Where the Great Cities Go, Do Other Cities Follow? Divergent Trajectories of LGBTQ Organizational Growth Across the United States During the AIDS Crisis

ABSTRACT Numerous studies examine how LGBTQ life differs between large, cosmopolitan cities like San Francisco and other, less prominent cities. Nevertheless, most of this research is done through case studies of one or a handful of LGBTQ communities, making it unclear how unique the large hubs of LGBTQ life truly are. This study leverages nationally complete data from the U.S. Gayellow Pages, a historical listing of local LGBTQ organizations, to evaluate how the organizational response of LGBTQ communities to the AIDS crisis—arguably the most prolific era of organizational creation in LGBTQ history—differed between large hubs and other cities. Findings make clear the risks of generalizing about LGBTQ life from large hubs alone. Although AIDS stimulated the creation of health-related and social movement organizations in large hubs, AIDS was more strongly associated with organizational creation outside of rather than within large hubs. The types of organizations created due to AIDS tended to be more varied outside of rather than within large hubs as well. These differences highlight the value of decentering the large hubs of LGBTQ life as units of analysis in the study of sexuality and space.

These four cities jointly comprise the "great cities" of LGBTQ life in the United States, and overreliance on findings from the great cities distorts theorizing about how LGBTQ people grow and sustain their communities (Stone, 2018).On the full distribution of LGBTQ communities and lifestyles across the country, the great cities are towering outliers (Forstie, 2020).
Theories of sexuality and space have yet to shift away from the great cities in part because few studies examine differences between LGBTQ communities on a large scale.Most empirical findings are driven by case studies of one or a handful of local communities (Brekhus, 2003;Brown-Saracino, 2017;Buring, 1997;Kennedy & Davis, 1993).The value of case studies for theory building is unequivocal, and case studies are an effective way to draw comparisons across settings (Flyvbjerg, 2006).Nevertheless, much of the field continues to see the great cities as exemplars of LGBTQ life rather than the outliers that they are.Large-scale evidence that directly compares the great cities to other settings is necessary to demonstrate how varied LGBTQ life is beyond the largest cities.
To help shift the paradigm of sexuality and space away from the great cities, this study analyzes the worst years of the AIDS crisis through the lens of LGBTQ organizations.Specifically, the study exploits nationally complete historical listings from the U.S. Gayellow Pages (GP)-a unique directory of locally owned and -operated organizations serving LGBTQ people across the country-to evaluate how the organizational response of LGBTQ communities to the worst years of the AIDS crisis diverged between the great cities and other cities across the United States.Sexualities scholars have analyzed specific organizations such as gay bars, schools, and LGBTQ-affirming churches to demonstrate differences between LGBTQ communities (Connell, 2015;Kane, 2013;Mattson, 2020).The study of sexuality and space, however, has yet the examine the diversity of organizational types that have served different LGBTQ communities over time.An exploration of LGBTQ organizations founded across the United States during a pivotal moment in LGBTQ history offers a valuable counterweight to the literature's overreliance on findings from the great cities.
The AIDS crisis, whose worst years occurred from the mid−1980s to the mid−1990s, was perhaps the most prolific era of organizational creation in LGBTQ history, and many of the organizations founded during the AIDS crisis left a legacy that influences LGBTQ life today.This era produced wellknown organizations such as the AIDS Coalition to Unleash Power (ACT UP) and the Gay Men's Health Crisis (GMHC), both of which helped bring the concerns of sexual minorities into public consciousness (Kayal, 1993;Roth, 2017;Schulman, 2021).The organizational infrastructure that developed during the AIDS crisis went on to help LGBTQ people during more recent viral scares such as the COVID−19 pandemic (Miles et al., 2021) and the spread of Mpox (Holloway, 2022) as well.Additionally, while the AIDS crisis disproportionately affected gay men (Epstein, 1996), the struggle against AIDS inspired many lesbians, bisexuals, and transgender people to fight for political rights alongside gay men (Brier, 2009;Stryker, 2004).These efforts produced numerous organizations that aimed to serve all sexual minorities (Schneider, 1992;Stein, 2012;Watkins-Hayes, 2019).
Using descriptive statistics and regression-based analyses, findings make clear how the impact of AIDS on LGBTQ organizations diverged within and outside of the great cities.As suggested by some case studies (e.g.Atkins, 2003;Buring, 1997;Merithew, 1995), AIDS inspired an organizational awakening outside of the great cities, stimulating the creation of all types of organizations throughout the United States.On the other hand, within the great cities, AIDS caused an across-the-board decrease in LGBTQ organizations, save for health-related and social movement organizations.These results have important implications for the study of sexuality and space.By demonstrating contrasts between LGBTQ organizations founded in the great cities and other parts of the U.S. during the AIDS crisis, this study highlights the value of comparative work that simultaneously measures trends within and outside of the great cities.If one were to draw conclusions from the great cities alone, one would come away with a misguided understanding of LGBTQ organizations founded during the AIDS crisis and their consequent legacy.Sexualities scholars should complement local case studies with population-level studies like ours.New, large-scale data sources on LGBTQ life, such as Facebook (Gilroy & Kashyap, 2021), Google search data (Lorber & Weiner, 2022), and recently digitized historical directories (Regan & Gonzaba, 2023), can be combined with existing large-scale data sources, such as the Census (Spring, 2013) and the General Social Survey (Lagos & Compton, 2021), to provide new insights into key moments in LGBTQ history.Indeed, exploiting these new data sources is an underappreciated way to wean the study of sexuality and space off of the great cities.

The birth of LGBTQ organizations in the great cities
To understand how organizations have historically served LGBTQ people, it is valuable to examine the AIDS crisis of the 1980s and 1990s.No other era was quite as consequential for LGBTQ organizations, especially in the great cities.As the AIDS epidemic unfolded, a broad infrastructure of health organizations and activist groups developed in the great cities (Armstrong, 2002;Epstein, 1996;Perrow & Guillén, 1990), resulting from the sheer magnitude of the AIDS crisis as well as "a perception that . . .existing community organizations were too timid in their approach to the institutions of the state and medical authority" (Altman, 1994, pp. 55-56).At the same time, other kinds of organizations, such as bathhouses and leather bars, closed down due to their association with the spread of AIDS as well as the moral panic over the supposedly deviant sexual norms that they represented (Rubin, 1997).Organizational turnover also occurred because the AIDS crisis encouraged some gay villages to become more homonormative.In other words, gay villages in great cities promoted the development of LGBTQ organizations closely aligned with cisgender heterosexual family values, such as upscale bars, and forced more non-traditional organizations, such as fetish clubs, underground (Mattson, 2015).
Although the great cities were hardest hit by the AIDS crisis, AIDS spread to all parts of the United States, and many LGBTQ communities across the country struggled for survival (Hackney, 1991;Lisbon, 2021;Rofes, 1998).At the onset of AIDS, many smaller cities had a less developed LGBTQ organizational infrastructure than did the great cities, making the former cities slower to respond to AIDS and more reliant on informal social networks for support (Perrow & Guillén, 1990;Rofes, 1998;Schneider, 1997).While scholars acknowledge the importance of local contexts to disaster management (Smiley et al., 2018), it remains unclear how local context influenced when and where LGBTQ communities coordinated their response to AIDS through organizations.If LGBTQ communities across the United States varied in the organizations they created during this moment of national crisis-both in terms of the number and kinds of organizations they created-then it is important to simultaneously analyze the great cities alongside other cities in order to isolate points of commonality and divergence between them.

The birth of LGBTQ organizations beyond the great cities
Despite the effect of AIDS on the great cities, non-heterosexual communities in other cities, such as Memphis, TN (Buring, 1997), Seattle, WA (Atkins, 2003), and Philadelphia, PA (Royles, 2020) were deeply affected by the crisis as well.Across the United States, the threat and stigma of AIDS motivated many non-heterosexual people to join the fight for LGBTQ equality (Brier, 2009;Stein, 2012;Stryker, 2004).This sentiment, along with AIDS activists' desire to create a nationwide network of organizations supporting LGBTQ life (Andriote, 1999), encouraged chapters of ACT UP and other national organizations to be founded across the country (Roth, 2017;Royles, 2020;Schneider, 1997).In many cities, AIDS deaths stimulated the creation of an organizational infrastructure that previously did not exist.Gary L. Atkins writes about the consequences of AIDS in Seattle, As the toll kept rising, the fight against AIDS generated a new infrastructure within Seattle's gay and lesbian community that might never have been there otherwise, developing a sense that community life was not just sexual liberation, or projecting a new identity, or even crusading for civil rights.Rather, it meant building the ongoing services and leadership that subsequent generations of gay men and lesbians would need.(Atkins, 2003, p. 346) Halfway across the country, another observer recounted that "the gay community in Detroit ha[d] become more organized and sophisticated" in response to the AIDS crisis (Merithew, 1995, p. 74).Gay bars, gay and lesbian homeowners' associations, and other organizations sprouted up in Detroit to mobilize queer people and fundraise for AIDS services (Merithew, 1995).
In sum, just like LGBTQ communities in the great cities, LGBTQ communities in other cities suffered greatly from AIDS.In those communities hit hardest, case studies suggest there was an organizational awakening in which sexual minorities founded new organizations of all types.We turn this assertion into the hypothesis below.Among cities that were not one of the great cities, the local AIDS death rate was positively associated with the number of new LGBTQ organizations founded, regardless of organizational type.

Data
This study analyzes agglomerations of LGBTQ organizations in metropolitan U.S. counties between 1987 and 1996, which were some of the worst years of the AIDS epidemic.We focus on counties rather than cities or neighborhoods because some historical data crucial to our study are only available at the county level, such as data on AIDS-related deaths.Beyond this rationale, LGBTQ organizations are rarely confined to the formal boundaries of particular neighborhoods or cities (Castells, 1983;Mattson, 2015;Stein, 2004).We would have liked to include non-metropolitan counties in our analyses, but too few of these counties had sizable enough LGBTQ organizational agglomerations during the study period to conduct meaningful analyses. 2 Although the first case of AIDS in the United States was detected in 1981, the Centers for Disease Control and Prevention (CDC) started sharing data on HIV/AIDS at the county level in 1987.By 1987, AIDS had already ravaged certain corners of the LGBTQ population, but to our knowledge, nationally complete county-level data on HIV/AIDS prior to 1987 are unavailable, preventing us from conducting an analysis that covers the pre−1987 period.The AIDS crisis steadily worsened until 1996, the final year of our analysis, when Highly Active Antiretroviral Therapy (HAART) became widely available across the country.HAART radically decreased AIDS death rates, ushering in a new phase of the epidemic in which AIDS became less acute and more of a chronic condition (Hall et al., 2011). 3While 1996 is an appropriate endpoint for our analysis, findings are consistent whether we shift the end of our study period one year earlier or later. 4 The main data source for this study is the Gayellow Pages (GP), a U.S. directory organized by locale that contains descriptions of locally owned and -operated organizations serving LGBTQ people.The GP is like other publications that offer directory listings for LGBTQ persons but is distinctive in its aim and scope.Its front cover indicates that the purpose is "informing the Lesbian, Gay, Bisexual & Transgender Community" and providing "vital information on the road or at home" so that readers of the directory will "find the everyday necessities of life."Due to the GP's comprehensiveness, it has formed the basis of numerous studies of LGBTQ life (Armstrong, 2002;Hayslett & Kane, 2011;Negro & Olzak, 2019;Negro et al., 2013).
There is no fee to obtain the GP or be included in its listings.To be listed, representatives of each organization must submit information about the organization on a postcard.The postcard asks for information such as name, address, and organization type.During the study period, the postcard also asked whether the organization was at least partly owned or -operated by someone who identified as LGBTQ, as well as whether the organization intended to exclusively serve LGBTQ clientele rather than a general audience.The GP was not published in 1990, 1992, or 1995, and the panel data set we construct is unbalanced.
The GP listed organizations in 182 categories during the study period.To make analyses tractable, we consolidate categories into six broad social domains: law, economy, cultural, religious, movement, and health organizations.When appropriate, we discuss individual categories in the analysis.Appendix Table A1 lists each of the 182 organizational categories as well as the social domains we assign to them.Organizations in the economy domain comprise almost 70% of listings, and the GP includes both for-profit and nonprofit organizations.Categories were largely stable in editions of the GP published during the study period.
In addition to GP data, we compile demographic measures from the U.S. Census Bureau's County and City Data Book as well as AIDSrelated data from the CDC's Wide-Ranging Online Data for Epidemiologic Research (WONDER) database.We also use a variable measuring the size of the LGBTQ population based on Census data.The absence of official statistics on the LGBTQ population makes counting LGBTQ people challenging.However, in 1990 and 2000, the U.S. Census included a proxy by allowing respondents to identify as unmarried partners regardless of the sex of the partners.Black et al. (2000) used data on same-sex unmarried partners from the 5% and 1% Public Use Microdata Samples to estimate the number of gay and lesbian households in each U.S. county in 1990.Smith and Gates (2001) updated these estimates for the 2000 Census.Although imperfect, we use these estimates to measure the number of LGBTQ people in each county included in the analysis. 5We linearly interpolate Census variables across years when necessary.
To achieve convergence in our regression models, we drop from our analysis any counties that had less than 10,000 residents at any point during the study period.This reduces our sample size by less than 0.5%.The final data set includes 53,216 organization-years spread across 580 counties such that 11,301 organization-years were in the eight counties that contained the great cities (i.e.New York City, Los Angeles, San Francisco, and Chicago) 6 and 41,915 organization-years were in the 572 counties that contained "other cities" (i.e.cities that were not one of the great cities but were still in a metropolitan county).The analysis uses 1990 Metropolitan Statistical Area (MSA) definitions, but results are similar when using 1980 MSA definitions.

Variables and methods
To assess how the AIDS crisis influenced the founding of LGBTQ organizations in U.S. counties, we rely on methods established by organizational ecology (Carroll & Hannan, 2000) and used in sexualities research (Gonsalves, 2021;Kane, 2013).Specifically, we model organizational foundings by linking a mixed-effects negative binomial model to the equation below.
where ORG it , the dependent variable, measures the number of organizations in a given social domain in county i in year t.AIDS i(t−1) , is the local HIV/AIDS death rate, 7 our key independent variable.This variable measures the number of deaths due to HIV/AIDS per 100,000 population in county i in the year before t.This value is logged in regressions.
Other factors may have also influenced organizational foundings during this time, so we include in our models X i(t−1) , a vector of lagged control variables.Some control variables speak to demographic characteristics, such as total population (logged) and the percentage of same-sex households in a county (Black et al., 2000;Smith & Gates, 2001).Data for the latter variable were not available until the 1990 Census, so we use it in separate regressions that start in 1991 rather than 1988.We also control for median personal income 8 because wealthier counties can presumably sustain more organizations.Additional controls include percent college educated since acceptance of LGBTQ identity is positively associated with education level (Adamczyk & Liao, 2019).To control for political conditions, we include a dummy variable for whether a county had ever enacted a nondiscrimination policy protecting the rights of sexual minorities. 9To account for the vitality of local LGBTQ organizational agglomerations, we capture the diversity of LGBTQ organizations in counties (calculated as a Simpson Diversity Index equal to , where p is the percentage of GP organizations in category c in each county in the year before t).Finally, ζ 1t is a county-specific random intercept, which accounts for the clustering of organizations within counties over time.the number of counties nationwide that had at least one GP organization grew by 25%.Expansion across counties was complemented by growth within counties as well.The number of counties that contained at least one GP organization jumped from 548 to 686 during the study period, and, at the median, there was a 44% increase in the number of GP organizations among counties that existed in the data set in 1987.

Descriptive statistics
Figure 2 breaks down growth patterns across counties by region and social domain.The number of organizations grew rapidly in several domains.Bolded lines show social domains that included most organizations in the data set.These domains included economy, movement, and health, which encompassed almost nine out of every 10 GP organizations in 1996.Since cultural, law, and religion organizations comprised very few of the listings in the GP, the remainder of this study combines cultural, law, and religion organizations into an "other" category.
Within counties containing the great cities, the most notable finding is a sizable decrease in economy organizations.In 1987, there were 1,073 economy organizations in these counties, but by 1996, that number decreased to 812. 10 Table 1 breaks down the change in number of organizations by social domain in each county that contained a great city.In almost every one of these counties, there was a substantial decrease in economy organizations, which pushed the overall organizational count to decline in these counties as well.The only domains that gained organizations were health and, to a lesser extent, social movement organizations. 11 In sum, despite the rise of health and social movement organizations, the great cities were associated with organizational loss more than gain.
Outside of the great cities, there was a substantial amount of organizational entry, as shown in Figure 2. Nonetheless, it is unclear whether this growth was attributable to the AIDS crisis, and whether organizational entry due to AIDS was more common across some social domains rather than others.We therefore turn to regression analysis to settle these issues.

Regression results
Table 2 includes summary statistics of key variables used in regression analyses.The table largely reinforces trends shown in Figure 2, but it is worth noting two key points.First, the median rate at which new organizations were founded outside of the great cities was positive but modest.During the study period, the median number of LGBTQ organizations in counties that contained other cities went from four to five.Nevertheless, each additional LGBTQ organization founded outside of the great cities likely had a major impact.
LGBTQ organizations outside of the great cities became lifelines for people affected by AIDS, even when these organizations did not explicitly label themselves as health or social movement organizations (Merithew, 1995;Schneider, 1997;Thompson, 2010).For example, gay bars handed out condoms, and food banks delivered groceries to people with AIDS (Lisbon, 2021).Second, the HIV/AIDS death rate grew substantially over the study period.On average, it almost tripled in counties containing other cities.This increased death rate was complemented by an expansion in the number of counties that experienced HIV/AIDS deaths as well.Between 1987 and 1996, the number of counties in the data set that experienced at least one death due to HIV/AIDS jumped from 428 to 616.The diffusion of AIDS over timewhich initially affected the great cities but went on to impact other cities around the United States (Atkins, 2003;Buring, 1997;Rofes, 1998)-coupled with growth in the number of counties that listed at least one organization in the GP, suggests that the two processes may have been related.Given the uneven distribution of LGBTQ organizations and AIDS deaths across the U.S., however, regression analyses will help clarify where and when the two were associated.
Figure 3 is a coefficient plot taken from mixed-effects negative binomial regressions that test the association between local HIV/AIDS death rates and LGBTQ organizational foundings.Each regression in the figure is specific to a particular social domain.Results shown in Figure 3 include a control variable for the lagged percentage of households that were same-sex couple, meaning that the models presented start in 1991.Full regression results, including models that exclude this control variable and start in 1988, are shown in Appendix Table A2. Figure 3 only includes counties that did not contain the great cities.We omit counties that contained the great cities from regression analyses because there were only eight such counties. 12Regardless, it is useful to compare the findings in Figure 3 to the findings in Table 1, which highlighted how the great cities experienced an overall loss of organizations.
No matter the social domain, regressions consistently show that, in counties containing other cities, AIDS was associated with the founding of all kinds of new LGBTQ organizations.This finding is consistent despite a p-value of 0.13 on the positive association between local HIV/AIDS death rates and the establishment of new social movement organizations.Average marginal effects further confirm the notable connection between AIDS and organizational entry.In counties containing other cities, a one-unit increase in the logged local HIV/AIDS death rate was predicted to increase the overall number of GP organizations the following year by  1987 1996 1987 1996 1987 1996 1987 1996 1987   Values comes from Black et al. (2000) and Smith and Gates (2001).
0.87.Based on this evidence, we uphold our hypothesis.Unlike the overall decline in organizations within the great cities, there was likely an organizational awakening in LGBTQ communities outside of the great cities due to the AIDS crisis.

Robustness of county classifications
Thus far, findings support our hypothesis but are based on imprecise distinctions between great cities and other cities.Whereas Stone (2018) does not treat Boston as one of the great cities of LGBTQ life, for example, Levine does (1979).Figure 4 reproduces the coefficient plots in Figure 3 using rural-urban continuum codes (RUCCs) from 1983.RUCCs enable us to split counties containing other cities into more refined classifications.We now classify counties into three groups: counties that contain the same four great cities as in previous analyses; other large counties (i.e.counties in metropolitan areas of 1 million or more people, such as Suffolk County, MA, which contains Boston); and midsize counties (counties in metropolitan areas of less than 1 million people, like Honolulu County, HI). 13 Sample size constraints do not allow us to refine county distinctions further.Appendix Table A3 contains the full regressions results on which Figure 4 is based.As before, regressions exclude counties that contained the great cities, although results do not change if these eight counties are included along with other large counties.A2.
Even when counties are classified using RUCCs, findings largely remain the same.In other large counties, organizational foundings were consistently, strongly associated with local HIV/AIDS deaths.Organizational foundings in midsize counties were also consistently positively associated with local HIV/ AIDS deaths, even though the positive association was not statistically significant in the case of movement organizations. 14Taken together, these findings support conclusions from previous analyses.

Discussion
Scholarship on sexual minority communities outside of the large, cosmopolitan centers of LGBTQ life has grown tremendously (Bell & Valentine, 1995;Forstie, 2020;Myrdahl, 2013;Podmore & Bain, 2020;Tongson, 2011), largely through case studies of one or a handful of LGBTQ communities (Brekhus, 2003;Brown-Saracino, 2017;Compton & Baumle, 2012;Gray, 2009;Kazyak, 2012;Kirkey & Forsyth, 2001;Krahulik, 2007;Todd, 2018).Despite the useful knowledge gained through these case studies, few studies can directly identify points of commonality and difference across LGBTQ communities on a large scale (Cooke & Rapino, 2007;Gates & Ost, 2004;Spring, 2013).This gap in the literature is especially problematic in the United States, where most studies of Coefficient plots of organizational foundings regressed on HIV/AIDS death rates using rural-urban continuum codes.Note.Regressions are done on two sets of counties: large counties other than those that contained great cities ("OLC") and midsize counties ("MSC").County-level HIV/AIDS death rates are logged and lagged by one year.Full regression results are shown in Appendix Table A3.
sexuality and space focus on the four "great cities" of LGBTQ life, namely New York City, Los Angeles, Chicago, and San Francisco (Stone, 2018).Without large-scale comparative work that can determine how unique the great cities truly are, the study of sexuality and space struggles to wean itself off of theories derived from great-city contexts.
Our study examined LGBTQ organizational growth patterns across the United States during the AIDS epidemic, a nationwide crisis whose impacts were felt by LGBTQ communities across the country, but in divergent ways.In cities that were not one of the great cities, AIDS stimulated the creation of numerous LGBTQ organizations of all types.Within the great cities, in contrast, LGBTQ organizations were more likely to close rather than open, despite the establishment of well-known, influential health and social movement organizations often discussed in the literature, such as the GMHC and ACT UP (Kayal, 1993;Roth, 2017;Schulman, 2021).This finding has two important implications for the study of sexuality and space, and for sexualities studies more broadly.First, by examining how the consequences of a national-level event varied across the United States, our findings caution against generalizing from the experience of one or a few LGBTQ communities.Large-scale, population-level analyses like ours can reveal the diversity of ways that LGBTQ people respond to key moments in history such as the AIDS crisis.Ongoing efforts to digitize historical records of LGBTQ life (Mattson, 2019;Regan & Gonzaba, 2023) and analyze new data sources such as Facebook (Gilroy & Kashyap, 2021) and Google search data (Lorber & Weiner, 2022) herald an exciting opportunity to do this kind of population-level research.Second, organizations are a useful site in which to understand how micro-level social patterns influence and are influenced by macro-level transformations within the LGBTQ population.The range of organizations that serve the LGBTQ population is vast (Gonsalves, 2021;Kane, 2013;Lockhart, 2022;Mattson, 2020;Velasco & Paxton, 2022), and as new data on LGBTQ organizations becomes available, scholars can use them as a window into the social conditions that shape and are shaped by LGBTQ communities.Methodologically, it would be useful for more sexualities scholars to draw on organizational ecology (Carroll & Hannan, 2000), as we did, to better understand dynamics within sexual minority communities.
The implications of our study hold despite two limitations to the analysis.First, our analytic approach presumed that organizations were founded the year they began listing themselves in the GP.There are many advantages to this approach, yet organizational directories may provide a biased view of where and when formal organizations cater to LGBTQ people (Knopp & Brown, 2021).Some organizations serving people with AIDS may not have listed themselves in the GP because they were afraid of retribution from homophobic locals.Our results should consequently be triangulated with other sources of data, especially those that go beyond the organizational categories of the GP and capture more characteristics of organizations.
Second, our study essentialized aspects of the LGBTQ experience to measure them quantitatively.Different LGBTQ organizations likely catered to different subgroups, and any temporary and informal organizations not listed in the GP may have been as important to LGBTQ communities as formal organizations were.A contemporary example of informal organizing comes from Grindr, a social media application that substitutes for some of the communitybuilding traditionally done through formal LGBTQ organizations (Collins & Drinkwater, 2017).While many people argue that social media applications have led to the decline of organizations such as the gay bar, they cannot replace the political, economic, and affective aims of many LGBTQ organizations, suggesting that at least some types of LGBTQ organizations will persist into the future (Bitterman & Baldwin Hess, 2021;Crooks, 2013).
Scholarship will likely address these limitations in the near future.Researchers are using organizational directories like the GP in creative ways, including as a census of LGBTQ organizations (Gonsalves, 2021;Mattson, 2019;Regan & Gonzaba, 2023); as a sampling frame for survey research (Rothblum & Rothblum, 2020); and as an opportunity to conduct critical analyses that reveal which segments of the LGBTQ population are most visible in geographic space (Knopp & Brown, 2021).We look forward to seeing how the field uses LGBTQ organizational directories in the future.

Notes
1. Throughout this paper, we use several terms to refer to sexual minorities, including LGBTQ, non-heterosexual, and queer.This language reflects the variety of terms used in scholarly and popular discourse.Our empirical analysis examines the 1980s and 1990s, when terms like LGBTQ had not yet become commonly used.We nonetheless use contemporary terms to inform current debates.When appropriate, we refer to specific sexual identity groups such as gays or lesbians.2. Bivariate analyses of non-metropolitan counties suggest that LGBTQ organizational growth was not associated with local HIV/AIDS deaths in rural areas.Regardless, the sample size of non-metropolitan counties that contained at least one GP organization was too small to make strong conclusions.3.More accurately, 1996 marked a transition in which the burdens of HIV/AIDS became stratified by more traditional forms of race-, class-, and gender-based inequality than by sexual orientation (Petrus, 2019).4. The Gayellow Pages was not published in 1995, so we technically tested models that ended in 1994 and 1997.In both cases, findings are robust.5. Lesbians frequently live in smaller cities than do gay men (Gates & Ost, 2004), so lumping together male and female same-sex couple households in the data set may bias our results.However, results are similar whether we control for all same-sex couple households, male same-sex couple households only, or female same-sex couple households only.6.These counties include Los Angeles County, CA; San Francisco County, CA; Cook County, IL; and the five counties that comprise New York City.Results are virtually identical if we restrict great cities data from New York City to New York County (i.e. Manhattan) only.Results are also virtually identical if we include surrounding counties as part of the great cities (meaning other counties within a great city's MSA).7. We focus on HIV/AIDS deaths because both HIV/AIDS cases and AIDS deaths were only available at the state level during the study period.8.All monetary variables in this study are measured in 1990 dollars. 9. Data for this variable come from Eskridge (1999, app.2), the National Gay and Lesbian Task Force (http://www.thetaskforce.org), and the Workplace database maintained by the Human Rights Campaign (http://www.hrc.org).10.In counties containing the great cities, the leading category of economy organizations was "BARS, RESTAURANTS, CLUBS, DISCOS," which comprised 41% of economy organizations in both 1987 and 1996.In fact, this category was so prominent in the data set that it may be useful to treat the category as distinct from other economy organizations, especially considering the role that many gay bars play in non-economic activities such as community organizing and public health (Thomas, 2011).Even when we treat this category separately from other economy organizations, findings do not change.11.It is possible that many organizations created during the AIDS crisis were established before 1987 rather than during our study period.If so, then the organizational decline after 1987 would mask overall organizational growth over time.Nevertheless, in the 1984 edition of the GP, counties containing the great cities had 1,944 organizations, higher than the 1,803 organizations in the 1987 edition.The organizational decline in counties containing the great cities, in other words, had already started prior to our study period.12.Even if we add counties that contained great cities to regression analyses, results do not change because there were so few counties in this group.13.To be specific, large counties have a RUCC of 0 or 1 but do not contain great cities, and midsize counties have a RUCC of 2 or 3. 14.In the case of other organizations, the coefficient for local HIV/AIDS deaths had a p-value of 0.09.4,254.62 2,931.80 2,325.86 1,685.26 2,325.86 1,685.26 3,657.59 2,652.68 2,989.61 2,110.71 4,684.85 3,272.26 3,478.37 2,399.47 6,010.01 4,151.16 2,233.25 1,554.99 3,126.62 2,189.90***p < .001,**p < .01,*p < .05(two-tailed).Models that exclude a control variable for percentage of households that were same-sex couple cover the 1988 to 1996 period.Because data on same-sex couple households were only available starting in 1990, models that include this control variable cover the 1991 to 1996 period.

Figure 1
Figure 1 is a bubble map that shows the number of GP organizations in U.S. counties in 1987 and 1996.The larger the bubble, the more GP organizations in a county.The figure affirms the importance of the great cities to LGBTQ life.Los Angeles County, San Francisco County, Cook County, IL (which contains Chicago), and the counties that comprise New York City were the places with the most LGBTQ organizational activity.Washington D.C. was a distant fifth place in both 1987 and 1996.Other counties in the top 10 during both years included King County, WA (which contains Seattle), Philadelphia County, PA, and San Diego County, CA.In addition, between 1987 and 1996,

Figure 2 .
Figure 2. GP organizations over time by region and social domain, 1987 to 1996.Note.The top panel includes organizational counts by domain for all counties that contained great cities.The bottom panel includes counts for all counties that contained other cities.

Figure 3 .
Figure 3. Coefficient plot from mixed-effects negative binomial regressions of county-level LGBTQ organizational foundings on county-level HIV/AIDS death rates.Note.This figure is restricted to counties that contained other cities.Regressions are done separately for each social domain.County-level HIV/AIDS death rates are logged and lagged by one year.Full regression results are shown in Appendix TableA2.

Figure 4 .
Figure 4. Coefficient plots of organizational foundings regressed on HIV/AIDS death rates using rural-urban continuum codes.Note.Regressions are done on two sets of counties: large counties other than those that contained great cities ("OLC") and midsize counties ("MSC").County-level HIV/AIDS death rates are logged and lagged by one year.Full regression results are shown in Appendix TableA3.

Table 1 .
Number of gayellow pages organizations in counties that contained the greatcities, 1987  and 1996.

Table 2 .
Summary statistics of key variables.Values are medians and ranges rather than means and standard deviations.HIV/AIDS-Related Deaths and Total Population are logged in regressions.
2 Values are measured in 1990 dollars. 3 Organizational Diversity is a Simpson Diversity Index of LGBTQ organizations in each county.4

Table A2 .
Mixed-effects negative binomial regressions of county-level LGBTQ organizational foundings on county-level HIV/AIDS death rates: Counts by domain.Models that exclude a control variable for percentage of households that were same-sex couple cover the 1988 to 1996 period.Because data on same-sex couple households were only available starting in 1990, models that include this control variable cover the 1991 to 1996 period.

Table A3 .
County-levelLGBTQ organizational foundings regressed on county-level HIV/AIDS death rates: Counts using rural-urban continuum codes.Other Large Counties Midsize Counties Other Large Counties Midsize Counties Other Large Counties Midsize Counties Other Large Counties Midsize Counties Other Large Counties Midsize Counties