Abstract
Why do municipal governments adopt lesbian, gay, bisexual, and transgender (LGBT) inclusive policies? The preponderance of literature suggests urbanism and social diversity are the most likely explanations for LGBT municipal policies. This research tests these assumptions using the morality politics model. Using rare-events logistic regression, municipalities in the state of Florida with LGBT antidiscrimination ordinances are compared with municipalities that do not have such policies. The results contradict theories of urbanism and highlight the shortcomings of the morality politics model. Specifically, the results indicate that even under highly salient conditions, LGBT advocacy resources play an important role in the policy adoption process.
Keywords
Despite recent gains for lesbian, gay, bisexual, and transgender (LGBT) rights at the federal level, many states and municipalities do not have LGBT antidiscrimination laws. This discrepancy has serious political and public policy ramifications given, for example, recent court decisions affirming same-sex marriage rights. That is, while same-sex couples may be legally married, many state and local laws do not prohibit discrimination against those individuals in employment, housing, and public accommodations, among other policy areas. Furthermore, transgender individuals face a disproportionate amount of discrimination, yet as of 2013, only 200 municipal and county governments had adopted any sort of gender identity–inclusive discrimination protections (HRC 2013; National Gay and Lesbian Task Force 2011). As the debate about LGBT rights increasingly becomes localized, the immediacy of state and local policy to the everyday lives of citizens makes it important to more clearly understand why local governments adopt LGBT antidiscrimination policies.
Much research has focused on state support and adoption of LGBT antidiscrimination policy; however, there have been fewer studies which examine local adoption of such policies (for state-level studies, see Taylor et al. 2012; Lax and Phillips 2009; McVeigh and Diaz 2009; Barth, Overby, and Huffmon 2009; Werum and Winders 2001; Haider-Markel and Meier 1996). Of the studies that examine local adoption of LGB antidiscrimination policies, the common conclusion considers local population size the most important factor in determining policy adoption. That is, municipalities with larger populations are more likely to adopt LGB protections than municipalities with smaller populations (see Wald, Button, and Rienzo 1996; Haeberle 1996; Wilson 1995). These studies concede, however, that population alone explains only part of the variation in policy adoption and posit theories relating to “moral regulation and social identity” may offer more explanatory value (Wald, Button, and Rienzo 1996, 1160). The purpose of this study is to reassess the question of why some cities adopt LGBT antidiscrimination policies and some do not.
This article will explore the determinants of municipal LGBT antidiscrimination policy adoption by comparing municipalities within the state of Florida which have adopted such policies with municipalities which have not. As the third most populous state in the country, Florida was chosen for two reasons. First, the demographic characteristics of the state are similar to the nation as a whole. Racial composition, income distribution, and economic indicators are all commensurate with national statistics. Second, with 410 incorporated municipalities, Florida is only slightly above the national average of 382, and the proportion of metropolitan areas, cities, towns, and villages also corresponds well with the nation as a whole (U.S. Census Bureau 2007). For these reasons, Florida represents somewhat of a microcosm of the United States, which make the conclusions derived from this analysis generalizable to the nation.
To test the morality politics theory, this analysis estimates a rare-events logistic regression model for municipalities in the state of Florida (King and Zeng 2001). First, the results refute the urbanism and community diversity hypotheses which contribute significance to municipal population and concomitant social diversity. Second, the results are consistent with the morality politics theory in that issue salience, partisan competition, and partisanship are significant indicators of policy adoption. In the end, however, the results illustrate some deficiencies in the morality politics model as advocacy/interest group resources are significant even under salient conditions.
Background
The morality politics framework has previously been used to explain local variation in LGB policy support (Haider-Markel and Meier 1996). Morality policy often involves issues that constitute deeply held religious beliefs, and it is those beliefs combined with strong partisanship and party competition which sustain morality policies. Because they represent an effort to regulate societal standards, morality policies easily evoke a response from the public and politicians who often seek to reduce the policy to a normative “right or wrong.” The low informational barriers to participation which result from such reduction increases issue salience, or the “importance and visibility” of a policy “to the public at large, and [its] prominence in public discourse” (Lax and Phillips 2009, 370). Salience, combined with partisanship, furthers morality politics debates as political parties are easily associated with certain policy preferences (i.e., prolife, prochoice). Furthermore, because strong party competition is more likely to produce openness to “citizen inputs,” the traditional morality politics model theorizes “competitive party systems” contribute to the development of “liberal policies” (Haider-Markel and Meier 1996, 338).
Recent literature, however, has exposed some caveats and limitations to the traditional approach. In the analysis that follows, the traditional morality politics model will be supplemented to reflect important contributions to the theory including factors such as the relationship between issue salience and public opinion as well as the influence of interest groups on LGBT policy adoption (Lax and Phillips 2009; Taylor et al. 2012).
The traditional morality politics model assumes high visibility, contingent upon the distribution of community values will either positively or negatively impact LGBT policy adoption (Haider-Markel and Meier 1996). In their study of state LGBT policies, Lax and Phillips (2009) demonstrate the importance of citizen values in the form of public opinion as a component of issue salience. The authors determine state LGBT policy is most responsive to “policy-specific opinion” (p. 383). However, despite favorable majority opinion, state LGBT policy does not always reflect the distribution of citizen values. That is, “even clear supermajority support seems insufficient for policy adoption” in certain policy instances such as state LGBT employment and housing discrimination protections (p. 383). Because strong public opinion and high issue salience provide a “clear signal” to policy makers, policies which boast both are more likely to produce responsiveness from democratic institutions (p. 383). Although this is not inconsistent with morality politics theory, the authors recognize institutional responsiveness and state policy congruence (i.e., enacting strongly supported LGBT policy), while heavily influenced by public opinion are also conditional on other factors which include institutional factors, issue salience, and interest group influence, the latter being ignored by the morality politics model.
The omission of LGBT population and interest group influences is perhaps the most serious deficiency of the morality politics model. That issue salience mitigates the effects of LGBT populations and interest groups on policy adoption at the local level cannot be assumed. For example, municipalities with robust LGB populations are more likely to have LGB policies than municipalities that lack a discernable LGB population (Wald, Button, and Rienzo 1996, Haberle 1996). Furthermore, personal contact with LGBT individuals can favorably influence attitudes toward LGBT policy (Barth, Overby, and Huffmon 2009; see also, Overby and Barth 2002). This influence can be greatly reduced, however, in “social contexts in which strong antigay messages are common” (p. 361). Morality politics assumes a salient policy debate, which would suggest strong pro- and anti-LGBT messages are commonplace. Do salient policy debates then mitigate all local LGBT demands and resources? Interestingly, in their examination of gay rights initiatives, Haider-Markel and Meier (1996, 341) reported no significance among their interest group variables suggesting interest groups had no influence in salient policy debates. Taylor et al. (2012, 90), however, construe LGBT interest groups to be an influence on policy diffusion which represent important external factors that “drive decisions to address LGBT rights.” This analysis contributes to the traditional morality politics model by including measures of LGBT population and interest group strength to test the claim that issue salience mitigates local demands and resources.
Along with salience and citizen values, the traditional morality politics model theorizes a relationship between LGBT policy adoption, partisanship, and partisan competition. Because of the heuristic nature of political parties, partisanship denotes a clear distinction in policy priorities. As Democrats are increasingly identified with liberal policies especially as relating to LGBT rights, it is expected that any measure of democratic partisanship would show a positive relationship with the municipal adoption of LGBT policy (Haider-Markel 2000). Furthermore, Haider-Markel and Meier (1996) argue strong partisan competition makes the adoption of liberal policies more likely. When partisan competition is low, it provides little incentive for elected officials to venture from their constituency’s values into a morality policy debate. However, a competitive environment produces more openness to citizen inputs which result in liberal policies (Holbrook and Van Dunk 1993; see also Key 1964). Although this differs from Mooney and Lee (1995), because Haider-Markel and Meier (1996) specifically examined LGB policy, their conception of partisan competition and its relationship to LGBT policy adoption make their interpretation more applicable to this study.
Finally, as indicated by Haider-Markel and Meier (1996), the morality politics model should take into account elements of the policy environment. Community diversity, urbanism, and socioeconomic conditions are influential in the adoption of inclusive policies, especially relating to sexual minorities (although the relationship with SES is undetermined; see Wilson 1995). Furthermore, the preponderance of literature suggests urbanism and social diversity are the most influential factors in municipal LGBT policy adoption. The morality politics model contradicts these assumptions placing no importance on the effects of community population or social diversity on the LGBT policy adoption process. The following sections describe the research design, operationalization of the dependent and independent variables, and the results of the analysis.
Research Design
Geographical Focus
This study compares municipalities in the state of Florida which have adopted LGBT antidiscrimination policies with municipalities which have not. The scope of the study has been limited to a single state in order to account for the effects of what Meier and McFarlane (1992, 692) refer to as “policy inertia,” or the concept that, “policies related to another policy are often good predictors of that policy choice” (see also Hofferbert and Urice 1985). Wald, Button, and Rienzo (1996, 1164) found municipalities and counties which had adopted LGB policies were more likely to be located in states that had LGB-inclusive hate crimes statutes and civil rights laws as well as states that had not adopted antisodomy laws. In order to mitigate this inertial effect, the state of Florida was chosen for its lack of statewide LGBT discrimination protections.
Dependent Variable: LGBT Policy
Using the municipalities in the state of Florida as the units of analysis, the existence of nondiscrimination laws inclusive of LGBT identity is the basis for a single dichotomous dependent variable (1 representing adoption of a policy; 0 otherwise). The classification “nondiscrimination law” may include policies that prohibit discriminatory practices in a number of policy areas the most common of which are employment, housing, and public accommodations (Colvin 2007). Although Taylor et al. (2012) distinguish policies based on content (i.e., employment, public accommodations, education, etc.) due to the relatively low number of observances of the dependent variable (n = 32), a municipality is said to have adopted an LGBT policy if it prohibits discrimination in any one of the major policy areas. Furthermore, the dependent variable represents municipalities which ban discrimination either on the basis of sexual orientation or gender identity. That is, some municipalities have banned discrimination based on sexual orientation, not gender identity, omitting protections for transgender individuals. Contrary to the recommendations of Taylor et al. (2012), and again due to the low number of observances, those policies that exclude gender identity will be included in the measure of the dependent variable. Data for the dependent variable were taken from reports by the National Gay and Lesbian Task Force (NGLTF) and the Human Rights Campaign (HRC; NGLTF 2011; HRC 2013).
The distribution of the dependent variable clearly represents discrete events which occur over the course of eighteen years. Wald, Button, and Rienzo (1996) attempted to compensate for this by altering the underlying assumption about their dependent variable. Instead of a dependent variable that represented either the existence or nonexistence of an ordinance, the variable indicates whether the municipality “maintained” an LGB policy “at the time the data were collected” (p. 1155). This study makes a similar assumption.
Policy Determinants: Independent Variables
Salience
Salience measures the “importance and visibility” of LGBT policy among Florida municipal residents (Lax and Phillips 2009, 370). Using data from the 2008 general election, the salience measure is the percentage of registered voters in each municipality which voted either for or against Amendment 2, a ballot measure that would amend the Florida state constitution to prohibit same-sex marriages or civil unions. Where municipal-level data are unavailable, the percentage of county voters which cast a ballot on Amendment 2 is substituted. Consistent with Mooney and Lee (1995) and Haider-Markel and Meier (1996), who postulated a negative relationship between issue salience and policy adoption, this variable is expected to be negatively associated with the adoption of LGBT policies by municipal governments.
Liberal and conservative values
Morality politics theory asserts issue salience leads to greater citizen involvement; and the impact of citizen involvement on policy adoption is a reflection of citizen values (Meier 1994; Haider-Markel and Meier 1996). Lax and Phillips (2009) provide a conceptualization of public opinion or citizen support/opposition to LGBT policy as a factor in the policy adoption process, which is contingent on issue salience and interest group resources. Measures of citizen values can be used to test these assumptions. Under salient conditions, if community values reflect conservative opinions on LGBT rights, then these values are expected to have a negative effect on antidiscrimination policy adoption. The converse (i.e., with liberal community values) is also hypothesized.
As there exist no polling data from which to derive issue-specific opinion, the distribution of citizen values is measured as a reflection of fundamentalist or conservative religious values and/or liberal religious values. The number of conservative and liberal religious institutions in each municipality is used to measure these values. Fundamentalist or conservative religious communities often constitute the driving force behind morality policy (Meier and Johnson 1990). Measures of religious fundamentalism provide a reliable indicator of conservative citizen values within a municipality. Similarly, liberal religious traditions that are more favorable to LGBT policies provide an indicator of liberal citizen values for a municipality. Drawing from Wald, Button, and Rienzo (1996), data from the Association of Statisticians of American Religious Bodies’ (ASARB) 2010 county-by-county study of congregations are used to create variables for liberal and conservative religious congregations in each Florida County (see Melton 1991; Wald, Button, and Rienzo 1996 for full descriptions of conservative and liberal denominations). The number of congregations in each county was then divided by the number of municipalities within the same county in order to provide a more accurate depiction of the distribution. The variables (Liberal Values and Conservative Values) represent the average number of conservative and liberal religious institutions in each municipality. Although Wald, Button, and Rienzo (1996) purposely excluded the Catholic Church in their indicator of conservative religious affiliation (see note 5), the denomination is included in this measure. To be sure of the impact, however, models will be estimated with and without a measure for Catholic congregations. If LGBT antidiscrimination policy is salient and the measures of religious institutions are accurate indicators of community values, then the conservative variable is expected to be negatively related to the dependent variable and the liberal variable to be positively related. This is not inconsistent with the morality politics model; however, it provides a novel way to test the conclusions drawn by Lax and Phillips (2009) at the local level.
Advocacy Resource
An important factor in the LGBT antidiscrimination policy adoption process, which is excluded by the morality politics model, is the role of supportive populations and interest groups. Data were collected on LGBT businesses and LGBT organizations for each municipality, as well as a census measure for percentage of the municipal population which identified as unmarried partners, which will be used as an indicator of LGBT population (see Wald, Button, and Rienzo 1996; Haeberle 1996 for discussion on LGB population variables; see Green 2013 for LGBT business/organization indicators).
A factor analysis reveals these three indicators measure a single concept, which will be referred to as the Advocacy Resource Scale. A municipality that scores low on the scale is lacking in LGBT resources while a municipality with a high score exhibits a strong LGBT network characterized by a robust LGBT population and advocacy groups and businesses with the potential to help further LGBT policy. If the theory that partisan competitiveness induces openness to citizen inputs holds for interest groups, then a competitive environment may allow for more access to elected officials by interest groups. In a salient policy environment, with competitive political parties, the significance of interest group resources may refute this assertion of morality politics at the local level.
Partisanship
Consistent with Wald, Button, and Rienzo (1996) and Haider-Markel and Meier (1996), partisanship is measured as the percentage of the democratic vote for president in each municipality. In this case, the average percentage of the vote for the democratic candidate for president from 1996 to 2012 has been calculated. It is expected that Democratic Party support will be positively associated with LGBT policy adoption.
Party Competition
Partisan competition is measured as the average difference between the percentage of votes cast for the Democratic and republican candidates for President, Governor, and U.S. Senate in each election beginning in 1996. From these data, a Scale of competition is constructed. As a municipality moves up the resulting scale, ranging from 23.55 to 139.71, the distance between the vote percentages of the two parties widens and competition is said to decrease. Smaller measures mean greater competition. It is hypothesized that a decrease in competition (moving from low to high values on the Competition scale) is expected to lessen the likelihood that a municipality will adopt an LGBT policy.
Policy Environment Variables
In order to test the significance of policy environment factors, specifically the urbanism and social diversity hypotheses, standard demographic variables such as population size, percentage of the municipal population which is African American and Hispanic, and median age are included in the analysis.
Finally, in order to better understand the relationship between SES and morality policy, a measure of municipal SES was developed. Factor analysis indicates measures of median family income, percentage of municipal population with bachelor’s degree or higher, percentage of municipal population below the poverty line, and municipal unemployment rate loaded on the same dimension. This became the basis of the SES Scale variable. Although it seems intuitive that SES would have little bearing on morality policy positions, Morgan and Meier (1980) contend higher SES communities should be less willing to impose conservative values on others. Furthermore, given the established relationship between education and LGBT policy support, it would not be unreasonable to hypothesize a positive relationship between SES and LGBT policy adoption.
Results
Although a logistic regression model is typically used with a dichotomous response variable, because the number of municipalities which have adopted LGBT policies is far less than the number of municipalities which have not adopted such policies, the standard logistic regression model would underestimate the probability of policy adoption (King and Zeng 2001). In this case, the proportion of occurrences of the dependent variable is roughly .07. With such a low proportion, using the rare-events logistic regression model developed by King, Tomz, and Wittenberg (1998) will produce more accurate coefficient estimates. Table 1 (see the table in the online supplement for information on the source for measures of the independent and dependent variables) shows the results of the analysis. Model 1 is the estimation of the coefficients with all variables in the model, and model 2 is the estimation with the insignificant variables from model 1 excluded. The salience, liberal/conservative values, partisanship, and competition variables are all significant and signed in accordance with the morality politics model.
Rare-event Logistic Regression.
Note: Dependent variable = LGBT policy. SES = socioeconomic status. n = 410.
*p < .1.
**p < .01.
***p < .001.
Salience and Liberal/Conservative Values
First, the results indicate LGBT antidiscrimination policy represents a salient debate at the municipal level. Furthermore, as issue visibility and prominence in the public discourse increases, municipalities appear less likely to adopt LGBT protections. Second, that issue salience along with conservative and liberal values indicators all show significance is consistent with Lax and Phillips’ (2009) findings relating to public opinion and policy responsiveness/congruence. Although the causal direction of the relationship cannot be specified here, the results comport with their interpretation that public opinion (in this case support or opposition for LGBT policy based on conservative or liberal religious values) is contingent on other factors including issue salience. That is, when salience is high, community attitudes toward LGBT policy do play an important role in policy congruence (whether or not democratic institutions reflect public opinion). Conservative attitudes (as indicated by conservative religious institutions) negatively relate to the existence of LGBT antidiscrimination policy under salient conditions while the opposite holds for liberal attitudes. Again, this is not inconsistent with morality politics theory, but it does deepen our understanding of the role community attitudes play in the adoption of LGBT policy.
Advocacy Resource
The significance of the advocacy resource variable in model 1 is inconsistent with Haider-Markel and Meier’s (1996) interpretation of the morality politics model, although the significance of the coefficient is reduced markedly from model 1 to model 2. Municipalities with more robust LGBT populations, LGBT businesses, and LGBT organizations are more likely to have antidiscrimination policies. Although the diffusion hypothesis was not specifically tested in this case, the significance of the advocacy resource variable suggests the morality politics assumption that internal variables such as partisanship and competition alone are the most important determinants of LGBT policy adoption may be flawed. It should be noted, however, that part of the variable represents internal municipal characteristics (e.g., LGBT population estimates) while some represent external influences (local chapters of national LGBT advocacy organizations). It is clear that municipalities where interest group resources are available and strong networks of advocates exist are more likely to have LGBT antidiscrimination protections.
Partisanship
As expected, the coefficient is positively related to the dependent variable. Municipalities with higher concentrations of Democratic voters are more likely to have LGBT antidiscrimination protections than municipalities with lower levels of Democratic support. This result is consistent with all previous research on LGBT policy support.
Competition
The significance of the competition coefficient supports the morality politics model, and its negative sign lends weight to the interpretation that partisan competition spurs contentious policy adoption (Haider-Markel and Meier 1996). In this case, as competition decreases (ascending value on the competition scale), municipalities appear less likely to adopt LGBT antidiscrimination policies. When issue visibility is high, strong competition between Democrats and republicans is likely to produce support for LGBT antidiscrimination protections. Furthermore, the significance of the Democratic partisanship variable reinforces the conclusions drawn from the significance of that variable alone, that municipalities in safe Democratic districts are more likely to have LGBT protections than municipalities in safe republican districts.
Additionally, the significance of the advocacy resource and the partisan competition coefficients supports the assertion that partisan competition opens the political system to the influence of LGBT interest groups. As Lax and Phillips (2009, 370) observed, “elected officials may feel it desirable or necessary to satisfy key interest groups instead of the median voter, for financial or other reasons”. This may also help explain why under salient conditions politically competitive municipalities are more likely to have LGBT antidiscrimination policies.
Policy Environment Variables
In line with the predictions of the morality politics model, large populations have no significance in predicting LGBT municipal policy in Florida. This is contrary to existing literature (Wald, Button, and Rienzo 1996; Haeberle 1996). Further contradicting previous studies, the model indicates SES has no significance as a determinant of antidiscrimination policy adoption. Only one of the policy environment variables, the percentage of municipal population which is Hispanic, shows significance. As Florida has a Hispanic population roughly 6 percent larger than the national average, it is not surprising that this measure shows significance (United States Census Bureau 2010). Using this variable as a measure of community diversity, however, does not support urbanism and social diversity literature because it is negatively signed.
Literature on Hispanic attitudes toward homosexuality shows similarities between Caucasians, African-American, and Hispanic attitudes, with variation explained by education, religious affiliation, and political ideology (Bonilla and Porter 1992; Herek and Gonzalez-Rivera 2006; see also Loftus 2001). The significance of the Hispanic variable in this model and its negative coefficient, however, suggests a negative relationship between Hispanic populations and LGBT policy adoption.
Conclusions
Evidence suggests LGBT policy does fit the morality politics model. Issue salience, partisanship, and partisan competition consistently show significance which supports morality politics theory. Although this is not contrary to previous literature, the significance of LGBT advocacy resources is. The results of this analysis provide an answer to a central question of this study. In reviewing the determinants of LGBT antidiscrimination policy at the municipal level, the urbanism hypothesis is not supported. Furthermore, though issue salience, partisan competition, and partisanship are significant predictors, the interaction of these variables with supportive populations and interest group resources highlights the deficiencies of the morality politics model. When conditions are salient, municipalities are less likely to adopt LGBT protections. However, salience does not completely mitigate the effects of interest groups on the LGBT policy adoption process. On the contrary, politically competitive municipalities, even when the issue of LGBT rights appears salient, are more likely to have LGBT antidiscrimination protections. LGBT populations and organizations do represent a resource available to mobilize around policy debates. Furthermore, competitive partisan environments may provide inroads for interest groups who wish to influence morality policy debates.
This study also has implications for LGBT policy, given the recent increase in court rulings affirming same-sex marriage rights. Many of the states that have been compelled to recognize same-sex marriages also have hostile political environments for LGBT people. While a right to marry may be recognized, the fact remains that many states and municipalities do not have LGBT antidiscrimination laws. That is, while same-sex couples may be able to legally marry, they may also have their employment legally terminated, or legally face discrimination in housing and public accommodations. The conflicting policy environment that results from judicial marriage determinations may further heighten issue salience which has negative implications for LGBT policy adoption. Future research should examine the implications of same-sex marriage decisions on the LGBT municipal policy process.
Also, while conservative religious affiliations have been shown to negatively impact LGBT policy support there has been little evidence to suggest the scope of influence held by liberal religious values on the policy process. This study suggests that liberal religious values play an important role in predicting policy adoption which should be further investigated.
The morality politics model does not traditionally contribute explanatory significance to demographic or “policy environment” variables, although Haider-Markel and Meier (1996) found levels of education showed significance in most of the cases of progay referenda votes. The significance of the Hispanic population measure in this study is anomalous. As previously stated, the Hispanic population in the state of Florida is roughly 6 percent larger than the national average, so the results may be due to an oversampling of the population. That the sign of the coefficient is negative coupled with the insignificance of the municipal population variable contradicts Urbanism and community diversity theories of LGBT policy adoption.
Future research should further explore the relationship between minority populations and LGBT policy. Previous studies of Hispanic and African American attitudes report more negative reactions when asked if homosexuality is immoral; however, when asked if immoral behavior merited a denial of civil liberties to gays and lesbians, the negative response was more muted (Bonilla and Porter 1992; see also Loftus 2001). As future research more thoroughly investigates the content and complexity of LGBT policy adoption, the community context should also be explored to better determine the nature of Hispanic attitudes, specifically, toward LGBT policies.
Finally, this study furthers our understanding of LGBT policy adoption by extending the morality politics model to the municipal level. This study encompassed only a single state; however, future research should extend the scope of the investigation to determine if the model holds over a larger geographic area.
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
