Institutions, Politics, and Mental Health Parity

Society and Mental Health 2(3) 154–171 Institutions, Politics, and Ó American Sociological Association 2012 DOI: 10.1177/2156869312455436 Mental Health Parity http://smh.sagepub.com

Elaine M. Hernandez1 and Christopher Uggen2

Abstract Mental health parity laws require insurers to extend comparable benefits for mental and physical health care. Proponents argue that by placing mental health services alongside physical health services, such laws can help ensure needed treatment and destigmatize mental illness. Opponents counter that such mandates are costly or unnecessary. The authors offer a sociological account of the diffusion and spatial distribution of state mental health parity laws. An event history analysis identifies four factors as especially important: diffusion of law, political ideology, the stability of mental health advocacy organizations and the relative health of state economies. Mental health parity is least likely to be established during times of high state unemployment and under the leadership of conservative state legislatures.

Keywords mental health, mental health policy, mental health services

Most citizens of the United States acquire health was too limited and attenuated to ensure mental care through employers. Although this employer- health parity. Following its sunset in 2005, based system has typically covered most physical Congress once again passed federal mental health illnesses, mental health services have historically parity legislation, the Mental Health Parity and been specifically excluded or minimally covered Addiction Equity Act of 2008 (effective July (Frank and McGuire 2005). Mental health parity 2010), which more forcefully prevents employers laws mandate that employers provide equal types from exempting patients with mental illness. of benefits for mental and physical illnesses Given the infrequency of federal legislation and (Mechanic and McAlpine 1999). An examination the weakness of the 1996 federal law, the battle for of the contested history of these laws brings to light mental health parity was fiercely contested at the the eroding boundary between mental and physical state level between 1980 and 2005. The purpose health; the shifting rights and obligations of gov- of this article is to understand and model the factors ernment, business groups, and citizens; and the precipitating passage of these state mental health emerging social acceptance of mental illness and parity laws and to situate our findings within its treatment (Goffman 1963; Kershaw 2008; a broader explanation of legal and policy change. Link et al. 1999; Martin, Pescosolido, and Tuch We use ideas from medical sociology, political 2000; McSween 2002; Scheff 1964). States have imposed a variety of mental health parity mandates on employers since the 1980s 1University of Texas at Austin, Austin, TX, USA (Caronna 2004), ranging from no mandates to com- 2University of Minnesota, Minneapolis, MN, USA prehensive mandates. In an attempt to consolidate Corresponding Author: the patchwork of state parity laws, the U.S. Elaine M. Hernandez, University of Texas at Austin, Congress passed the Mental Health Parity Act Population Research Center, 1 University Station, G1800, (MHPA) of 1996. The MHPA was an important Austin, TX 78712, USA symbolic accomplishment to be sure, though it Email: [email protected]

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 Hernandez and Uggen 155 sociology, and the sociology of law to understand treatments through employer-based health insur- variation in state mental health parity legislation ance plans (Mechanic 2003:1229). Beyond equal- between 1980 and 2005. Bridging these literatures, ity, fundamental economic forces are consistently we draw on Amenta’s (1998) institutional politics at the forefront of debates about mental health par- theory to explain state passage of parity ity, as with health care more generally. laws. From a political perspective, we consider Mental illnesses are frequently chronic, requir- government ideology, and from an institutional per- ing long-term care, and individuals with mental ill- spective, we consider economic pressures, organi- nesses run a high risk of having comorbidities that zational structures, and diffusion of law. may require expensive long-term care (Mechanic At the most basic level, mental health parity 2003, 2004; Sturm and Pacula 1999). When offer- seeks to provide equality for people with a mental ing health insurance to such populations at risk illness, allowing their continuous care. Beyond for incurring expenses, insurance providers con- the economic and health care concerns associated sider two key economic forces: moral hazard and with mental illness, such laws symbolically seek adverse selection (Frank, Koyanagi, and McGuire to ‘‘delabel’’ (Trice and Roman 1970) and destig- 1997). ‘‘Moralhazard’’refers to situations in which matize mental illness (Mechanic 2003). Parity leg- increased coverage motivates individuals to islation thus provides a window into the political demand more services, particularly as the cost asso- and institutional conditions that helped tear down ciated with each service drops. Adverse selection the wall between physical and mental health cover- takes place when individuals enroll in specific age, resulting in an increasing acceptance of, and health plans because they anticipate needing spe- even expectation for, equal mental health coverage. cific types of coverage, resulting in a higher proportion of ‘‘high-risk’’ individuals in those plans. Employers and insurers are particularly sensitive BACKGROUND to the problem of moral hazard, and there is some evidence that moral-hazard problems in mental In the United States, the state-level battle over health care are especially acute (Frank et al. employer coverage of mental health services fol- 1997; Newhouse 1993). Yet evidence suggests lowed significant shifts in health care reimburse- that the introduction of managed behavioral health ment and mental health care in the United States. care was not associated with significant increases Driven in part by the expense of programs such in expenses (Sturm 1997; Sturm, Zhang, and as Medicare and Medicaid, the economic burden Schoenbaum 1999). For instance, Goldman, of medical care became a major concern of federal McCulloch, and Sturm (1998) found that although and state governments (Caronna 2004; McKinlay the number of employees using mental health and Marceau 2005; Quadagno 2005). These eco- services at a firm increased after managed behav- nomic concerns prompted health policy that used ioral care was introduced, more costly expenses market mechanisms to contain costs in the 1970s were less likely to be incurred (e.g., inpatient and 1980s (Caronna 2004:50), such as the gradual admission). introduction of health maintenance organizations Given these issues, legislators must balance (Mechanic 2004; Shi and Singh 2001; Tausig, calls for equality against calls for fiscal restraint. Michello, and Subedi 1999). At the same time, ad- In the early 1980s, individual states began consid- vances in mental health treatment and the shift from ering legislation mandating equal coverage for hospital-based to community-based treatment con- mental health care. Although this legislation had tributed to a greater need for mental health service an important impact because it mandated continu- coverage (Joint Commission on Mental Illness and ous mental health care in employer-based health Health 1961; Mechanic and Rochefort 1990; plans, previous federal legislation known as the Tausig et al. 1999). 1974 Employee Retirement Income Security Act Within this context of economic concerns about (ERISA) provided an exemption for large employ- health care reimbursement and broader changes in ers that self-insure. ERISA specifically provides the structure of mental health services, debates self-insuring employers exemptions from state about mental health care reimbursement emerged. health insurance mandates such as mental health A fundamental rationale for mental health parity parity laws (Buchmueller et al. 2007). is the equality it promotes by providing fair access Over the next decade, the increasing prevalence and the option of coverage for mental health of state-level parity legislation, as well as the push

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 156 Society and Mental Health 2(3)

Figure 1. Mental Health Parity State Mandates (2005) Source: National Congress of State Legislatures and Mental Health America. for equitable coverage despite economic concerns, plans that did offer mental health coverage had set the stage for debate about mental health parity at great flexibility to restrict coverage. Given these the national level. This debate began in the midst of exemptions, at least three quarters of employer- President Clinton’s 1993 health care reform pro- sponsored health plans were still placing greater re- posal, as mental health advocates and powerful strictions on mental or behavioral health coverage state actors, notably Senator Pete Domenici and in 1997 (Buck et al. 1999). In essence, the federal the late Senator Paul Wellstone, made the case law functioned as a symbolic expression rather for parity (Amenta and Poulsen 1996; Hausman than a mandate to provide care (Gitterman, 2002). The debate over mental health reform cul- Sturm, and Scheffler 2001; Otten 1998). minated in the first federal mental health law, the Following the national legislation, state-level federal MHPA, albeit in a limited and attenuated mental health parity legislation also expanded in form (U.S. Government Printing Office 1996): the 1990s but faced the same discord over employer exemptions, resulting in considerable variation in In the case of a group health plan (or health state mental health parity laws and the coverage insurance coverage offered in connection they mandate. We focus our analysis on the years with such a plan) that provides both medical 1980 to 2005, examining data from the National and surgical benefits and mental health ben- Conference of State Legislatures (NCSL) as well efits. . . . If the plan or coverage does not as Mental Health America (MHA; formerly known include an aggregate lifetime limit on sub- as the National Mental Health Association). When stantially all medical and surgical benefits, the MHPA was allowed to sunset in 2005, most the plan or coverage may not impose any mental health policy advocates concentrated their aggregate lifetime limit on mental health attention on passing federal legislation, efforts benefits. (110 U.S. Statute 2945) that culminated in passage of the Mental Health Parity and Addiction Equity Act of 2008. This Beyond these stipulations, if plans imposed lim- makes the period from 1980 to 2005 especially its on coverage (e.g., annual limits on benefits) they appropriate for a state-level analysis. Figure 1 were required to set equal limits for medical, surgi- shows the spatial distribution of mental health par- cal, and mental health benefits. Nevertheless, the ity legislation during this period. By 2005, five fine print of the federal MHPA permitted exemp- states had parity laws that applied to treatment for tions for companies both large and small, and those mental health and substance abuse under private

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 Hernandez and Uggen 157

Government ideology Economic pressure • Low conservatism • Low unemployment

Organizational structures Diffusion of law • Mental health association State • Laws in adjacent states/region parity law

Figure 2. Conceptual Framework insurance: Maryland (1994), Minnesota (1995), social expenditures (Amenta 1998:19). Amenta Vermont (1997), Connecticut (1999), and Oregon and Poulsen (1996) used this theory to highlight (2005). In addition to these fully comprehensive the institutional and political conditions that pro- laws, six states had passed comprehensive parity moted social spending in the United States follow- laws (Indiana, Kentucky, Maine, New Mexico, ing the New Deal. For instance, they identified Rhode Island, and Washington) with narrowly voting rights as an important institutional factor defined limitations and exemptions. Twenty-seven that affected passage of Old-Age Assistance pen- states had passed limited parity laws that apply only sions. In a similar manner, they posited that politi- to select populations, such as those with severe cal conditions, including prospending actors, play mental illnesses. an equally crucial role in the passage of such programs. Although institutional politics theory has yet to EXPLAINING STATE VARIATION IN be applied to state mandates such as mental health MENTAL HEALTH PARITY LAWS parity laws, the theory offers an appropriate framework for examining the expansion of mental health A Conceptual Model of Social Policy coverage. In a manner similar to the state-level re- Reform and State Mental Health sponses to the New Deal, states responded to the Parity federal legislation based upon a combination of institutional and political conditions (Amenta We next develop a basic conceptual model that uses 1998:170). Minnesota, for example, was among institutional politics theory (Amenta 1998) to the first states to pass mental health parity legisla- explain state passage of mental health parity laws. tion. During the late 1980s and early 1990s, this Designed to explain sources of government social state exemplified a highly democratized political provision, institutional politics theory aims to system led by a reform-oriented party system resolve differences between institutional and polit- with a history of social spending support (Amenta ical theory (Amenta and Poulsen 1996:34) and and Poulsen 1996:40). In this political atmosphere, delineate how they combine to influence social Minnesotans elected a powerful ally for mental policy (Amenta 1998:19). Amenta (1998:19) health parity legislation, the late Senator contended that political systems that are underde- Wellstone. With the help of Wellstone, a dynamic mocratized or include party systems that are political actor, mental health parity legislation patronage oriented significantly hinder the passage was brought to the national forefront in an example of government social spending policies. of vertical policy diffusion from the state to the Underdemocratized political systems are charac- national level (Shipan and Volden 2006). terized by restrictive democratic processes, such To explain the passage of mental health parity as unfair voting or political assembly rules, and laws among all states, we isolate specific institutional patronage-oriented party systems maintain power and political conditions salient for mental health leg- by rewarding their supporters with individual ben- islation, as outlined in Figure 2. In so doing, our efits. Conversely, reform-oriented regimes, defined framework builds on institutional politics theory but as governments that are aligned with groups and is also based on theoretical perspectives from medical constituencies that support spending, will promote sociology, political sociology, and the sociology of

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 158 Society and Mental Health 2(3) law. From a political perspective, we identify govern- employer-based health insurance (see, e.g., ment ideology, and from an institutional perspective, Cubbins and Parmer 2001). Insuring individuals we identify economic pressures, organizational with mental illnesses is potentially expensive, as structures,anddiffusion of law as important drivers they are more likely to require high-cost, long- of mental health parity legislation. term care (Mechanic 2003:1229–30). At the state level, there are conflicting objectives between the mental health care providers who supply their Political Conditions care and the corporations that insure them. In the simplest terms, providers want their care to be Government ideology. As with other state poli- reimbursed, but insurance companies want to cies, debates about mental illness take place avoid insuring the potentially high-cost risk pool against the backdrop of partisan politics and state composed of individuals with mental illnesses. power (Erikson, Wright, and McIver 1993; These competing objectives clash at the state Quadagno and Meyer 1989; Starr 2011). To level when corporations and associations lobby understand the political conditions that promote individually and collectively against passage of social provision at the state level, Amenta and mental health mandates (Mintz 1995; Mintz and Poulsen (1996:37) drew from a social democratic Palmer 2000; Quadagno 2005). States are likely model, anticipating that redistributive public poli- to be particularly sensitive to corporate and associ- cies would be most likely to pass when left-wing ation resistance during bad economic times political parties are in office. They thus predicted (Imershein, Rond, and Mathis 1992), when firms greater state-level public spending on social provi- can argue that extending benefits could lead to sion under Democratic leadership. job loss and higher unemployment. Therefore, We too hypothesize that mental health parity states facing greater economic strain should be laws were more likely to pass under Democratic less likely to pass mental health parity laws. leadership. Following the era of federal involve- Organizational structures. Our model of social ment, the 1980s ushered in an era of conservative policy also emphasizes the role of larger organiza- leadership that supported the application of market tional structures, including interest groups and mechanisms to contain rising costs (Caronna associations (Amenta, Bonastia, and Caren 2004). Within a decade, however, physicians, pol- 2001; Caronna 2004; Quadagno 2005). Mental iticians, and the public resisted this market-based health advocacy groups have a vested interest in approach to cost containment (Mechanic 2001), re- aiding passage of mental health parity laws. jecting the rationing that had become associated Although it is difficult to establish a causal rela- with the introduction of managed care. By the early tionship, previous evidence shows that voluntary 1990s, the policy environment was primed for associations can indeed influence policy forma- health reform that protected patient rights to equal tion (Skocpol et al. 1993). The National Alliance coverage (Caronna 2004; Quadagno 2005; Wright, on Mental Illness (NAMI), founded in 1979, is Erikson, and McIver 1987). the largest grassroots group dedicated to mental The fragmented U.S. system of government fos- health advocacy, research, and education. NAMI ters a wide spectrum of state government ideolo- advocates for inclusive health policy, such as gies and political environments, offering varying mental health parity, but there has been significant support for mental health parity legislation. We variation in state NAMI representation over recent therefore do not expect that partisan power or affil- decades. We predict that the presence of a state- iation alone will dictate policy formation at the level NAMI association will be positively related state level. Above and beyond party affiliation, to passage of state mental health parity legislation. we anticipate that states with more conservative Aside from the presence of a state-level NAMI government ideologies (Berry et al. 1998; Erikson association, we also anticipate that its stability et al. 1993) and elected officials will be especially and duration—the degree to which the association resistant to mental health parity laws. has become institutionalized—will affect the passage of parity laws. On the other side of the debate for mandated Institutional Conditions mental health coverage, business associations, Economic pressures. Economic factors clearly such as the National Business Group on Health affect the characteristics and distribution of (NBGH; formerly the Washington Business

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 Hernandez and Uggen 159

Group on Health) represent powerful employers together in time; therefore, we expect that states that influence health policy at the national level will pass laws in quick succession. As Grattet et (Mintz 1995). Along with the NBGH, the largest al. (1998:289) hypothesized, the likelihood of national business lobbying group, the U.S. a state adopting legislation is time-specific and Chamber of Commerce, strongly opposed mental contingent on period effects. Finally, we antici- health mandates in the 1990s (Wood 2005). Yet pate regional clustering of parity laws, as states the NBGH and the Chamber of Commerce primar- look to their neighbors for effective models ily represent the interests of firms at the national (Berry and Berry 1990; Grattet et al. 1998). level; the strength of these lobbying groups may More specifically, we hypothesize that states thus bear little relation to state-level mental health will pass mental health parity laws if other states policy. Instead, we focus on the presence of self- in close regional proximity have passed laws. insuring employers in each state that have greater Beyond spatial proximity, regions also provide economic motivation to prevent passage of state a proxy for ‘‘similarsocial and political conditions’’ parity laws. Even though self-insuring firms are (Grattet et al. 1998). Erikson et al. (1998) stipulated exempt from state-level legislation, most opposed that a ‘‘theory of state electoral politics must take any mental health parity legislation during the into account the wide variation in the ideological time period included in our analysis. Thus, orientations of state political parties’’ (p. 731); although this is an indirect measure, we anticipate quite simply, a Democratic state legislator in that the presence of Fortune 100 companies and Mississippi may vote differently than a higher proportion of self-insuring firms in a state a Democratic state legislator in Minnesota. We will have a negative influence on passage of mental consider this regional variation in government health parity legislation. ideology in our diffusion framework, and we Diffusion of law. Neoinstitutional models of hypothesize that states within the same region legal diffusion suggest that legal change occurs will pass similar laws (Behrens, Uggen, and in predictable spatial and historical patterns Manza 2003; Berry and Berry 1990; Eyestone (Edelman 1990; Edelman, Uggen, and Erlanger 1977). 1999; Frank, Camp, and Boutcher 2010; Grattet, On the basis of the four factors described—go- Jenness, and Curry 1998; Jenness 1999; vernment ideology, organizational structures, eco- McCammon et al. 2001). In their analysis of the nomic pressures, and diffusion of law—we draw homogenization of hate crime laws, for example, the following four sets of hypotheses: Grattet et al. (1998) noted that policy makers often have a limited understanding of ‘‘what constitutes Hypothesis 1: States with more conservative optimal policy, especially in terms of acceptability elected officials will be less likely to pass to key constituencies’’ (p. 288). Their policy mental health parity laws. choices are thus contingent on the information Hypothesis 2a: States with mental health and experience they can draw from other policy associations (NAMI), particularly well- making bodies. At the state level, this transfer of established ones, will be more likely to information through social policy networks results pass mental health parity laws. in states mimicking policies adopted in other Hypothesis 2b: States with more Fortune 100 states (Berry and Berry 1990; Eyestone 1977; companies and/or higher percentages of Grattet et al. 1998; Walker 1969). firms that self-insure will be less likely to We anticipate that state mental health parity pass mental health parity laws. laws will spread across time and space in a manner Hypothesis 3: States will be less likely to pass consistent with these models. First, states often fol- mental health parity laws during times of low federal leadership, particularly regarding equal high unemployment than times when unem- opportunity legislation (Edelman 1990; Hill ployment is low. 1986). We hypothesize that states will react to Hypothesis 4a: States will be more likely to pass the elevated importance of mental health parity mental health parity laws following the fed- legislative reform in the mid-1990s, and they eral MHPA of 1996. will be more likely to pass mental health parity Hypothesis 4b: States will be more likely to pass laws shortly after the 1996 federal MHPA. mental health parity laws if other states in Second, changes to state laws generally cluster their region passed parity laws.

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 160 Society and Mental Health 2(3)

Figure 3. Smoothed Hazard Function Estimate for Time to Passage of Any Parity Law

DATA AND METHODS before the federal MHPA of 1996 and a parallel analysis for the period after the federal legislation. Data on state mental health parity laws were coded To test the association between government using information from the NCSL and MHA for ideology and parity legislation, we use a well- laws passed between 1980 and 2005. A table clas- established measure of state government ideology sifying the 2005 mental health parity laws, adapted from Berry et al. (1998:327–28), which is based from materials provided by MHA and the NCSL, is on roll-call voting scores of state congressional del- 1 provided in the Appendix and mapped in Figure 1. egations. These scores are then weighted by the bal- To test our conceptual model (see Figure 2), we ance of partisan strength within each state. We use consider both the simple presence of parity laws the updated government ideology scores created by and the extent to which they provide compre- Berry et al. in their revised 1960 to 2006 govern- hensive coverage. Laws were coded at year of pas- ment ideology series data. For the purposes of sage as ‘‘limited’’ or ‘‘comprehensive’’ based upon this article, we reverse-coded these ideology scores the type of services required or the exceptions also that higher scores indicate more conservative lowed within each health plan. Comprehensive par- state ideologies (on a scale ranging from 1 to ity laws (identified as ‘‘best’’ or ‘‘good’’ by the 100), and we divided the resulting scores by 10 so MHA) apply to all or nearly all treatment of mental that all scores represent a 10-point change in gov- illness and substance abuse covered by private ernment ideology. health plans. Limited parity laws apply to select We operationalize state economic pressures groups only or provide partial protection for those using the state unemployment rate for each year. with mental illnesses. The data are taken from the U.S. Census Bureau’s Between 1980 and 2005, only 11 states passed Statistical Abstract series for 1980 to 2005, provid- comprehensive mental health parity laws, whereas ing a time-varying index of the prevailing eco- 29 states passed limited mental health parity laws. nomic environment and strain present in each For our analysis, we included states passing com- state.3 prehensive parity legislation in the total number We include several measures of state-level of states passing (at least) a limited state mental organizational structures. We used the health parity law. As the Appendix and the hazard Encyclopedia of Associations: Regional, State function in Figure 3 show, most laws were passed and Local Organizations for the years 1996 to 2 in the late 1990s and early 2000s. Overall, 74 per- 2005 to determine whether states had NAMI asso- cent of states (38) had passed at least some form of ciations in each year. States were coded 1 if they parity law by the end of 2005. To test the hypothe- were home to NAMI associations in a given year ses above, we conduct an event history analysis of and 0 otherwise. Annual state-level association factors influencing law passage for the period information was not available prior to 1988, so

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 Hernandez and Uggen 161 we created a dichotomous measure to indicate specified in discrete intervals (years). The depen- when each state NAMI association was established dent variable indicates whether the event (passage for the earlier period between 1980 and 1995.4 As of a mental health parity law) did or did not occur an indicator of the stability, duration, and solvency within a given interval for all units (states) at risk of the NAMI associations during the later period, for the event. The data are analyzed in a ‘‘state- we computed a measure of the years since a state- year’’structure, with the outcome coded 1 for years level association had been established, ranging in which a first parity law was passed and 0 other- from 0 to 22 years. If any state NAMI association wise. Once a state passes such a law, it is removed was not solvent in a particular year, but was estab- from the risk set and contributes no subsequent ob- lished in the following year, the count restarted at 0. servations to the analysis. Because state legislation We also use a measure of the percentage of operates on yearly rather than continuous cycles, ‘‘private-sector establishments that offer health discrete-time methods have been especially well insurance that self-insure at least one plan’’ taken suited to studies of state-level legal change, includ- from the 1993 Medical Expenditure Panel Survey ing the passage of state hate-crime laws (Grattet et (U.S. Department of Health and Human Services al. 1998), felon voting restrictions (Behrens et al. 1996). The survey is not conducted annually, so we 2003), and worker compensation legislation use the earliest available survey data from 1993 (Pavalko 1989). and apply the same proportion of self-insuring firms Discrete-time models require some method of for all of the years in the analysis for the second accounting for duration dependency. The most typ- period (1996–2005). Finally, we include a measure ical specification involves temporal dummy varia- of the number of Fortune 100 companies in the state bles, though other transformations often produce for each year, ranging from 0 to 22 firms per state. better fitting models (Box-Steffensmeier and With regard to the diffusion of law, we estimate Jones 2004). Following Allison (2010), we in- the effects of both region and timing. We use a stan- spected the hazard distribution in Figure 3 and dard Census Bureau four-category region classifi- compared fit statistics for several specifications cation to represent the Northeast, Midwest, South, of time dependency, including individual year and West as separate indicator variables. As a mea- dummy variables, a linear trend, and a quadratic sure of regional diffusion, we include a variable model with both the linear trend and its square. representing the percentage of states that have The best-fitting models for the first period and passed mental health parity laws in each of nine di- for our comprehensive parity models include the visions (subcategories of regions) in each of the year term alone, but the best-fitting model for immediately preceding years between 1980 and the second period includes both year and year- 2005 (see, e.g., Grattet et al. 1998). squared terms. We will first examine the duration State-level analysis is well suited to assessing structure of the data using a nonparametric hazard variation in health policy (e.g., Mintz and Palmer techniques, before turning to parsimonious multi- 2000). Unlike cross-sectional studies, we use an variate models that assess the effects of govern- event history approach to model the time-varying ment ideology, economic strain, associations, factors associated with legal change (Allison and legal diffusion. 2010; Yamaguchi 1991). We estimate the likelihood of mental health parity law passage using a discrete-time logistic regression model: RESULTS In Figure 3, we show the hazard distribution for log½Piy=ð1 PiyÞ5ay1b1Xiy11 ...1bkXiyk: time to passage of any parity law. This plot illus- In this equation, Piy represents the conditional trates how the national picture changes over time, probability of a mental health parity law being as a reflection of both national- and state-level pro- passed in state i during year y, given that a law cesses. The first seven laws were passed in the has not already been passed in that state. 1980s and were limited in scope, and the first com- Explanatory variables are represented as X1, X2,. prehensive law was not passed until 1994. By 1996, ..,Xk, with time-varying covariates denoted by a total of 14 states had passed some form of mental a y subscript. b represents the effect of the indepen- health parity legislation. Many states passed laws in dent variables, and ay represents a set of constants the late 1990s, as indicated by the upward trend in corresponding to each year. In these models, time is the hazard distribution. The peak rate of passage

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 162 Society and Mental Health 2(3) occurs in the early 2000s but declines slightly in passed a law (limited or comprehensive) more 2005. The original federal MHPA of 1996 was al- quickly than they passed comprehensive laws. lowed to sunset in 2005, and mental health policy After a law has been passed, the state contributes advocates had begun to focus their attention on no more observations or state-years to the analysis. passing new federal legislation. Fewer states were thus at risk for passing limited Next, we consider the effects of the independent laws (model 2) relative to comprehensive laws variables on the time until passage of these laws. (models 3a and 3b) during much of the period Descriptive statistics for the time-varying and from 1996 to 2005. On the basis of initial analyses time-invariant independent variables are shown in considering each variable in turn, we selected the Table 1. For all states from 1980 to 2005, the unem- conservative government ideology measure (rather ployment rate ranged from 2 percent to 18 percent, than an indicator of partisan control) to represent with an average rate of about 6 percent. The conser- the effect of government ideology.5 Given that vative ideology measure ranged from 0.21 to 10, the number of events is always less than 50 in with a mean of 5; approximately one third of all a state-level analysis of the time until a first law states had Republican-controlled legislatures, and is passed, we estimate parsimonious models (i.e., about half had Republican governors. During the fewer than 12 covariates). first period, about 70 percent of states had estab- We first summarize multivariate predictors of lished NAMI organizations. Between 1996 and any law passage, including comprehensive laws, 2005, half the states had NAMI organizations, for the pre-MHPA period in model 1 of Table 2. which had been in operation for an average of Odds ratios greater than 1.0 represent positive ef- 2.13 years. The average percentage of private-sec- fects, and odds ratios less than 1.0 represent nega- tor establishments that self-insure at least one plan tive effects. During this period, only our measures was 21 percent, and on average, states had about of diffusion predicted parity legislation: The odds two Fortune 100 companies. The regional means ratio of 1.09 for ‘‘percentage in census division reflect the proportion of state-years contributed with law’’means that the estimated odds of passing by a given region in the analysis, prior to any ad- a law rose by 9 percent for each corresponding per- justments for censoring (e.g., the 9 states in the centage point increase in the rate of passage in northeastern region constitute 0.18 of the total neighboring states. Apart from these findings, how- observations). ever, the model fails to explain passage of parity Inspection of the hazard rate revealed a clear ris- laws during this early period. Our measures of gov- ing and falling pattern for the second period, and ernment ideology, economic pressures, and organi- a comparison of the linear model with the quadratic zational structures did not significantly predict suggested that the squared term is necessary to when the 13 laws passed prior to the MHPA federal appropriately model the duration structure of pas- legislation. sage of any parity law after 1996. For the first For the post-MHPA period, government ideol- period, and analysis of comprehensive legislation ogy, economic pressures, and diffusion all pre- during the second period, a linear specification of dicted law passage, as shown in model 2 of time provides the best fit to the data. All models Table 2. Consistent with our conceptual model, provided a good fit relative to equations specifying states were significantly less likely to pass a mental time as a set of individual year dummy variables. health parity law during times of high unemploy- Although the statistical power to detect effects ment. The odds ratio of 0.53 indicates that the esti- is limited by the small number of states passing mated odds of passage drop by 47 percent with each laws, particularly comprehensive laws, we built percentage point increase in the unemployment parsimonious multivariate models to consider the rate. States characterized by a more conservative effects of the most important predictors on passage government ideology were also significantly less of first mental health parity law. The sample size in likely to pass any law after 1995, although provid- these models is the total number of state-years (e.g., ing a unit-change interpretation of this coefficient 50 states over 16 years in model 1 of Table 2 and 50 is less straightforward because of the metric. Our states over 10 years in models 2, 3a, and 3b) that measures of organizational structures, however, have not yet passed a mental health parity law did not significantly predict law passage, nor did and are therefore at risk for passing such legisla- we find differences by region. As with the pre- tion. The smaller number of cases in model 2 rela- MHPA period, the percentage of states in a division tive to models 3a and 3b indicates that more states that had passed a law remained a significant

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 Table 1. Description of Variables

Variable Description Mean or % Minimum Maximum Expected Effect

Government ideology Conservative ideology a Conservative state legislator ideology 5.04 0.21 10.00 - Republican control Republican control of both legislative houses 0.28 0 1 - Downloaded from Republican governor Republican governor 0.52 0 1 - Economic pressures Unemployment rate State unemployment rate 5.93 2.20 18.00 -

smh.sagepub.com Organizational structures State NAMI association, period 1 NAMI association at state level (1980–1995) 0.69 0 1 1 State NAMI association, period 2b NAMI association at state level (1996–2005) 0.50 0 1 1 State years since NAMI association was establishedb Years since NAMI association established 2.13 0.00 22.00 1 b

atASA-American SociologicalAssociation onNovember13,2012 Fortune 100 companies Number of Fortune 100 companies 1.95 0.00 22.00 - Percentage self-insuringb Percentage of private-sector establishments that offer 21.01 6.70 34.30 - health insurance that self-insure at least one plan Diffusion of law States in region with law Percentage in nine-category census region that have 29.08 0.00 100.00 1 passed law Northeast Connecticut, Maine, Massachusetts, New Hampshire, 0.18 0 1 1 New Jersey, New York, Pennsylvania, Rhode Island, Vermont South Alabama, Arkansas, Delaware, Florida, Georgia, 0.32 0 1 - Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, West Virginia West Alaska, Arizona, California, Colorado, Hawaii, Idaho, 0.26 0 1 - Montana, Nevada, New Mexico, Oregon, Utah, Washington, Wyoming Midwest Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, 0.24 0 1 1 Missouri, Nebraska, North Dakota, Ohio, South Dakota, Wisconsin

Note: NAMI = National Alliance on Mental Illness. 163 a. The conservative ideology measure is a continuous variable, where 0 equals the most liberal and 10 equals the most conservative. The other two government ideology variables are binary indicators, where 1 equals Republican control (of the governorship or the legislature) and 0 equals other. See text for further details. b. Measures included in the analysis of the second period only (1996–2005). 164 Table 2. Multivariate Predictors of First Mental Health Parity Law, before and after the Federal Mental Health Parity Act of 1996

Any Limited, Comprehensive, or Fully Comprehensive Legislation Comprehensive or Fully Comprehensive Legislation Period 1 (1980–1995) Period 2 (1996–2005) Period 2 (1996–2005) Model 1 Model 2 Model 3a Model 3b Variable exp(b) exp(b) exp(b) exp(b) Downloaded from

Government ideology Conservative ideology 0.85 0.80* 0.63** 0.57** Economic pressures smh.sagepub.com State unemployment rate 1.08 0.53** 0.67 0.61 Organizational structures State NAMI organization present 0.70 — 0.74 — State NAMI organization stability — 0.94 — 1.34** atASA-American SociologicalAssociation onNovember13,2012 Fortune 100 companies — 0.99 0.84 0.67 Percentage self-insuring — 0.96 1.04 1.07 Diffusion of law Percentage in region with law 1.09*** 1.05*** 1.01 1.00 Northeast (vs. Midwest) 1.68 0.63 2.01 1.25 South (vs. Midwest) 0.88 1.18 0.34 0.27 West (vs. Midwest) 0.15* 1.95 2.62 2.97 Year 0.99 1.27 1.01 1.02 Year2 — 0.99 — — Constant 0.01** 0.26 0.19 0.19 x2 28.03*** 27.39*** 11.14 17.16* –2 log likelihood (df) 102.07 (8) 128.22 (11) 60.40 (10) 54.38 (10) Events 13 24 7 7 Number of cases 719 238 430 430

Note: NAMI = National Alliance on Mental Illness. The U.S. Census Bureau divides the four regions into nine divisions, which are used for the measures of diffusion of law. Models 1 and 2 predict the passage of any parity legislation, including limited or comprehensive laws. Models 3a and 3b predict the passage of comprehensive legislation only. The measure of NAMI organization stability indicates the number of years the state NAMI organization has been present. See text for further details. *p \ .10, **p \ .05, and ***p \ .01 (two tailed). Hernandez and Uggen 165 predictor of state law passage. In short, diffusion of basic hypotheses in our conceptual model of mental law, economic pressure, and conservative govern- health social policy: Parity laws tended to pass in ment ideology emerge as statistically significant states characterized by less conservative ideology, predictors of passage of mental health parity laws low unemployment, long-standing advocacy or- after the federal MHPA of 1996. ganizations, and geographical proximity to states In the final two models of Table 2, we show that had established parity. Yet the relative salience multivariate predictors of comprehensive law pas- of these factors differed depending on the period sage (see models 3a and 3b). Only seven states and type of law. had passed comprehensive laws, so the models Differentiating between states that passed any become unstable when too many covariates are parity laws versus those that passed a more compre- considered in the same equation. In contrast to hensive mandate, as well as the period in which the ‘‘any parity’’ analysis, we fail to detect a they were passed, reveals subtle but important dis- significant unemployment effect on passage of tinctions. Prior to the passage of the federal legisla- a comprehensive law. Rather, our measure of gov- tion, the diffusion of laws within census divisions is ernment ideology emerges as a powerful predictor the only significant predictor of any parity laws. of comprehensive laws: States led by politicians This measure continues to influence legislation fol- with more conservative government ideologies lowing the federal MHPA of 1996, but unemploy- have been significantly less likely to pass these ment rates, government ideology among state laws. In model 3b, our measure of the stability legislators, and the presence of a mental health anddurationofNAMIassociationsemergedas advocacy organization also emerged as significant a significant predictor, with the estimated odds predictors in the later period. of passing a comprehensive law rising by 34 per- During the second period, lower unemployment cent with each year since a state NAMI associa- rates were significantly associated with the passage tion was first established. Neither Fortune 100 of any type of parity legislation. From a practical companies nor the proportion of employers self- perspective, the negative effect of economic pres- insuring was associated with passage of compre- sure makes sense: States are less likely to prioritize hensive legislation. Consistent with the geo- mental health parity when there are pressing eco- graphic scattering of comprehensive parity states nomic concerns on the public agenda. Corporate shown in Figure 1, we did not detect significant and association lobbying against state health man- regional effects in these models. Net of our other dates may also increase during times of economic measures, states with more conservative legisla- strain, as business groups actively resist govern- tors were significantly less likely to pass compre- ment mandates. For example, a U.S. Government hensive laws, but those with well-established Accountability Office (2000) survey indicated NAMI associations were significantly more likely that employers made benefit changes after the fed- to pass them. eral MHPA as part of their cost containment efforts. Nevertheless, neither the number of Fortune 100 companies nor the percentage of establishments DISCUSSION that self-insure is a statistically significant predictor of parity legislation. Over the past 50 years, there have been remarkable Although states were significantly less likely to changes in how mental illnesses are treated in the pass any parity laws during times of high unem- United States. Building from theories of institu- ployment, this economic pressure did not influence tional politics (Amenta 1998; Amenta and the likelihood they would pass more comprehen- Poulsen 1996), our aim in this study has been to sive legislation (nor did our regional measures of understand and model the conditions that contrib- legal diffusion). Conservative government ideol- uted to state mental health parity legislation. We ogy and stable advocacy organizations emerged hypothesized that a spectrum of political and insti- as the most salient predictors of comprehensive tutional forces would influence state passage of parity laws. The importance of government ideol- these legislative mandates. Using discrete-time ogy for passage of these more stringent state parity logistic regression event history models, we esti- laws supports our prediction that legislation was mated the effects of government ideology, organi- more likely to pass when left-wing political parties zational structures, economic pressure, and were in power. Although conservative ideology diffusion of law. Our results generally support the shows a marginal negative association with

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 166 Society and Mental Health 2(3) passage of general laws (limited or comprehensive) of parity laws. At the beginning of the 1990s the in the later period (p\.10), it emerges as a power- health care policy environment was primed for ful and significant predictor of laws requiring more legislation that protected patients’ rights comprehensive coverage of mental illness treat- (Quadagno 2005), and regional policy diffusion ment. Associations, such as NAMI, provided addi- appears to be important in this early period. tional momentum for the passage of state-level Once mental health parity emerged more promi- comprehensive legislation. Over the years, though, nently as a national issue (culminating in the these state-level NAMI associations were not MHPA of 1996), partisan political pressures ap- always solvent, and the degree to which these peared to play a bigger role, as indicated by the were stable or well established—essentially the strong effect of conservative government ideol- degree of institutionalization of each NAMI asso- ogy during the more recent period. ciation—predicted passage of comprehensive par- Although not included in our analysis, we ity legislation. On the other hand, neither the would be remiss to neglect discussion of several number of Fortune 100 companies nor the percent- key political actors, representing a variety of polit- age of self-insuring companies was associated with ical ideologies, who paved the way for national par- the passage of any type of law. Our inability to ity legislation (Amenta and Poulsen 1996). detect such effects may be due to the indirect nature Although our account emphasizes structural forces, of these measures; we lack a direct indicator of the individual legislators also cite more personal fac- strength of business association lobbying against tors. The federal MHPA of 1996 was successfully parity legislation. brought to the forefront of health policy reform Aside from economic pressures, government by Senator Domenici and Senator Wellstone. ideology, and organizational structures, the spatial Upon Senator Wellstone’s untimely death, other distribution of the basic parity laws is at least par- powerful political figures took up the cause to tially consistent with institutional accounts of legal pass federal legislation in the 2000s, including the diffusion (Behrens et al. 2003; Edelman 1990; late Senator Edward Kennedy and his son, Patrick Frank et al. 2010; Grattet et al. 1998). We found Kennedy. The serious mental illness of close family evidence of regionalization when we considered members appeared to motivate several of these the percentage of states in a division that had passed legislators to champion the cause of mental health at least a limited parity law in the preceding year. parity at the federal level, regardless of their polit- Although the region indicators are not statistically ical ideologies (Barry, Huskamp, and Goldman significant in multivariate models predicting com- 2010; McSween 2002). prehensive parity legislation, northeastern states Our overarching goal has been to advance sub- such as Connecticut and Vermont emerged as insti- stantive knowledge on the important question of tutional leaders in mental health care reform in the mental health parity and to situate such knowledge 1990s (the midwestern state of Minnesota is a nota- within a broader explanation of policy change. ble exception to this regional pattern). The plot of There are some important limitations of the current our hazard function also affirmed our hypothesis analysis, including the relatively small number of that states would be more likely to pass mental states passing laws and the challenge of measuring health parity laws, limited or comprehensive, fol- institutional and political processes in a time-vary- lowing the federal MHPA of 1996. Although we ing structural-level analysis. Although we have cannot definitively determine the cause of this taken care to include covariates related to both peak in state laws during the late 1990s and early our state-level predictors and passage of parity 2000s, it may reflect a state-level effort to fill laws, omitted variable bias remains a concern in gaps in the federal legislation. Until recently, how- an observational study such as this one. For exam- ever, no state-level legislation was capable of over- ple, subsequent research would benefit from further riding the ERISA exemption for firms that self- consideration of factors such as health maintenance insure; thus, mental health advocates moved to organization market penetration, a variable that pass additional federal legislation later in the was not available to us for individual states prior 2000s. to 2002, or annual state-level data indexing the Taken together, these effects reinforce our stigma associated with mental illness. Despite hypotheses, and the model proposed by Amenta these limitations and potential omissions, we (1998) and Amenta and Poulsen (1996), that both observe a consistent pattern of variation in mental political and institutional factors influence passage health parity legislation from 1996 to 2005. States

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 Hernandez and Uggen 167 characterized by less conservative government thus ensuring more comprehensive coverage for ideologies, low unemployment, and neighboring those with a mental illness. states with parity laws were likely to pass general Despite the recent focus on federal legislation, (limited or comprehensive) parity provisions. The state parity law continues to evolve. A 2009 most comprehensive laws requiring insurers to NAMI survey suggests both continued refinement extend comparable benefits for mental and physical and substantive changes to state mental health par- health were passed in states with less conservative ity laws in Colorado (2007), Idaho (2006), North ideology and well-established mental health advo- Carolina (2007), Ohio (2006), Oregon (2006), cacy organizations. Washington (2007), and West Virginia (2007). In response to the current patchwork of state and federal law, Tovino (2011) proposed amendments that CONCLUSIONS would conform the state laws to minimum federal requirements and expand the reach of state mental Although our analysis focuses on state legislation, health parity law to encompass all health plans sub- our model suggests that shifting government ideol- ject to state insurance regulation. ogy and partisan power may help explain the devel- These recent changes and proposals should not opment of federal parity legislation between 1980 obscure the contested history of state parity legisla- and 2005. On December 31, 2005, Congress allowed tion prior to the 2008 federal law. Declared the the federal MHPA of 1996 to sunset, with hopes of ‘‘decade of the brain’’ by presidential proclamation new federal mental health parity legislation on the (Bush 1990), the 1990s brought about substantial horizon. During subsequent years, legislators in debate over health care reform and coverage of both the U.S. House of Representatives and Senate treatment for mental illnesses (Wolff 2002:791). proposed new versions of federal mental health par- Despite such attention, though, the decade did not ity legislation. Led by Senator Domenici, a powerful bring about a level of equality in mental health Republican, Congress eventually found bipartisan reimbursement sufficient to ‘‘symbolically des- support for the Paul Wellstone and Pete Domenici tigmatize’’ mental illness (Mechanic 2003). The Mental Health Parity and Addiction Equity Act of most recent federal legislation represents a signifi- 2008, which took effect for health plan years begin- cant step toward providing parity, and it suggests an ning on or after July 1, 2010. The law was passed emerging national consensus about the need for immediately prior to the election of President equal mental and physical health coverage. If, as Barack Obama and preceded the comprehensive Mechanic (2003) argued, mental health parity Health Care and Education Reconciliation Act of laws symbolically undermine the stigma associated 2010, signed into law on March 23, 2010 (Light with mental illness, the issue of reimbursement for 2011).6 Unlike the previous federal law, this most mental health services is significant for its sym- recent mental health parity legislation does not allow bolic declarative effects as well as its tangible pub- for ERISA exemptions for firms that self-insure, lic health policy impacts.

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 168 Society and Mental Health 2(3)

Appendix Mental Health America Categorization of Parity Laws (1980–2005)

Fully Comprehensive Comprehensive Limited No Parity or Parity Lawsa Parity Lawsb Parity Lawsc Mandate Laws

Connecticut (1999) Indiana (1999/2001/2003) Arizona (1997/2001) Alabama Maryland (1994) Kentucky (2000)d Arkansas (1997/2001) Alaska Minnesota (1995) Maine (1995/2004) California (1974/1999) Florida Vermont (1997) New Mexico (2000) Colorado (1997) Georgia Oregon (2005) Rhode Island (1994/2001) Delaware (1998/2001) Idaho Washington (2005)d Hawaii (1988/2004) Kansas Illinois (2001) Mississippi Iowa (2005) New York Kentucky (1986) North Dakota Louisiana (1982/1999) Pennsylvania Massachusetts (2000) Wisconsin Michigan (1988) Wyoming Missouri (2004) Montana (1999) Nebraska (1999) Nevada (1999) New Hampshire (1994/2002) New Jersey (1999/2002) North Carolina (1991/1997) Ohio (1985) Oklahoma (1999) South Carolina (2000/2005) South Dakota (1998) Tennessee (1998) Texas (1991/1997) Utah (2000) Virginia (2004) Washington (1987) West Virginia (2004) a. Parity applies to all mental health and substance abuse disorders under private insurance plans. No exemptions. b. Not quite comprehensive parity because of certain exemptions and/or limitations. c. Parity applies only to select groups, such as those with severe mental illnesses or state and local employees, or protects against only certain types of discrimination. d. State passed an earlier limited parity law and a later comprehensive parity law.

ACKNOWLEDGMENTS NOTES We would like to thank Cynthia Goff for her helpful in- 1. For the purposes of this article, we consider both sights about health care policy, as well as Kim Gardner, ‘‘fully comprehensive’’ and ‘‘comprehensive’’ laws Andrew Halpern-Manners, Wesley Longhofer, Karen to be comprehensive mental health parity laws. The Lutfey, Donna McAlpine, Heather McLaughlin, John NCSL adopts a more complex categorization scheme Robert Warren, and Suzy McElrath for helpful comments but is largely consistent with the MHA classification. and other assistance. Support for this research was pro- Our analysis is based on the more focused and intui- vided by a predoctoral National Research Service tive MHA categorization for the period from 1980 to Award from the National Institute of Mental Health 2005. Although the grouping strategy was slightly dif- (T23-MH19893) to Dr. Hernandez and a Robert Wood ferent between the MHA and the NCSL, auxiliary Johnson Foundation Investigator Award in Health analysis revealed that the results remained sub- Policy Research to Dr. Uggen. stantively the same. Two states, Kentucky and Washington, passed earlier limited parity laws and

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 Hernandez and Uggen 169

later comprehensive laws and are included in both sets Perspective: Concepts, Images, Arguments, and of analyses. Research Strategies.’’ Annual Review of Sociology 2. All state-level mental health parity legislation was 27:213–34. passed after 1980, except for an earlier, limited law Amenta, Edwin and Jane D. Poulsen. 1996. ‘‘Social in California, passed in 1974. Thus, California is left Politics in Context: The Institutional Politics Theory censored and hence excluded from analysis of limited and Social Spending at the End of the New Deal.’’ parity laws, but it contributes observations to the com- Social Forces 75:33–61. prehensive law analysis. When we include California Barry, Colleen L., Haiden A. Huskamp, and Howard H. in analysis of the period from 1970 to 2005, we find Goldman. 2010. ‘‘A Political History of Federal identical results, though we lack information on Mental Health and Addiction Insurance Parity.’’ some covariates between 1970 and 1980. Milbank Quarterly 88:404–33. 3. We also considered a measure of logged gross domes- Behrens, Angela, Christopher Uggen, and Jeff Manza. tic product by state (formerly gross state product) from 2003. ‘‘Ballot Manipulation and the ‘Menace of the Census Bureau for each state in each year. Negro Domination’: Racial Threat and Felon However, the Bureau of Economic Analysis strongly Disenfranchisement in the United States, 1850–2002.’’ cautions against using these data for time-series anal- American Journal of Sociology 109:559–605. ysis between 1963 and 2006 because of changes in Berry, Frances Stokes and William D. Berry. 1990. ‘‘State coding over this period. Lottery Adoptions as Policy Innovations: An Event 4. The NAMI measure for the first period indicated when History Analysis.’’ American Political Science each association began but does not account for vari- Review 84:395–415. ation in association presence after it was established Berry, William D., Evan J. Ringquist, Richard C. Fording, (e.g., because of insolvency of state associations). and Russell L. Hanson. 1998. ‘‘MeasuringCitizen and Using the Encyclopedia of Associations, the NAMI Government Ideology in the American States, measure for the second period does account for such 1960–93.’’ American Journal of Political Science variation. 42:327–48. 5. In our bivariate models (not shown), our measure of Box-Steffensmeier, Janet M. and Bradford S. Jones. 2004. conservative government ideology significantly Event History Modeling: A Guide for Social reduced the likelihood of passing any parity laws Scientists. Analytical Methods for Social Research and comprehensive parity laws separately, but other Series. Cambridge, UK: Cambridge University Press. measures of conservative ideology did not predict Buchmueller, Thomas C., Philip F. Cooper, Mireille law passage. State presence of a NAMI association Jacobson, and Samuel H. Zuvekas. 2007. ‘‘Parity for did not predict passage of laws, but the years since Whom? Exemptions and the Extent of State Mental the NAMI association had been established did predict Health Parity Legislation.’’ Health Affairs 26: the passage of a comprehensive parity law (during the w483–87. second period only). Our measures of the percentage Buck, Jeffrey A., Judith L. Teich, Beth Umland, and of self-insuring firms and the number of Fortune 100 Mitchell Stein. 1999. ‘‘Behavioral Health Benefits in firms in a state were not associated with law passage. Employer-sponsored Health Plans, 1997.’’ Health Although we did not find regional differences overall, Affairs 18:67–78. we found that states were significantly more likely to Bush, George. 1990. ‘‘PresidentialProclamation 6158. U. pass a parity law if other states in their division had S. Senate and U.S. House of Representatives Joint passed a law. Resolution to Designate the Decade Beginning 6. As of this writing, national health care reform remains January 1, 1990 as the ‘Decade of the Brain.’’’ the subject of fierce political debate (Jacobs and House Joint Resolution 174/Senate Joint Resolution Skocpol 2010; Starr 2011). Consistent with our con- 173, Public Law 101-58. ceptual model and findings, Republican leadership is Caronna, Carol A. 2004. ‘‘The Misalignment of actively mobilizing against the Obama health care Institutional ‘Pillars’: Consequences for the U.S. initiatives. Health Care Field.’’ Journal of Health and Social Behavior 45:45–58. Cubbins, Lisa A. and Penelope Parmer. 2002. ‘‘Economic REFERENCES Change and Health Benefits: Structural Trends in Employer-based Health Insurance.’’ Journal of Ò Allison, Paul D. 2010. Survival Analysis Using SAS :A Health and Social Behavior 42:45–63. Practical Guide. 2nd ed. Cary, NC: SAS Institute. Edelman, Lauren, Christopher Uggen, and Howard Amenta, Edwin. 1998. Bold Relief: Institutional Politics Erlanger. 1999. ‘‘The Endogeneity of Legal and the Origins of Modern American Social Policy. Regulation: Grievance Procedures as Rational Myth.’’ Princeton, NJ: Princeton University Press. American Journal of Sociology 105:406–54. Amenta, Edwin, Chris Bonastia, and Neal Caren. 2001. Edelman, Lauren B. 1990. ‘‘Legal Environments and ‘‘U.S. Social Policy in Comparative and Historical Organizational Governance: The Expansion of Due

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 170 Society and Mental Health 2(3)

Process in the American Workplace.’’ American Federal Hate Crime Law in the U.S. 1985–1998.’’ Journal of Sociology 95:1401–40. Social Problems 46:548–71. Erikson, Robert S., Gerald C. Wright, and John P. McIver. Joint Commission on Mental Illness and Health. 1961. 1993. Statehouse Democracy. Cambridge, UK: Action for Mental Health. New York: Basic Books. Cambridge University Press. Kershaw, Sarah. 2008. ‘‘The Murky Politics of Eyestone, Robert. 1977. ‘‘Confusion, Diffusion, and Mind-body.’’ The New York Times, March 30, Innovation.’’ American Political Science Review 71: Section WK, p. 1. 441–47. Light, Donald W. 2011. ‘‘Historical and Comparative Frank, David John, Bayliss J. Camp, and Steven A. Reflections on the U.S. National Health Insurance Boutcher. 2010. ‘‘Worldwide Trends in the Criminal Reforms.’’ Social Science and Medicine 72:129–32. Regulation of Sex, 1945 to 2005.’’ American Link, Bruce G., Jo C. Phelan, Michaeline Bresnahan, Ann Sociological Review 75:867–93. Stueve, and Bernice Pescosolido. 1999. ‘‘Public Frank, Richard G., Chris Koyanagi, and Thomas G. Conceptions of Mental Illness: Labels, Causes, McGuire. 1997. ‘‘The Politics and Economics of Dangerousness, and Social Distance.’’ American Mental Health ’Parity’ Laws.’’ Health Affairs 16(4): Journal of Public Health 89:1328–33. 108–119. Martin, Jack K., Bernice A. Pescosolido, and Steven A. Frank, Richard G. and Thomas G. McGuire. 2005. Tuch. 2000. ‘‘Of Fear and Loathing: The Role of ‘‘Integrating People with Mental Illness into Health ‘Disturbing Behavior,’ Labels, and Causal Insurance and Social Services.’’ Pp. 223–37 in Attributions in Shaping Public Attitudes Toward Policy Challenges in Modern Health Care, edited People with Mental Illness.’’ Journal of Health and by David Mechanic, Lynn B. Rogut, David C. Social Behavior 41:208–23. Cohen, and James R. Knickman. New Brunswick, McCammon, Holly J., Karen E. Campbell, Ellen M. NJ: Rutgers University Press. Granberg, and Christine Mowery. 2001. ‘‘How Gitterman, Daniel P., Roland Sturm, and Richard M. Movements Win: Gendered Opportunity Structures Scheffler. 2001. ‘‘Toward Full Mental Health Policy and U.S. Woman’s Suffrage Movements, 1866 to and Beyond.’’ Health Affairs 20:68–76. 1919.’’ American Sociological Review 66:49–70. Goffman, Erving. 1963. Stigma: Notes on the McKinlay, John B. and Lisa D. Marceau. 2005. ‘‘TheEnd Management of Spoiled Identity. Englewood Cliffs, of the Golden Age of Doctoring.’’Pp. 189–214 in The NJ: Prentice Hall. Sociology of Health and Illness: Critical Goldman, William, Joyce McCulloch, and Roland Sturm. Perspectives, 7th ed., edited by Peter Conrad. New 1998. ‘‘Costs and Use of Mental Health Services York: Worth. before and after Managed Care.’’ Health Affairs McSween, Jean L. 2002. ‘‘The Role of Group Interest, 17(2):40–52. Identity, and Stigma in Determining Mental Health Grattet, Ryken, Valerie Jenness, and Theodore R. Curry. Policy Preferences.’’ Journal of Health Politics, 1998. ‘‘The Homogenization and Differentiation of Policy and Law 27:773–800. Hate Crime Law in the United States, 1978 to 1995: Mechanic, David. 2001. ‘‘The Managed Care Backlash: Innovation and Diffusion in the Criminalization of Perceptions and Rhetoric in Health Care Policy and Bigotry.’’ American Sociological Review 63: the Potential for Health Care Reform.’’ Millbank 286–307. Quarterly 79:35–54. Hausman, Ken. 2002. ‘‘Plane Crash Takes Life of Mechanic, David. 2003. ‘‘Policy Challenges in Improving Congressional Ally.’’ Psychiatric News 37:1. Mental Health Services: Some Lessons from the Past.’’ Hill, Herbert. 1986. Black Labor and the American Legal Psychiatric Services 54:1227–32. System. Madison: University of Wisconsin Press. Mechanic, David. 2004. ‘‘The Rise and Fall of Managed Imershein, Allen, Philip Rond, and Mary Mathis. 1992. Care.’’ Journal of Health and Social Behavior 45: ‘‘Restructuring Patterns of Elite Dominance and the 76–86. Formation of State Policy in Health Care.’’ Mechanic, David and Donna D. McAlpine. 1999. American Journal of Sociology 97:970–93. ‘‘Mission Unfulfilled: Potholes on the Road to Jacobs, Lawrence R. and Theda Skocpol. 2010. Health Mental Health Parity.’’ Health Affairs 18:7–21. Care Reform and American Politics: What Everyone Mechanic, David and David A. Rochefort. 1990. Needs to Know. New York: Oxford University Press. ‘‘Deinstitutionalization: An Appraisal of Reform.’’ Jenkins, J. Craig and Teri Shumate. 1985. ‘‘Cowboy Annual Review of Sociology 16:301–27. Capitalists and the Rise of the ‘New Right’: An Mintz, Beth. 1995. ‘‘BusinessParticipation in Health Care Analysis of Contributors to Conservative Policy Policy Reform: Factors Contributing to Collective Formation Organizations.’’ Social Problems 33: Action within the Business Community.’’ Social 130–45. Problems 42:408–28. Jenness, Valerie. 1999. ‘‘Managing Difference and Mintz, Beth and Donald Palmer. 2000. ‘‘Business and Making Legislation: Social Movements and the Health Care Policy Reform in the 1980s: The 50 Racialization, Sexualization, and Gendering of States.’’ Social Problems 47:327–59.

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012 Hernandez and Uggen 171

Newhouse, Joseph P. 1993. Free for All? Lessons from the Abuse Benefits under Managed Care?’’ Journal of RAND Health Insurance Experiment. Cambridge, Behavioral Health Services and Research 26: MA: Harvard University Press. 203–10. Otten, Alan L. 1998. Mental Health Parity: What Can It Tausig, Mark, Janet Michello, and Sree Subedi. 1999. A Accomplish in a Market Dominated by Managed Sociology of Mental Illness. Upper Saddle River, Care? New York: Milbank Memorial Fund. NJ: Prentice Hall. Pavalko, Eliza K. 1989. ‘‘State Timing of Policy Tovino, Stacey A. 2011. ‘‘ReformingState Mental Health Adoption: The Case of Workmen’s Compensation Parity Law.’’ Houston Journal of Health Law & in the United States, 1909–1929.’’ American Journal Policy 11:455–504. of Sociology 95:592–615. Trice, Harrison M. and Paul M. Roman. 1971. ‘‘Delabeling, Quadagno, Jill. 2005. One Nation Uninsured: Why the U. Relabeling, and Alcoholics Anonymous.’’ Social S. Has No National Health Insurance. New York: Problems 17:538–46. Oxford University Press. U.S. Department of Health and Human Services. 1996. Quadagno, Jill and Madonna Harrington Meyer. 1989. ‘‘Medical Expenditure Panel Survey. Tables of ‘‘Organized Labor, State Structures, and Social Private-sector Data by Firm Size and State.’’ Policy Development: A Case Study of Old Age Washington, DC: U.S. Government Printing Office. Assistance in Ohio.’’ Social Problems 36:181–96. U.S. Government Accountability Office. 2000. ‘‘Mental Scheff, Thomas J. 1964. ‘‘The Societal Reaction to Health Parity Act: Despite New Federal Standards, Deviance: Ascriptive Elements in the Psychiatric Mental Health Benefits Remain Limited.’’ GAO Screening of Mental Patients in a Midwestern State.’’ Report HEHS-00-95. Washington, DC: U.S. Social Problems 11:401–13. Government Accountability Office. Shi, Leiyu and Douglas A. Singh. 2001. Delivering U.S. Government Printing Office. 1996. ‘‘Departmentsof Health Care in America. 2nd ed. Gaithersburg, MD: Veterans Affairs and Housing and Urban Develop- Aspen. ment, and Independent Agencies Appropriations Shipan, Charles R. and Craig Volden. 2006. ‘‘Bottom-up Act, 1997.’’ Public Law 104–204. Retrieved Federalism: The Diffusion of Antismoking Policies November 11, 2010 (http://www.gpo.gov/fdsys/pkg/ from U.S. Cities to States.’’ American Journal of PLAW-104publ204/pdf/PLAW-104publ204.pdf). Political Science 50:825–43. Walker, Jack L. 1969. ‘‘The Diffusion of Innovations Skocpol, Theda, Marjorie Abend-Wein, Christopher among the American States.’’ American Political Howard, and Susan Goodrich Lehmann. 1993. Science Review 63:880–99. ‘‘Women’s Associations and the Enactment of Wood, Elijah Patton. 2005. ‘‘Children and Mental Health Mothers’ Pensions in the United States.’’ American Parity.’’Policy Matters: Critical Issues in Health Care Political Science Review 87:686–701. Access. Washington, DC: National Conference of Starr, Paul. 2011. Remedy and Reaction: The Peculiar State Legislatures. American Struggle over Health Care Reform. New Wolff, Nancy. 2002. ‘‘Risk,Response, and Mental Health Haven, CT: Yale. Policy: Learning from the Experience of the United Sturm, Roland. 1997. ‘‘How Expensive is Unlimited Kingdom.’’ Journal of Health Politics, Policy and Mental Health Care Coverage under Managed Law 27:801–32. Care?’’ JAMA 278:1522–37. Wright, Gerald C., Robert S. Erikson, and John P. McIver. Sturm, Roland and Rosalie Liccardo Pacula. 1999. ‘‘State 1987. ‘‘Public Opinion and Policy Liberalism in the Mental Health Parity Laws: Cause or Consequence of United States.’’ American Journal of Political Differences in Use?’’ Health Affairs 18:182–92. Science 31:980–1001. Sturm, Roland, Weiying Zhang, and Michael Schoenbaum. Yamaguchi, Kazuo. 1991. Event History Analysis. 1999. ‘‘How Expensive Are Unlimited Substance Newbury Park, CA: Sage.

Downloaded from smh.sagepub.com at ASA - American Sociological Association on November 13, 2012