danmelton.net

Data and Methodology

To test the above, I collect data for 529 central cities from the FBI on Hate Crimes; the Human Rights Campaign and various reputable Internet sources for city and state civil rights laws; and the 2000 US Census for same-sex partnered households. The following covers each measure:

Same-Sex Partnered Households

I use Census 2000 summary 3 file data to obtain same-sex partnered households at the tract level. Same-sex households should not be construed as a measure of overall gays and lesbians. The census only measures gays and lesbians that check a box indicating an unmarried partner; research shows that these households are on average more educated, have higher incomes and are mostly white (Badgett 2002). Also, bisexual and transgender persons are not measured independently. This measure is only a proxy and should be taken with caution in interpreting the results. See Bell and Binnie (2004) and Badgett (2002). Same-sex partnered households are totaled and considered concentrated if they exceed each location quotient computed for that tract. For instance, if a tract has a location quotient of 1 or greater, then all of its same-sex partnered households would be included in the concentration percentage. After all tract results are totaled, that number is divided by the total number of same-sex partnered households in the entire central city, suburb or region.

Central Cities

Central cities are defined as the largest place in a metropolitan region. Some metropolitan regions have several central cities. For instance, Kansas City is comprised of Kansas City, MO and Kansas City, KS as well as outlying non-central suburb cities. Using 1999 census-defined designations, I focus on central cities as the unit of analysis for three reasons. First, agencies reporting hate crimes tend to be located in the urban core, encompassed by central cities. Second, central cities tend to be the primary adopters of civil rights ordinances for gays and lesbians. Third, central cities provide a cross boundary specific measure on each of the variables, where as a regional analysis would have to account for differing agencies reporting hate crimes, or individual cities within a region that may or may not have civil ordinances.

Political Tolerance

Gastil (1988) pioneered measures of political tolerance using civil rights and similar laws at the country-level. He writes that civil liberties “are rights to free expression, to organize or demonstrate, as well as rights to a degree of autonomy such as is provided by freedom of religion, education, travel, and other personal rights.” To measure civil liberties, Gastil created a ‘status of freedom’ index, assigning a value of 1 to 7 to countries, with 7 being the least politically tolerant. He added each law to an index, and then comparatively created the scale. Similarly, I measure political tolerance using civil rights ordinances passed by city governments. To measure political tolerance, I use research conducted by the Human Rights Campaign, plus additional web sources, to find pro and anti-gay rights laws for central cities. Focusing on laws enacted up to 2001, I construct an additive index by subtracting the adoption date of each law from 2002; scores for each law are added to create a total for that city. I use an additive index for three primary reasons. First, a categorical index limits the distribution, placing cities with far different sets of political tolerance near each other numerically. Second, an additive index makes it less likely that two cities will have the same scores over a categorical index. Additionally, if a city did not have a pro or anti-gay ordinance, but the county did, I included the county’s adoption date. Finally, an additive index captures the historical element of civil rights ordinances. While Gastil did not account for time since law adoption, I believe this to be an important aspect of how cities are perceived as tolerant. Presumably, the longer a civil rights law has been enacted, the more tolerant a city will be perceived.

Social Tolerance

To measure social tolerance, I rely on anti-gay/lesbian hate crimes reported to the FBI. This data was compiled and distributed via the National Archive of Criminal Justice Data in partnership with Interuniversity Consortium for Political and Social Research (ICPSR). The data begins in 1992 after Congress passed national hate crimes legislation requiring that police agencies track and report hate crimes. To maintain continuity across data sources, I use 1999 to 2001 reported crimes for central cities. , Hate crimes are bias motivated where the individual is picked out solely because of a perceived trait such as gender, religion, race, disability or sexual orientation. Victims call the police, who ask FBI-guided questions to determine whether or not the crime was motivated by hate. If so, the crime is reported to the FBI. The FBI provides training guides and technical assistance to reporting agencies.

Before moving on, I should address issues related to reporting of hate crimes by police and victims. Wide claims by both scholars and advocacy groups suggest that police officers are reluctant to report an incident as a hate crime . Green et al compared police reported FBI hate crimes with anti-violence project reports. They find that both sources strongly correlate with each other and the location and densities of gays. Advocacy agencies reported many more hate crime incidents than those recorded by the FBI. Yet, Green finds that these also include generalized heterosexist incidents. Hate crimes meet a higher litmus test than heterosexist incidents. The individual must be targeted solely due to prejudice and the perpetrator generally has to have a background of prejudice. For instance, name calling during the incident does not imply a hate crime. Even so, they admit that FBI hate crimes may be underreported, but at a proportional rate.

Others might argue that bias could exist in reporting procedures between metropolitan areas due to police prejudice or victim awareness of hate crime reporting. Cities that have higher hate crimes could merely be an artifact of less biased police procedures. Perhaps police in cities like San Francisco or New York have been pressured to report hate crimes by active gay and lesbian communities over those in cities such as Baton Rouge or Biloxi. Advocacy groups work to ensure that police officers are held accountable to reporting hate crimes; they also work to inform the public about the need to report them. The National Coalition of Antiviolence Projects reports that 18 metropolitan areas host anti-violence projects with the express purpose of advocating on behalf of gay and lesbian victims, including hate crimes . I compared the prevalence of hate crimes before and after the start dates of these programs to ascertain whether or not an antiviolence program might indeed affect the reporting of hate crimes. Ten projects had start dates within the ten years of hate crime data, resulting in 480 cases. No significant difference could be found between the years before and the years after the start of an antiviolence project even with lagged variables and de-trending. While not an exhaustive data analysis, it appears that reported hate crime rates are not affected by the presence of an advocacy group.

Concentration

To measure same-sex partnered households, I use location quotients similar to Florida’s analysis. Location quotients provide a standardized measure that accounts for the group population while controlling for the larger base population. Using census tracts as a proxy for neighborhoods, I create a location quotient for same-sex partnered households at the tract and central city level. I use the total pool of central cities as the base population, rather than national or regional populations. To obtain percent concentrated, I use the tract level location quotients rather than a flat percent. Location quotients utilize a ratio method that (1) considers the proportion of the group that locates to an area (i.e. neighborhood) from the entire region (i.e. metropolitan area) and (2) relates that to the population that located to the area compared to the entire population for the region. I present data at four levels of location quotients, 1, 2, 3 and 5. After 5, the population concentrated in tracts approaches zero across all metropolitan regions.

Method

To test the above model, I first categorize central cities by location quotients. Splitting central cities into very low through very high location quotients, represents the propensity of same-sex households to locate to central cities. Cities with high location quotients would have higher proportions of same-sex households and thus more likely to exhibit characteristics of the expanded and assimilated aspects of the model. After presenting location quotients categories and relationships with tolerance and concentration, I cluster the four measures to see if the results fit the model presented above.