journals and not in one or a few cohesive journals. Among the leading journals are NVSQ, Nonprofit Management and Leadership (NML), and Voluntas. Brass et al. focused only on nonprofits that work internationally (NGOs). They concluded that NGO research has become methodologically more rigorous over time.

The study of the nonprofit sector is inherently interdisciplinary and includes scholars from various disciplines. In the first issue of NVSQ, founding editor David Horton Smith (1972, p. 2) noted,

This is the first issue of a scholarly journal representing a new intellectual endeavor—an attempt by people with a serious intellectual interest in some aspects of the various forms of voluntary action to develop an interdisciplinary, interprofessional, and international forum for their research, theory, and policy thinking.

In fact, Smith argued that the voluntary action field is relevant to some 30 different fields of study.

The field started with having a few elective courses in some universities and grew rapidly. Mirabella (2021) reported 292 colleges and universities with courses in nonprofit management. In addition, Mirabella found 92 universities offering one or two graduate courses, usually financial management and generic nonprofit management. Even more impressive is the fact that “As of March 2009, there were one-hundred and sixty-eight colleges and universities with graduate degree programs, throughout the United States, that have a concentration in nonprofit management” (Mirabella 2021, p. 1). In 1991, the Nonprofit Academic Centers Council (NACC) was established. In 2018, NACC listed 52 centers that taught and gave a degree/certificate in nonprofit studies (ranging from leadership, management to governance and public policy). The field of nonprofit education saw tremendous growth and so knowledge production in the field grew at the same pace.

Studying Nonprofit Academic Publications: A Review

Given the diversity of the field and recent development of formal nonprofit management programs, some studies of nonprofit scholarship (cf. Brass at al., 2018) searched through a large set of journals to find articles dealing with NGOs and used them as a sampling framework. Similarly, Allison et al. (2007) and Jackson et al. (2014) studied academic articles published in English and a wide range of peer-reviewed journals focusing on keywords such as nonprofit organizations, third sector, philanthropy, civil society, voluntary associations, voluntary sector, voluntarism, charitable organizations, grassroots organizations, community based organizations, nongovernmental organizations, and NGOs. ² Brass et al. (2018) covered 5 years (2000–2005) and the latter authors (Allison et al., 2007; Jackson et al., 2014) added articles from 2000 to 2011. These authors, however, did not analyze the articles’ contents; rather, they documented the immense growth in academic productivity between the first and second periods.

Jackson et al. (2014) also studied nonprofit/philanthropy dissertations. A more comprehensive attempt to examine dissertations was carried out by Shier and Handy (2014). Shier and Handy confirmed the interdisciplinary nature of the field and suggested five key themes in the studied 3,790 dissertations. They reported five main themes: (a) resources (human and financial); (b) organizational effectiveness and performance; (c) organization development (context, processes, and culture); (d) intra-organizational context (leadership, structure, etc.); and (e) interaction and collaboration (with other organizations, government, etc.).

Ma and Konrath (2018) might be the only study to conduct advanced quantitative analysis of knowledge production in the field. These authors included all relevant publications from 1925 to 2015. They also provided an analysis of articles in NVSQ, NML, & Voluntas for the years 1986 to 2015 that included 2,848 bibliographical records. They concluded that “scholars in this field have been actively generating a considerable amount of literature and a solid intellectual base for developing this field towards a new discipline” (Ma & Konrath 2018, p. 1). Ma and Konrath reported an explosion in the scope of nonprofit research especially since the 1980s. They also focused on field cohesion. Academic cohesion “examines whether the literature has formed several inter-connected research themes that can distinguish this research field—a prerequisite for forming its disciplinary identity” (Ma & Konrath 2018, p. 2).

Others focused on one journal and covered a specific time frame using qualitative approaches (cf. Brudney & Durden, 1993, who studied JVAR/NVSQ between 1972 and 1992; Bushouse & Sowa, 2012, who studied policy implementation for practice in NVSQ between 2000 and 2010). As such, this article is not the first study analyzing articles regarding the nonprofit sector. However, this is the first attempt to analyze the first 48 years of a key journal since its inception and assess the stability and change of thematic topics.

This research is also unique for quantitative analysis using structural topic model (STM; Roberts et al., 2014), which allows researchers to detect “latent topics to be automatically inferred from text” (p. 1066) with the assistance of machine-learning techniques. One recent example of using this method is Hopkins (2018) analyzing the rhetoric of the health care debate. When studying linguistic texts, latent topics can be viewed as groups of related words. For example, in the field of organizational culture research, Schmiedel et al. (2018) used 428,492 reviews of Fortune 500 companies from the online platform Glassdoor, where employees can evaluate organizations. From these reviews, they inductively identified topics that mattered to employees and explored their relationship with employees’ perception of organizational culture. In finance, topic modeling was used to extract textual risk disclosures from annual reports (Bao & Datta, 2014). Similarly, Grothe-Hammer and Kohl (2020) used STM to analyze the change in focus over time of organizational sociology within the field of sociology.

This study focuses on NVSQ to understand the cohesion in the field, its thematic continuity, and new thematic trends. Others can use the findings from this study to assess what areas of inquiry are underrepresented and need extra attention. Another contribution we offer is analyzing knowledge production through one journal only for the longest period of time. As noted above, to date the studies of knowledge production from one journal in the nonprofit field ranged from covering 5 years to 18 years.

The 48 years that we cover in this study are not equal in many ways. To try and compare an article from 1975 with one from 2015 would be somewhat misleading. The number of people studying nonprofits grew dramatically between these 2 years, the number of manuscripts submitted grew, over time the expected statistical rigor intensified, interest from international scholars increased, and many historical events took place within and outside the field. To that end, our first aim was to assess key periods in the history of NVSQ.

NVSQ History and Periods ³

In this section, based on ARNOVA documents and personal discussions with three of the journal editors, we discuss the history of NVSQ with the intent of dividing the journal timeline into three distinct eras. These three eras will be the basis for analyzing longitudinal shifts in nonprofit scholarship focus.

The first three editors of the journal when it was published under the title of Journal of Voluntary Action Research (JVAR) were David Horton Smith, Richard D. Reddy (for 2 years), and later Jon Van Til. Combined they are responsible for the first 21 volumes and 469 articles of the journal (Vol. 1–21). These editors included the organizational founder and committed members who agreed to take over the journal. The organization known as Association for Voluntary Action Scholars (AVAS) was small in scale (about 200 members), very informal, and with no paid staff. The journal was initially self-published by Boston College Press, then published by Transaction Periodicals Consortium and then by Jossey-Bass Publishers (Smith, 2003). When Brudney and Durden (1993) analyzed the first 20 years of JVAR/NVSQ, they found that sociology was the discipline represented most heavily, although scholars from 35 disciplines have provided articles. The second most represented discipline was social work. They also found that 17% of the authors had a non-academic affiliation. We can call this 21-year period of the journal “the era of the founder editors.”

The second era of the journal started in 1992 with the first use of ARNOVA search committees to select the editor, suggesting further professionalization of the journal. In the early 1990s, the organization underwent a massive strategic planning initiative. As of 1992, paid staff were hired, and the topics of the conferences were widened to include civil society, philanthropy, and organizational behavior. The organization grew to about 600 members and an office was established in Indianapolis, IN. The journal’s name was changed to Nonprofit and Voluntary Sector Quarterly (NVSQ) and the publication of the journal was transferred from Jossey-Bass to the more prestigious Sage Publications. Carl Milofsky became the fourth editor in 1992 (volume 22) and 6 years later he was succeeded by Steve Rathgeb Smith. Together, these two editors served till 2004 covering volumes 22 through 33. Combined they are responsible for 280 articles. We can call this 12-year period of the journal “the era of committee selected editors.”

The third era of NVSQ development started when it became clear that the field of nonprofit studies was growing, and one editor would not be able to cope with the demands of the job. The organization’s membership exceeded 1,200 members and dozens of new academic nonprofit/philanthropy teaching programs as well as research centers were established in the United States and worldwide. This period also saw proliferation of nonprofit-related journals some of which focused on more specific aspects of nonprofit studies such as management and leadership, policy, teaching, and marketing. In 2009, due to an increased interest in the field with a concomitant increase in manuscripts submitted, NVSQ expanded from four to six issues per year, still retaining the name “Quarterly.” The journal ranking based on its impact factor placed it at the top group of all nonprofit and public administration journals (Bernick & Krueger, 2010). As such, in 2004, an ARNOVA search committee selected the team of Dwight Burlingame and Wolfgang Bielefeld as co-editors. They were later succeeded by a team of three editors. The team included Jeff Brudney, Femida Handy, and Lucas Meijs. Each combination of multiple editors included at least one editor whose disciplinary background was public administration and one with a social science background. Combined, these two teams took the journal from volume 34 to volume 45. At the time of writing this article, new trio editors (Angela Bies, Chao Guo, and Susan Phillips) began editing responsibilities with three issues already published. In this article, we covered three volumes under their editorship. Together, the team editors are responsible for volumes 34 to 48 and 767 articles. We can call this 15-year period of the journal “the era of multiple editors.”

When viewing the above history of NVSQ, it is logical to view the journal evolution as composed of three eras: “the era of the founder editors” (vol. 1–21; 1972–1993); “the era of committee selected editors” (vol. 22–33; 1993–2003), and “the era multiple editors” (vol. 34–48; 2003–2019). ⁴ However, when assessing the development of knowledge one can also look at individual editors, number of issues of the journal published in each year (started with two and now six), the year when the journal title was changed, and the academic backgrounds of the editors (specifically, sociology, political science, public administration, and others).

Based on a review of nonprofit and philanthropic programs in the United States, Mirabella et al. (2019) concluded educational programs reflected convergence among course offerings by disciplinary orientation (such as public administration) and homogeneity among curricular offerings for each disciplinary group (such as business and social work). Overall, based on the large heterogeneity of programs, they concluded that “The field of nonprofit management and philanthropic studies has yet to come into its own” (p. 63). If this is the case, the thematic topics published in NVSQ are likely to fluctuate over time and be based on the home discipline of the editor(s). Consequently, we sought to test whether the prevalence of thematic topics was related to the three time periods used to mark the journal’s history and to the academic backgrounds of the editors.

Methods

Step 1

Our first step was to place data from all 1,516 articles in the first 48 volumes of NVSQ into one big data file. We incorporated the articles’ title and abstract into this database. We could not use key words because in earlier volumes there were no key words. Table 1 reports our initial stage of analysis. On the left column is the period and years covered followed by the numbers of years, and the number of articles in each period. We then used a simple machine reading of counting terms to identify possible tokens, ⁵ where we obtained 169,937 different terms (see right column in Table 1). It was thus clear that simple machine reading of terms would be inappropriate.

OPEN IN VIEWER

Table 1.

Basic Descriptive Statistics for Our Textual Corpus.

Period	Years	No. of articles	No. of terms
P1 (1972–1992)	21	469	24,782
P2 (1993–2004)	12	280	32,385
P3 (2005–2019)	15	767	112,770
Total	48	1,516	169,937

We screened out numerous irrelevant terms such as “the,” “if,” “questions,” and so forth. Those words that are usually termed as stopwords were removed from the data. This elimination allowed us to save processing time and keep only items that were substantive to the study, but the reduced list was still sizable. Other scholars used their educated/subjective judgment to determine what topics to study. For example, Brass et al. (2018) employed 11 keywords that they referred to as sectors, based on common usage by NGO and other international development practitioners and academics (cf. Oxfam America and World Bank). In that way, they superimposed preexisting keywords on the data rather than allowing the data to speak to them. In our study, it was clear that given so many possible topics, we needed a sophisticated method to determine which topics are relevant and more commonly detected.

Step 2

For statistical text analyses of the 1,516 articles and 1,352 abstracts published in NVSQ, this study relied on structural topic modeling (STM; Roberts et al., 2014) that allows researchers to detect “latent topics to be automatically inferred from text” (p. 1066).⁶ STM is an extension of Latent Dirichlet Allocation (LDA; Blei et al., 2003) model which generates a series of words which latently represent underlying potential number (usually k) of topics in a collection of texts. However, compared with the LDA model, STM allows social scientists to test hypotheses about how the prevalence of a topic in a document is related to the document’s characteristics. In other words, STM is fitted for our research which aims to test how academic topics published in NVSQ transform across the predetermined three periods (before 1993, 1993–2003, after 2003). The logic behind topic models is the assumption that meaningful topics emerge out of relations between words rather than reside within single words (DiMaggio et al., 2013). Conceptually, STM is akin to exploratory factor analysis that takes a set of scores and places them together under a comprehensive title according to their interrelationships. However, unlike factor analysis that seeks the appropriate factor-item structure in numeric data, STM searches the optimally represented topics based on the textual similarities and differences between documents in textual data.

To determine the optimal number of latent topics in topic modeling, our study relied on Griffiths and Steyvers’s (2004) algorithm, which searches the log-likelihood of the topic model based on the identified number of latent topics (k). That is, how many different latent topics (groups of words) would best represent the material generated from the 1,516 studied articles. To determine the optimal number, a two-stage approach was taken. First, this study calculated model-fit statistics (i.e., log-likelihood) of each determined topic model whose number of topics ranges from 15 to 65, by taking the sequence of “five” (i.e., 15, 20, 25, . . . , 55, 60, 65). As shown in Figure 1, the magnitude of log-likelihood statistics became highest around 35 and then dropped and resurged around 45.

OPEN IN VIEWER

Figure 1.

The model-fit statistics: Identifying the optimal range of latent topics.

Figure 1 provided a general view suggesting that the optimal number of latent topics would be between 35 and 45. Figure 2 inspects each possible number of latent topics in the interval suggested by Figure 1. Based on the results found in Figure 1, this study examined model-fit statistics (i.e., log-likelihood) of each determined topic model whose number of topics ranges from 25 to 40, by taking the sequence of “one” (i.e., 25, 26, 27, . . . , 38, 39, 40). As shown in Figure 2, the log-likelihood statistics reached a maximal point when k = 37. Guided by Figure 2, we conclude k = 37 would be the optimal number of latent topics.

OPEN IN VIEWER

Figure 2.

The model-fit statistics: Finding the optimal number of latent topics.

Based on Griffiths and Steyvers’s (2004) algorithm-based results, STM with k = 37 was finally estimated. As briefly introduced before, the results of topic models, such as STM or LDA, provided researchers a list of words that latently represent underlying topics in corpus. Two nonprofit experts read the list of words per topic sorted by STM and came up with a semantic label on each topic. Readers may find the top 10 words classified for each topic in the appendix.

Step 3

In STM, each prevalence of the 37 thematic topics was analyzed by a linear regression analysis using the following covariates related to NVSQ articles: (a) three time periods defined above and (b) the editor’s academic backgrounds. Among the family of topic models, only STM can simultaneously estimate the relationship between topic prevalence and textual features in documents with a formal statistical significance test (Roberts et al., 2014). Table 2 shows how trends in the prevalence of thematic topics change across three time periods with a statistical confirmation. Based on STM findings, we also graphed the overall status of and the prevalence trends in thematic topics in NVSQ. STM and graphing of its estimated results were achieved by relying on STM (version 1.3.3), tidystm (version 0.0.0.90), and tidyverse (version 1.2.1) packages in R, an open-source statistical software.

OPEN IN VIEWER

Table 2.

Results Testing Difference in Topic Prevalence Across Three Time periods.

Source	Needy populations	Welfare agency & provision	Association & membership	Charitable giving	Managerial & administration issues	Democracy & political participation	Qualitative methods	Quantitative methods
Intercept	0.052*** (0.009)	0.055*** (0.009)	0.073*** (0.008)	−0.006 (0.011)	0.066*** (0.012)	0.051*** (0.009)	0.003 (0.010)	0.051*** (0.012)
Control variable: Editor’s background (baseline = other disciplines)
Sociology	0.003(0.011)	−0.024* (0.010)	0.036** (0.011)	0.009(0.013)	−0.001(0.015)	0.007(0.012)	0.000(0.012)	−0.015(0.015)
Political science	−0.008(0.007)	−0.011(0.007)	−0.012(0.007)	0.014(0.010)	−0.007(0.009)	−0.006(0.007)	0.006(0.009)	−0.006(0.010)
Public administration	0.005(0.012)	−0.002(0.012)	0.024* (0.011)	0.010(0.015)	−0.015(0.016)	0.001(0.013)	−0.002(0.014)	0.014(0.016)
Three time periods (Baseline = Period 1 [1972–1993])
Period 2 (1993–2003]	−0.030* (0.012)	−0.001(0.012)	−0.077*** (0.012)	0.027* (0.013)	−0.028(0.016)	−0.033** (0.012)	0.010(0.013)	−0.003(0.017)
Period 3 (2003–2019]	−0.029* (0.012)	−0.025* (0.012)	−0.069*** (0.011)	0.039** (0.014)	−0.034* (0.017)	−0.032* (0.013)	0.027* (0.014)	−0.039* (0.017)

Note. Unstandardized coefficients with their standard errors in parentheses. The number of simulated draws from the posterior distribution was set to 200. STM results testing the difference in prevalence for the other 29 or 30 topics were not reported because there was no statistically significant difference across three time periods. Detailed results for the other 19 topics can be obtained by request from the first author.

†

p < .10. *p < .05. **p < .01. ***p < .001.

The terms that are aggregated into a thematic topic are not always intuitively clear. A few articles dealing with a certain topic may also include tokens (words and phrases) that seem unrelated. Yet, these tokens may be aggregated together. As such the logic of naming a thematic theme is to search for the common ground of as many conceptually interrelated tokens as possible. In the case of thematic topic 18, we could not find any common thread and as such labeled it Miscellaneous.

Results

Before testing the relationships between prevalence of topics and characteristics of documents in the corpus, we examined the overall distribution of 37 latent topics and how each topic prevalence varies across years. First, the expected prevalence of total 37 topics across the whole period (1972—2019) is reported in Figure 3. These 37 thematic topics ranged from appearing in 1.85% (IT related topics) to 4.21% (Volunteerism 1) of the whole corpus. The following topics appeared at over 3%: “Volunteerism 1,” “Association & membership,” “Quantitative methods,” “Advocacy & empowerment,” “Needy populations,” “Government & public policy,” “Charitable giving,” “Welfare agency & provision,” “Managerial & administrative issues,” “Hybrid organizations,” “Various capitals,” and “Organizational mission & performance.”

OPEN IN VIEWER

Figure 3.

Prevalence of appearance of the 37 thematic topics in the first 48 years of NVSQ.

For each of the 37 thematic topics, Figure 3 reports its percent appearances in NVSQ articles. It should be emphasized that one article may contain more than one thematic topic. Yet, Figure 3 is a sort of measurement of the popularity or prevalence of academic thematic topics in NVSQ through the last half century.

The yearly change of 37 topics can be found in Figure 4. An eye-balling review of the 37 graphs of each thematic topic over the 48 studied years and with consideration to the three studied periods reveals few interesting trends. The time trend of each topic publication prevalence differs across topics. For example, some topics such as “Association & membership,” “Managerial & administrative issues,” “Democracy & political participation,” and “Needy populations” show a declining time trend, indicating academic papers discussing such topics were more likely to be published early on and less so in the later years of the journal. However, other topics such as “IT related issues,” “Organizational mission & performance,” “Charitable giving,” “Financial considerations 2,” and “Arts & cultural heritage” show an increasing trend, indicating more publications in the later years of NVSQ. Other topics show temporal stability in terms of prevalence of publication in NVSQ, such as “Generational issues,” “Government & public policy,” “Innovation & civic engagement,” and “Employment issues.” While it is informative to examine the whole distribution of 37 thematic topics’ prevalence and their year change, it is not an accurate assessment of change over time. There can be a year or two in which one thematic topic was popular followed by a few years of decline. Such aberration can be the result of a special issue or one editor’s intellectual pursuit. What would be more meaningful is comparing longer periods of time. To that end, we introduced and justified dividing the journal history into three studied time periods. Results testing differences in prevalence for all 37 thematic topics across the three time periods are provided in Figure 5 and Table 2. Table 2 includes also the possible impact of the editors’ academic disciplines as a control variable. Eight thematic topics have statistically significant differences across the three time periods. The other 29 topics whose STM results were not reported in Table 2 and are not colored red in Figure 5 showed no significant difference of topic prevalence across the three time periods. Put differently, of the 37 machine derived thematic topics of NVSQ articles, more than three-quarters (78.4%) showed no change in frequency of publications. The significant change in the prevalence of thematic topic publication is the exception rather than the rule.

OPEN IN VIEWER

Figure 4.

Yearly change of prevalence of 37 thematic topics in NVSQ.

OPEN IN VIEWER

Figure 5.

Statistical significance test of the thematic topics periodic prevalence using STM.

STM results revealed that six of the eight thematic topics reported in Table 2 were more frequently published before 1993, but only two thematic topics were more frequently published after the first studied period.

Specifically, STM results revealed that six topics reported in Table 2 were actively studied before 1993, but those topics received less attention after 1993. Specifically, when controlling editor’s academic backgrounds, probability of topic occurrence for “Association & membership” was significantly higher in the first period, significantly lower in the second period (b = −.077, p < .001) and significantly lower in the third period (b = −.069, p < .001). Such patterns were similar for “Needy populations” (b = −.030, p < .05 for Period 2; b = −.029, p < .05 for Period 3), and “Democracy & political participation (b = −.033, p < .01 for Period 2; b = −.032, p < .05 for Period 3). Despite a slightly different pattern, the topics of “Welfare agency & provision,” “Managerial & administrative issue,” and “Quantitative methods,” have been less studied after the second period (b = −.025, p < .05; b = −.034, p < .05; and b = −.039, p < .05, respectively), but there was no statistically distinguishable difference between the first and second time periods. Thus, these three thematic topics showed decline in prevalence only in the third period.

Unlike those six topics, two topics in Table 2 revealed increasing trends. Specifically, the thematic topic of “Charitable giving” was rarely studied between 1972 and 1993. It then increased in the second period (b = .027, p < .05), and finally further increased in the third period (b = .039, p < .01), which is statistically different from that detected at the first period. Similarly, “Qualitative methods” was also rarely apparent in the first period and continued to be low in the second period (b = 0.01, not significant) but significantly increased at the third period (b = .027, p < .05).

Regarding the editor discipline, as can be seen in Table 2, it had very little statistical impact. The few exceptions included “Association & membership” where editors from sociology and public administration were more likely to include such papers. Another exception was regarding “Welfare agency & provision,” a topic that was less frequent under editors who were sociologists. Overall, an editor’s discipline had little impact on thematic topics’ frequency. It is possible that as the flagship journal of nonprofit studies, submissions were from different academic disciplines focusing on different aspects of the nonprofit domain. As such, editors had to respect submissions from different disciplines and only limited ability to reshape the thematic topics featured in NVSQ in the shape of their own academic discipline.

Limitations

This study focused on one leading journal (NVSQ) and the one that has appeared continuously for almost half a century. Yet, our findings are limited to this journal and cannot be generalized to publications in non-English journals or any of the other 60 journals that David Horton Smith suggested are relevant to nonprofit studies. The field of nonprofit studies, as Ma and Knorath (2018) demonstrated, is significantly larger than NVSQ. However, NVSQ primacy and continuity may indicate how the field as a whole has evolved since 1972. Our division of the three NVSQ periods may be challenged by some readers who have a different interpretation of the journal and field development. The titles awarded to the 37 thematic topics were not easy to come by and some readers who carefully review the appendix may come up with different (perhaps better) thematic headers. Reliance on machine learning approach is becoming more and more common and assumed to be objective, yet it can ignore common sense and there may be critical issues/topics that some readers would feel missing as we relied on an automated content analysis. Finally, regarding the editor’s disciplines one can argue that their training is different from their actual discipline at the time of being an editor and that we could have grouped them into a smaller number of categories. These issues are open for future research.

Summary and Conclusions

We studied the evolution of thematic topics published in the field of nonprofit studies through the analysis of NVSQ’s first 48 years. We reviewed 1,516 articles that were published between 1972 and 2019. We divided the 48 years into three periods based on the journal’s levels of maturity and editor selection. These three periods were the era of the founder editors (vol. 1–21); the era of committee selected editors (vol. 22–33), and the era of multiple editors (vol. 34–48). To that end, we used structural topic modeling (STM). We obtained 37 distinct thematic topics. We then analyzed if there were significant changes over the three studied periods regarding the rate of publishing papers related to these 37 thematic topics.

The findings indicated that change and transition in the focus of nonprofit studies happens slowly and thematic stability is the norm. This stability likely reflects authors’ choice of path dependence. That is, novices and seasoned scholars studying nonprofit-related themes realize the thematic focus of previous papers in NVSQ, and accordingly submit similarly thematic-related papers to NVSQ. Whereas authors who assess their study topic to be different from the perceived NVSQ focus may deem their work beyond NVSQ thematic fields. This leads them to submit these papers elsewhere despite the editors’ willingness to review and be amenable to new thematic topics beyond NVSQ’s traditional. ⁷ Nevertheless, it is also an indication of academic maturity, where the focus is to drill down refinements and variations while the field itself has more established and coherent.

Stability may be a double-edged sword. On one hand, it reflects a field’s maturation and the existence of a committed group of scholars invested in this academic domain. On the other, it may hamper a disciplinary field from progress. Stability of a top journal may cause scholars to plan research projects that will fit the journal’s dogma and deter them from undertaking ground-breaking innovative approaches that may only be published in lesser journals. This issue cannot be determined by our data and should serve as a topic for future research.

The eight thematic topics that showed significant change over the three periods belonged to three key groups. The first group includes two thematic topics related to research methodology: “Qualitative method” and “Quantitative methods.” ‘Qualitative methods’ showed an increase through the periods and “Quantitative methods” showed a decrease. Interestingly, the change of these two thematic topics indicates that qualitative research has been growing over the years, at least in NVSQ. The second group is about one thematic topic: “Charitable giving.” It showed an increase pattern through the periods. This thematic topic is an indication that NVSQ—and ARNOVA—is moving away from focusing almost exclusively on voluntary action to incorporating the wider world of nonprofit studies.

Five thematic topics saw decline from the first studied era to later eras. Four of these topics are traditionally related to discipline of social work (“Needy populations” and “Welfare agency & provision”), discipline of sociology (“Association & membership”), and discipline of political science (“Democracy & political participation”). It is possible that these declines are the results of relatively lower number of ARNOVA members from these academic fields and the rise of new nonprofit programs that have lower focus on the declining thematic topics.

The decline in “Managerial & administrative issues” is notable. It is possible that the appearance and inclusion in social sciences citation index (SSCI) of NML directed more manuscripts of this thematic theme away from NVSQ. It is also possible that many international universities included NVSQ as a high-level journal in promotion reviews, which in turn increase submissions from non-administration faculty. Finally, while NVSQ is rated high among public administration journals, it is still one of many options for nonprofit scholars from public administration departments. Relatedly, nonprofit scholars in public administration departments and business schools still need to publish in traditional disciplinary journals and may choose NVSQ less frequently.

The overall stability and the few changes discussed above can indicate that the formative years of the journal helped solidify the field as a cohesive field of study. It may be that these trends are short-lived and that 10 years from now, a similar study would find that different topics entered the field and others may have declined. Regardless, in almost 50 years the field of nonprofit studies is not strongly fluctuating, and the literature seems more focused on fine-tuning than expanding new thematic topics.

Another important finding is that the journal editors did not bring their disciplinary focus with them to the journal. We found only two out of 37 thematic topics that were influenced by the editor’s discipline (“Association & membership” and “Welfare agency & provision”). This suggests that editors of the journal in its first 48 years saw nonprofit research as the field of study for NVSQ rather than their home disciplines. It is a sign of cohesion and maturity to this field. Our findings suggest that the editors’ disciplinary background did not seem to affect the journal thematic trajectory. As such, possible claims about an editor’s idiosyncratic influence of the journal thematic topics should be studied separately and maybe not even in the immediate future.

Upon the findings of this study, we recommend a few future research directions. One possible future study is concerning the globalization of the field and the journal. When and how more articles from non-USA countries started to be published and what frequency can be a new focus on future research. Similarly, the geographic location of editors may also have an impact on papers published. Another future research possibility is the impact of special issues, which are usually organized around specific coherent themes, on the progression of thematic topics. Finally, similar studies should be carried out to other leading journals in the field, such as Voluntas and NML.

Overall, we see these findings as indicating solidifying intellectual and professional boundaries in this interdisciplinary field. Contrary to Mirabella et al. (2019), who saw the field as diverse and fragmented based on the nonprofit training programs, our findings suggest centrality and stability of thematic themes throughout half a century. Our findings support nonprofit research as an evolving yet cohesive field of study.