Defining Worker Legal Mobilization

There are various ways that scholars have generally conceptualized legal mobilization. Many studies have focused on the highest levels of the dispute pyramid, analyzing processes of court litigation (e.g., Reference BursteinBurstein 1991; Reference Burstein and MonaghanBurstein & Monaghan 1986; Reference EppEpp 1998; Reference McCannMcCann 1994; Reference Nielsen, Nelson, Lancaster and PedrianaNielsen, Nelson, Reference RosenbergLancaster, & Pedriana 2008b; Rosenberg 1991). Yet rights are often accessed outside of the courts, either collectively through social movements, or individually through administrative enforcement agencies. Many of the claims that are mediated by these agencies never reach the courts, and factors shaping the emergence and fate of these claims have been explored both qualitatively (e.g., Reference Albiston, Nielsen and NelsonAlbiston 2005; Reference BumillerBumiller 1992; Reference MerryMerry 1990) and quantitatively (e.g., Reference HirshHirsh 2008; Reference Hirsh and KornrichHirsh & Kornrich 2008; Reference Wakefield and UggenWakefield & Uggen 2004).

Federal and state labor laws have become even more significant in combating workplace violations in an increasingly nonunionized economy. Discrimination charges under Title VII of the Civil Rights Act represent only one aspect of workplace violation claims filed by workers, though scholars have focused overwhelmingly on this area of employment law. The fact that the EEOC has the most extensive data available compared to other labor enforcement agencies more than likely accounts for this fact.Footnote ¹ Other employment statistics, most notably wage and hour violations, vary widely due to differences in state regulations and varying enforcement standards.

Under the Fair Labor Standards Act (FLSA), the U.S. Department of Labor is authorized to enforce employment protection laws, and in 2008 more than 23 states had enacted standards that exceeded the federal requirements (U.S. Department of Labor Wage and Hour Division 2008: n.p.). State and local fair employment practice agencies may also process federal claims, which are incorporated into EEOC statistics and are readily available. However, there is no uniform data collection or reporting mechanism for federal and state agencies tracking wage and hour violations (Reference Bernhardt and McGrathBernhardt & McGrath 2005). Several surveys have assessed the extent of wage and hour violations, particularly for low-paying jobs and immigrant workers (e.g., Reference MehtaMehta et al. 2002; Restaurant Opportunities Center of New York 2005; Reference ValenzuelaValenzuela et al. 2006), but they do not provide representative data about factors driving claims based on these violations nationwide.

In general, scholars have not addressed the legal mobilization process for immigrant populations, primarily because of limited available data.Footnote ² While the growing immigrant workforce is a key demographic for organized labor (Reference NessNess 2005), foreign-born workers are less likely to be unionized compared to native-born workers; nationally, Latinos overall have the lowest levels of union membership (Reference Milkman and KyeMilkman & Kye 2007). Labor scholars interested in the claiming process for immigrants have tended to employ qualitative methods to examine occupational health and safety issues, and wage and hour violations (e.g., Reference AzaroffAzaroff et al. 2002; Reference ChoCho et al. 2007; Reference NissenNissen et al. 2008; Reference SmithA. Smith 2005; Reference SmithK. Smith 2003), while the context of discrimination for immigrant workers has been mostly overlooked. This is due in large part to the immediacy of worker safety and pay issues, compared to the broader and sometimes more elusive kinds of discrimination. The investigative process for discrimination, versus other areas of workplace violations, is also distinct. One EEOC director I spoke to compared the process for his agency to that of the U.S. Department of Labor:

Thus, while discrimination may require a “higher standard of proof” and be a more complex claim for workers to file, it is nonetheless still significant to minority and immigrant workers.

The Context of National Origin Discrimination

Discrimination, in particular based on national origin, was a prominent concern amongst legislators and advocates following the imposition of employer sanctions under the Immigration Reform and Control Act of 1986. Many feared that employers would make assumptions about workers' immigrant status in efforts to avoid violating the law. Consequently, the federal government dedicated specific resources to monitor such situations by establishing the Office of Special Counsel for Immigration-Related Unfair Employment Practices at the U.S. Department of Justice, which was charged with ensuring that employer sanctions did not unfairly impact immigrant-origin workers and did not accelerate the exploitation of undocumented workers. Workers are protected from discrimination based on national origin under Title VII of the 1964 Civil Rights Act (Equal Employment Opportunity Commission 2008a). The EEOC defines national origin discrimination as “treating someone less favorably because he or she comes from a particular place, because of his or her ethnicity or accent, or because it is believed that he or she has a particular ethnic background.” It can also include “treating someone less favorably at work because of marriage or other association with someone of a particular nationality” (Equal Employment Opportunity Commission 2002a: n.p.). In practice, national origin discrimination complaints often overlap with other kinds of discrimination, such as race (Reference Nielsen, Nelson and LancasterNielsen, Nelson, Reference Parker& Lancaster 2008a; Parker 2005).

National origin discrimination can include practices such as unlawfully enacting a citizenship or language requirement for hiring or promoting a worker, or unlawfully harassing or retaliating against a worker based on his or her ancestry. These protections extend to all workers regardless of their nativity or immigration status. The EEOC compliance manual acknowledges that “federal law prohibits employers from employing individuals lacking work authorization” but also emphasizes that “employers who nonetheless employ undocumented workers are prohibited from discriminating against those workers” (Equal Employment Opportunity Commission 2002a: n.p.).

Following the 2002 Supreme Court case Hoffman Plastics v. National Labor Relations Board,Footnote ³ the EEOC issued a statement declaring its commitment to “root out discrimination against undocumented workers” and its commitment to not“inquire into a worker's immigration status or consider an individual's immigration status when examining the underlying merits of a charge”Footnote ⁴ (U.S. Department of Labor Wage and Hour Division 2007: n.p.). The EEOC has also led many outreach initiatives targeted at immigrant workers in several U.S. cities (Equal Employment Opportunity Commission 2002b). But while national origin discrimination is an often overlooked workplace concern, usually overshadowed by a focus on occupational health and wage and hour violations, these charges are nonetheless significant for the growing immigrant workforce.

This study assesses the aggregate variation in levels of national origin discrimination claims for all 50 states (excluding the District of Columbia) over the period 2000–2005. This analysis is a pilot study to help legal and public policy scholars identify key institutional variables driving legal mobilization and determine future research in other areas. This analysis focuses on the impact of civil society on legal mobilization, controlling for the economic and policy context, for two vulnerable populations: nonunion workers and Mexican immigrants. I propose four hypotheses, which I outline below.

The Third Sector—Tool for Public Action

Labor unions have typically been the primary advocate for workers. Unionized workplaces are generally better paid, have more generous benefits, and have higher levels of worker satisfaction (Reference Bennett and KaufmanBennett & Kaufman 2007; Reference Budd, Bennett and KaufmanBudd 2007). Unions can mobilize on behalf of their members by filing unfair labor practices complaints, but the 21st century has seen the lowest rates of union membership since the 1930s (Reference LichtensteinLichtenstein 2002). Most American workers are not represented by a collective bargaining contract and instead rely on federal and state bureaucracies to govern the conditions of their employment. When a labor violation occurs, workers can file a complaint with the appropriate agency, but they may confront a number of obstacles such as lack of knowledge, fear, or resistance from their employer. In the absence of a union, the third sector has become increasingly important for helping workers to overcome such obstacles to achieve their individual employment rights (Reference Jolls and FreemanJolls 2004; Reference PollertPollert 2005).

While conservatives tend to see the third sector as a replacement for government, other scholars, such as Reference Salamon and SalamonSalamon (2002), characterize nongovernmental organizations (NGOs) as a crucial “tool for public action” and a key element in an emerging “third-party government” structure. Low cost or pro bono legal services are perhaps the most visible aspect of nonprofit assistance for legal mobilization (e.g., Reference ZemansZemans 2006), but they constitute only a small part of the role that civil society can play in the workplace. NGOs are well-known in the international community as monitors and advocates for improving labor standards and promoting government accountability (Reference Doh and GuayDoh & Guay 2004; Reference GereffiGereffi et al. 2001; Reference Keck and SikkinkKeck & Sikkink 1998; Reference O'RourkeO'Rourke 2003). In the United States, these third sector organizations can also help mediate barriers that workers face in accessing their rights. Civil society organizations can raise consciousness about workplace rights and may provide workers with information and resources to file claims.Footnote ⁵

Mexican Immigrants and Undocumented Status

Largely due to limited data, scholars know relatively little about the mechanisms driving immigrant legal mobilization.Footnote ⁶ Immigrants are a growing segment of the American workforce (Pew Hispanic Center 2008), and undocumented immigrants constituted roughly 40 percent of all immigrants arriving in the United States in 2005 (or about 5 percent of the American civilian labor force [Reference PasselPassel 2006:ii]). These workers experience employment violations at high rates and confront significant barriers to accessing their rights in an increasingly nonunionized workplace. For many undocumented workers, there is a reluctance to engage government agencies for fear of deportation, language barriers, and lack of access to legal counsel (Reference FineFine 2006; Reference GordonJ. Gordon 2007; Reference ValenzuelaValenzuela et al. 2006).

Mexican immigrants are not the only category of workers eligible to file discrimination claims based on national origin. However, they do represent one of the most vulnerable populations by virtue of their economic standing and immigration history. The largest concentrations of Mexican immigrants are in traditional receiving states, which border Mexico, but like other immigrants, they have begun moving into other states, primarily in the South (Reference BatalovaBatalova 2008). Mexican immigrants represent the largest foreign-born population in the United States and have one of the lowest socioeconomic profiles (Pew Hispanic Center 2008). The immigration history of Mexicans is distinct from that of other immigrant communities. The vast majority (80–85 percent) of Mexican immigration in recent years is undocumented, and more than half (57 percent) of the overall undocumented U.S. population is Mexican (Reference PasselPassel 2006:4). Undocumented workers face both the challenge of being foreign-born and of lacking legal status.

While the Mexican population is an imperfect proxy for the undocumented population across the United States, the probability that a Mexican immigrant is undocumented is relatively high and has been shown to impact community norms for accessing rights and benefits. This point is still relevant despite the likelihood that most Mexican immigrants who file claims are likely to be documented, either native-born or with legal resident status. Therefore not all Mexicans who file claims need to be undocumented in order for high percentages of undocumented workers in the community to have an influence on the overall rates of legal mobilization. For example, Reference Van Hook and BeanVan Hook and Bean (2009) discuss the role of “employment-based cultural repertoires” for Mexican immigrants accessing welfare benefits. Their findings suggest that the effect of undocumented status is not simply an instrumental liability that hinders access to institutional rights, but is also a legal definition that emphasizes an individual's labor function in society while imposing cultural and psychological barriers to accessing rights.Footnote ⁷

Civil Society as a Mediating Institution

Undocumented workers, like many low-income workers, are often unable to afford private legal assistance when a workplace violation occurs. This is compounded by the fact that legal aid societies that provide alternative low-cost or pro bono assistance may be reticent or unable to serve some workers.Footnote ⁸ Such legal assistance is largely funded through the Legal Services Corporation, a federally funded program that prohibits serving undocumented workers, a growing population that is particularly at risk (Legal Services Corporation 2006). In the absence of direct access to legal counsel, civil society organizations often play a mediating role for promoting legal mobilization. These are the institutions closest to local communities, and they possess a unique advocacy relationship to these communities that earns them trust (Reference BloemraadBloemraad 2006; Reference RamakrishnanRamakrishnan 2005).

The challenges facing immigrant, and especially undocumented, workers are well known to labor standards enforcement agencies, as are the benefits that civil society organizations can provide in bridging this gap. For example, the outreach manager of the San Francisco EEOC explained, “The challenge is getting them to trust us. And we work very closely with other organizations to do that …. I'm a strong believer in collaborating with migrant workers” (Interview with outreach manager, EEOC, San Francisco, 4 Aug. 2006). She and the agency director went on to discuss the importance of building credibility in the community.Footnote ⁹

Post-9/11, similar challenges in Asian- and Arab American communities emerged. However, demographically, the Mexican community represents the largest ongoing constituency for the San Francisco agency, and community relationships have been key to reaching these workers. According to the EEOC outreach manager for the agency, one example was the recent success in a high-profile case involving a Mexican immigrant agricultural worker who had suffered sexual harassment:

This example suggests that civil society organizations may be particularly significant in workplaces with higher concentrations of workers who face barriers to legal mobilization, as exemplified by the Mexican community. Research shows that such organizations function as important agents and intermediaries for implementing existing policies and enforcing workers' rights more broadly, particularly for immigrant communities (Reference De Graauw, Ramakrishnan and Bloemraadde Graauw 2008; Reference PutnamPutnam 2000; Reference SkocpolSkocpol 2003; Reference VerbaVerba et al. 1993; Reference WeilWeil 2003). Though the third sector is diverse in terms of mission and purpose, civic organizations have been used to communicate information, particularly to the immigrant community, about their rights and resources available to them, and they are an important component of immigrant incorporation processes (Reference BloemraadBloemraad 2006; Reference Cordero-GuzmánCordero-Guzmán 2005). Community-based organizations can provide information to workers about their rights, encourage them to come forward when a violation occurs, and help them by guiding them through the process of submitting a claim. Beyond this important service, these organizations possess cultural literacy that may also help break through cultural barriers that make immigrant workers less likely to want to complain (Reference ChoCho et al. 2007; Reference FineFine 2006; Reference GordonJ. Gordon 2007).

Last, unions may provide workers with better working conditions and alternative grievance mechanisms. Consequently, civil society organizations may be particularly important at this present juncture when union membership is waning. Nonprofit organizations inform workers of their rights through the creation of social networks, facilitate governance through referrals to government agencies, and in some cases provide guidance in navigating the claims-making process—all of which are functions traditionally provided by organized labor.

Given these four hypotheses, the next section reviews the analytical model used to assess these relationships.

Analytical Model

This analysis models the rate of national origin discrimination claims in a state-year, λ_it, with a negative binomial regression model. The negative binomial regression model is commonly used with data such as these, where the analysis requires modeling the occurrence of discrete events, which are over-dispersed—that is, more variation than would be expected with an ordinary Poisson model (Reference Cameron and TrivediCameron & Trivedi 1998).Footnote ¹⁰ A specific expression of the negative binomial regression model is used for panel data, which includes state fixed effects (Z_i) to control any unobserved variation amongst the different states that may emerge.

$\text{Equation}\hspace{0.17em}\text{4}.1:\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\text{In}({\lambda }_{\text{it}}/{\eta }_{\text{it}}{)=\alpha }_{\text{i}}+{\beta \text{X}}_{\text{it}}+{\text{YZ}}_{\text{i}}+\in$

The model includes an exposure variable, η_it, which measures the “population at risk” for filing a national origin claim in each state (i) – year (t). In conjunction with the dependent count variable (i.e., the number of national origin claims in each state-year), the exposure variable acts as a denominator to provide a predicted rate of claims-making. The Appendix provides a conceptual overview of some ways to conceptualize and define this variable. The model uses group C, “all workers of color (non-white and/or Hispanic).” Data are drawn from the American Community Survey (U.S. Census Bureau, 2000–2005). (For a similar approach, see Reference RubensteinRubenstein 2001, which models rates of sexual orientation discrimination claims across states using estimates of the lesbian, gay, bisexual, and transgendered population in the workforce as the exposure variable.)

Conversely, the other alternative “exposure populations” presented in the Appendix are inappropriate. For example, to use group B, “all workers,” would inflate the count of workers who are likely to file a national origin claim and consequently deflate predicted levels of claims-making. Alternatively, to use group D, “all workers of color who are foreign-born,” would underestimate the count of potential claimers, thus inflating rate estimates. Indeed, a worker need not be foreign-born to experience national origin discrimination (Equal Employment Opportunity Commission 2002a).Footnote ¹¹

Measuring Legal Mobilization

Limited data are available on the individual characteristics of workers who actually file claims. Not only are these data kept strictly confidential and not readily released to the public by labor standards enforcement agencies, but many important characteristics of claimants (such as socioeconomic status and immigration status) are not recorded in any systematic manner. As a result, an individual-level analysis that interrogates the factors driving claims-making is very difficult, and researchers are left largely with aggregate analyses.

Certainly, aggregate analyses reveal different mechanisms than do individual analyses. They cannot predict an individual's propensity to engage in an outcome, and they cannot substitute for the precision that micro-level data provide (Reference Geronimus and BoundGeronimus & Bound 1999; Reference WooldridgeWooldridge 2002). Aggregate analyses, however, can offer information on contextual factors driving aggregate levels of claiming (Reference EppEpp 1990; Reference McVeighMcVeigh et al. 2003). Within this framework, the objective of this analysis is to account for variation in levels of legal mobilization across economic and policy contexts. In the United States, worker protection and enforcement vary widely from state to state, as do the demographic and industrial makeup of the workforce, and the extent of civil society resources available to promote legal mobilization for aggrieved workers.

This analysis draws on data provided by the EEOC, obtained through a public records request. The data include all claims filed under federal statute across the country for national-origin discrimination between 2000 and 2005. In 2005, national origin discrimination charges accounted for 11 percent of all charges processed by the EEOC, and 15 percent of all Title VII charges.Footnote ¹² I chose to focus on national origin claims specifically for the purposes of theoretical and methodological clarity.

First, focusing on one specific statute eliminates the risk of double-counting actual charges, as a single charge may include multiple statute allegations. In 2005, for example, the 55,461 actual charges filed represented 62,197 statute allegations. (There were 3,149 claims filed under Title VII color provisions, 14,349 claims under Title VII national origin discrimination provisions, and 44,699 claims under Title VII race provisions, for a total of 62,197 statutes. Further information available from author upon request.)

Second, by focusing on cases involving national origin discrimination, I can analyze specific information about the claimant population at risk. Title VII protections cover all workers regardless of nativity or ancestry, and the population most likely to file a particular claim varies for each. The EEOC does not collect data on the nativity or immigration status of its claimants, though other data shed light on the demographic composition of charging parties.Footnote ¹³ For example, Reference Nielsen, Nelson, Lancaster and PedrianaNielsen, Nelson, Lancaster, and Pedriana (2008b) analyzed a random sample of employment discrimination litigation cases from 1987 to 2003, focusing specifically on race, sex, age, and disability claims. They found that across all categories, African Americans file the largest share of claims (44 percent), followed by whites (33 percent), other (13 percent), Hispanics (7 percent), and Asians (3 percent) (2008b:9).

Based on my own assessment of the data, close to half of all national origin discrimination claims are filed by Hispanic claimants, compared to only 6 percent of those who file claims on the basis of color discrimination, and 3 percent who file based on race discriminationFootnote ¹⁴ (see Table 1). This distribution is likely an artifact of how the federal government defines Hispanic as a separate ethnic group that is not mutually exclusive to other racial categories (Reference HitlinHitlin et al. 2007; Reference RodriguezRodriguez 2000). Similarly, national origin discrimination is not exclusive to other bases for discrimination claims, such as race and color (Equal Employment Opportunity Commission 2002a). As a result, these distributions alone do not necessarily speak to the specific discrimination experiences of different groups, but they do, however, make analysis of them more complex. Consequently, an analytical approach that combines various bases of discrimination would be complicated by the difficult task of defining the appropriate “at-risk population.” Furthermore, given that this study's focus is on the Mexican immigrant workforce, and the impact of undocumented status on that community, this analysis concentrates solely on national origin discrimination claims. Nonetheless, many of the lessons learned here may be useful in other national origin communities.

Table 1. National Origin Discrimination Claims, by Race and Hispanic Status of Claimants

The geographic distribution of national origin claims is represented in Figure 1. In 2005, claims levels varied from seven claims per 100,000 minority workers in Mississippi, to 327 in Nebraska (further information available from author upon request). One of the benefits of disaggregating claims data by state is that it provides a way to assess the influence of various contextual factors such as industrial composition, demographic makeup of the workforce, and labor standards.

Figure 1. Number of National Origin Discrimination Claims per 100,000 Workers Who Are Not Non-Hispanic White United States, 2005

One of the challenges, however, is to disentangle which mechanisms affect the rates of employer violations of these discrimination protections, versus those that propel legal mobilization by aggrieved workers. Reference McVeighMcVeigh et al. (2003) offer an instructive framework for understanding legal mobilization given this dilemma, using the aggregate context of hate crime reporting. They argue that the “differences in the number of hate crimes reported in various regions do not simply reflect differences in the number of criminal acts motivated by bias” and may instead reflect “different incentives to call acts of bias to the attention of local authorities, as well as different incentives that influence law enforcement agents to respond to and to report, hate crimes” (2003:846). As such, the authors treat the level of hate crimes reported as the measure of a “successful social movement outcome, rather than a valid measure of a particular type of crime” (2003:846).

However, two processes—employer violation practices and workers' propensity to engage in legal mobilization—could impact observed levels of claims-making simultaneously. For example, more stringent labor standards might dissuade employers from discriminating against workers, while simultaneously encouraging aggrieved workers to come forward. Similarly, a tight labor market for employers (i.e., one with lower rates of unemployment that foster competition for talent) may also discourage discrimination, while also providing workers with fewer incentives to pursue claims against a hostile employer when alternative jobs are available. Therefore, an aggregate analysis such as this cannot distinguish between these different mechanisms. However, by controlling for economic and policy differences such as partisan strength of labor enforcement standards, this analysis seeks to account for these complex countervailing forces.

Key Independent Variables

The primary independent variable in this analysis is the size of civil society. Social scientists have discussed the importance of civil society in various contexts, including civic and political participation, social movements, and legal mobilization. Due to its wide-ranging influence, this sector has been defined and operationalized in many different ways. Broadly speaking, most social scientists define civil society as the third sector, which functions between the market and state (Centre for Civil Society 2004). The most complex assessments incorporate a large number of indicators to measure the cultural and structural elements of civil society (Reference HeinrichHeinrich 2005). Some, such as Reference PutnamPutnam (2000), even extend the definition of civil society to include broader forms of social capital through various networks, while others, such as Reference SampsonSampson et al. (2005), focus on concrete collective action events. Institutionally, scholars tend to measure civil society by the number of voluntary nonprofit associations (Reference Skocpol and FiorinaSkocpol & Fiorina 1999).

In this analysis, civil society is operationalized as the number of registered nonprofit organizations in a state-year. The National Center for Charitable Statistics (NCCS) provides a count of the number of official 501c(3) registered organizations for every state-year,Footnote ¹⁵ broken down by National Taxonomy of Exempt Entities (NTEE) code.Footnote ¹⁶ This analysis uses NCCS data to test the effect of both the aggregate count of all 501c(3) organizations per 1,000 residents, and the count of organizations in the NTTE major category “R: Civil Rights, Social Action, and Advocacy” (CRSAA organizations) per 1,000 residents (see also Reference McVeighMcVeigh et al. 2003 for a similar operationalization). CRSAA organizations represent “private nonprofit organizations whose primary purpose is to protect and promote the broad civil rights of groups and civil liberties of individuals, to work for the realization of specific social or political goals or to encourage the participation of people in the public policy debate” and include “organizations that work to improve relations between racial, ethnic, and cultural groups; advocacy and citizen action groups that work to change public policy and opinion in a variety of areas; organizations that use courts to protect and enlarge civil rights and liberties; and organizations that promote voter education and registration” (National Center for Charitable Statistics 2008: n.p.). These CRSAA organizations are more directly involved in issues surrounding discrimination protections, though past research suggests that voluntary associations of all kinds could be relevant to disseminating information and providing resources for a variety of issues, particularly for immigrant communities (Reference BloemraadBloemraad 2006; Reference Cordero-GuzmánCordero-Guzmán 2005; Reference De Graauw, Ramakrishnan and Bloemraadde Graauw 2008; Reference MarwellMarwell 2007; Reference RamakrishnanRamakrishnan 2005). (The analysis presented below focuses on results for the models including CRSAA organizations, though the findings do not differ significantly from the model measuring all organizations.)

This analysis also includes a measure of the size of the Mexican immigrant population in a state-year. The focus on the Mexican immigrant population is theoretically motivated by high levels of undocumented status in this community, as described above, which is likely to deter formal claims-making and benefit most from nongovernmental civil society resources. While other immigrant groups may share similar characteristics, no other immigrant group is as well distributed across the United States to warrant a national analysis. Data come from the American Community Survey for each state-year.

The model also tests the effect of the strength of organized labor using a measure of union membership for each state-year. These data are drawn from published news releases from the Bureau of Labor Statistics (Bureau of Labor Statistics 2008a), and represent the total union affiliation of employed wage and salary workers.

Two multiplicative terms are included: (1) a two-way interaction between the density of nonprofit organizations and the size of the Mexican immigrant population, and (2) a two-way interaction between the density of nonprofit organizations and union membership levels, which tests the mediated effect of civil society at different levels of organized labor participation. (Tests confirm that multicollinearity between the size of civil society and union membership is not driving model results.Footnote ¹⁷)

Control Variables

In addition to these key variables, the study also controls for several economic and policy factors that shape workplace conditions and may influence levels of worker legal mobilization. First, a measure of industrial concentration is included based on data from the American Community Survey (ACS). Data from the EEOC do not include detailed information on the industrial distribution of claimants and rely on the 1987 Standard Industrial Classification (SIC) system. Claimants are classified by agency investigators, which results in further inconsistent categorization. (Available data reveal that one-fifth of all claimants could not be categorized.) Drawing on the North American Industrial Classification scheme, this analysis controls for the proportion of the workforce in four industrial sectors: agriculture, forestry, fishing and hunting (1); construction (4); manufacturing (5); and arts, entertainment, recreation, accommodations, and food services (13). Based on current population survey data, these low-wage industries have a higher share of foreign-born workers than native-born workers (Pew Hispanic Center 2008) and are also major destinations for undocumented workers (Reference PasselPassel 2006). Additional information could be provided by using the full industrial distribution, though this would be at the expense of statistical efficiency.

The economic climate has also been shown to drive claims-making. The model controls for poverty and unemployment levels in a state-year. The poverty rate is based on the percent of individuals who are living in a family whose income falls below the federal threshold for poverty (100 percent of the threshold coded in the ACS.) The model also includes a measure of the annual average unemployment rate reported by the Bureau of Labor Statistics for the particular state-year (Bureau of Labor Statistics 2008b). Reference Wakefield and UggenWakefield and Uggen (2004) have found that high rates of unemployment are a key factor driving levels of EEOC complaints; Reference Donohue and SiegelmanDonohue and Siegelman (1991) have found that the single most important factor driving the increases in the number of employment discrimination suits filed in federal district court during the 1970s and 1980s was a rise in the unemployment rate.

National origin discrimination is a federal protection under Title VII of the Civil Rights Act, though claims levels are likely to be influenced by the overall climate of workplace protections in a state. Employers in places with more worker-friendly policies may be less likely to discriminate, just as workers in these places may be more encouraged to make claims. The model controls for state labor standards using a measure of the minimum wage. The Fair Labor Standards Act sets the national minimum wage that must be given to workers, but states can also formally legislate higher or lower standards, or choose not to enact any standard at all (U.S. Department of Labor Wage and Hour Division 2008). States with no official standard are assigned the federal minimum, while states with rates below the federal minimum are assigned the highest state official rate.Footnote ¹⁸

Lasty, because bureaucracies are influenced by the political climate in which they operate, I control for two measures of partisan strength. Partisan control can impact the appointment of key administrators in regulatory agencies and courts (e.g., Reference Brace and HallBrace & Hall 1997; Reference Cross and TillerCross & Tiller 1998; Reference Tiller and CrossTiller & Cross 1999). Reference Wakefield and UggenWakefield and Uggen's (2004) analysis produces null results for the effects of partisan strength, though they offer several reasons to believe that economic and political changes affect levels of claims-making in other contexts. Reference McVeighMcVeigh et al. (2003) also control for political factors, arguing that partisan strength may affect the saliency of civil rights organizations. This model includes two dummy variable measures for each state-year: (1) whether a Democratic governor is in power, and (2) whether the state legislature has a Democratic majority.Footnote ¹⁹ Data are drawn from reports published by the National Conference of State Legislatures (http://www.ncsl.org, accessed 25 March 2009).

The results discussed below are calculated as mean predicted rates from the full model, using mean values in a Democratic context (see Table 2 for descriptive statistics).Footnote ²⁰ Additional data are also provided from interviews with labor standard enforcement agents in the San Francisco Bay area and the Houston metropolitan area in 2005–2006.

Table 2. Descriptive Statistics