Making the National Local: Specifying the Conditions for National Government Influence on State...

27
State Politics and Policy Quarterly, Vol. 4, No. 3 (Fall 2004): pp. 318–344 Making the National Local: Specifying the Conditions for National Government Influence on State Policymaking Mahalley D. Allen, University of Kansas Carrie Pettus, University of Kansas Donald P. Haider-Markel, University of Kansas abstract The national government can force or entice state governments to act on policy through a variety of actions, including providing monetary incentives and sanctions. We examine how and under what conditions actions of the national government influence the diffusion of policy across the states. We test our hypotheses on the cases of the diffusion of partial birth abortion laws, truth-in-sentencing laws, and hate crime laws using event history analysis on pooled cross-sectional data from the 50 states. Our results suggest that, in addition to fiscal incentives, the national govern- ment can influence state policymaking when it sends strong, clear signals to the states concerning its preferences and the potential for future action. But even national-level signals that are weak and ambiguous may influence state policymaking indirectly. National policymakers with strong domestic policy agendas can be frustrated by the constitutional structure of government in the United States. Although the framers of the United States Constitution established a com- promise system that had both federal and national characteristics, just how much influence the national government had over states was left to evolve (Krane 1993; Madison [1778] 1999). Outside of foreign policy, national de- fense, and interstate commerce, the relationship between national govern- ment activity and state government activity was left to practice and the in- terpretation of the United States Supreme Court. So today, after over 200 years of practice and interpretation, to what extent can the national govern- ment influence state policymaking, and under what conditions is such in- fluence most likely? Researchers have long examined and debated some of the top-down and bottom-up forces in our federalist system (Dubnick and Gitelson 1981; Eye-

Transcript of Making the National Local: Specifying the Conditions for National Government Influence on State...

State Politics and Policy Quarterly, Vol. 4, No. 3 (Fall 2004): pp. 318–344

Making the National Local: Specifying theConditions for National Government Influenceon State Policymaking

Mahalley D. Allen, University of KansasCarrie Pettus, University of KansasDonald P. Haider-Markel, University of Kansas

abstract

The national government can force or entice state governments to act on policythrough a variety of actions, including providing monetary incentives and sanctions.We examine how and under what conditions actions of the national governmentinfluence the diffusion of policy across the states. We test our hypotheses on the casesof the diffusion of partial birth abortion laws, truth-in-sentencing laws, and hatecrime laws using event history analysis on pooled cross-sectional data from the 50states. Our results suggest that, in addition to fiscal incentives, the national govern-ment can influence state policymaking when it sends strong, clear signals to the statesconcerning its preferences and the potential for future action. But even national-levelsignals that are weak and ambiguous may influence state policymaking indirectly.

National policymakers with strong domestic policy agendas can befrustrated by the constitutional structure of government in the United States.Although the framers of the United States Constitution established a com-promise system that had both federal and national characteristics, just howmuch influence the national government had over states was left to evolve(Krane 1993; Madison [1778] 1999). Outside of foreign policy, national de-fense, and interstate commerce, the relationship between national govern-ment activity and state government activity was left to practice and the in-terpretation of the United States Supreme Court. So today, after over 200years of practice and interpretation, to what extent can the national govern-ment influence state policymaking, and under what conditions is such in-fluence most likely?

Researchers have long examined and debated some of the top-down andbottom-up forces in our federalist system (Dubnick and Gitelson 1981; Eye-

fall 2004 / state politics and policy quarterly 319

stone 1977; Gray 1973, 1994; Hamilton and Wells 1990; Kettl 1983; Lowry1992), and a variety of specific influences have been found. For example,Welch and Thompson (1980) showed that federal financial incentives in-fluence the rate at which policy innovations diffuse across the states; Soss etal. (2001) discovered that changes in national welfare policy influenced statewelfare policy; Mossberger (1999) found that state experimentation withenterprise and empowerment zones influenced national policymakers; andHedge (1983) demonstrated that a state’s reliance on federal aid influencesits budgetary process. Scholars have also attempted more general explana-tions of federalism in policymaking. For example, Lowry (1992) includedvertical and horizontal dimensions in his framework for understanding en-vironmental regulatory policymaking in the states, and Mooney (2000) de-veloped a similar framework for morality policymaking. Scholars have alsoexamined federalism’s influence on policy implementation (Goggin et al.1990; Grogan 1999), including its influence on bureaucratic responsivenessto democratic institutions (Gerber and Teske 2000; Wood 1992).

Clearly, national policymakers have domestic agendas and, in this feder-alist system, they might use incentives in an effort to make states follow thoseagendas. Most federalism scholarship has focused on fiscal federalism as themechanism for vertical diffusion, that is, the means by which national poli-cymakers influence the states. But are there other ways in which the actionsof the national government can influence state policymaking?

We address this question by exploring the nature of vertical policy dif-fusion in the context of other internal and external influences on state pol-icy adoption (Berry and Berry 1990). To test our hypotheses, we examine onecase of traditional fiscal federalism (truth-in-sentencing laws) and two cas-es where the influence of federalism is perhaps less direct and obvious (par-tial birth abortion and hate crime). For each case, we use event history anal-ysis to analyze pooled cross-sectional data from the American states. Ourresults suggest that national influence on state policymaking occurs when thenational government sends strong, clear signals to the states concerning re-wards, punishments, and the likelihood of future national government ac-tions. Nonetheless, national forces are just one among many forces that in-fluence state policymaking.

federalism and the vertical diffusion ofpolicy innovations

The question of how national government activity might influence statepolicy must begin with the notion of incentives. If national officials can pro-

320 allen, pettus, and haider-markel

vide appropriate incentives for state officials to act, then, all else being equal,state officials should be more likely to do so. In this respect, fiscal federalismis perhaps the most direct and obvious national government activity aimedat influencing state policymaking (Hamilton and Wells 1990). Fiscal feder-alism refers to national government taxing, spending, and revenue-sharingpolicies regarding subnational governments. National revenues may be trans-ferred to subnational governments in the form of grants-in-aid, categoricalgrants, block grants, and project grants. These funds may be provided eitherto implement a national program or as incentives for subnational govern-ments to act in other ways (Chubb 1985; Eyestone 1977). For example, pri-or to 1996, states were responsible for distributing Aid to Families with De-pendent Children (AFDC) grants. When AFDC grants from the nationalgovernment increased, states expanded benefits, but when grants decreased,state benefits remained the same, largely because of state contextual factors(Volden 1999). Thus, national government fiscal incentives can influencestate policymaking by making states dependent on intergovernmental reve-nues or by making a program or policy more affordable (Chubb 1985; Dub-nick and Gitelson 1981; Eyestone 1977; Hamilton and Wells 1990; Peterson,Rabe, and Wong 1986).1

National government mandates requiring states to act are often inter-twined with fiscal incentives. National mandates may originate in Congress,the Supreme Court, or the executive branch, and they may come in the formof direct orders, crosscutting regulations, crossover sanctions, or partial pre-emption. With mandates, the national government forces states to act on painof some type of penalty, financial or otherwise. For example, in 1973, theSupreme Court struck down state laws that completely ban abortion; begin-ning in 1984, Congress began a series of mandates that states expand Med-icaid coverage; and the 1970 Clean Air Act and its amendments establishednational ambient air quality standards, requiring states to establish programsand provide funds to meet these standards. Each of these national govern-ment mandates increased the adoption of innovative policies by states (Gog-gin et al. 1990; Gray 1994; Grogan 1999; Krane 1993).

In their discussion of carrot-and-stick incentives (financial or otherwise)provided to states by the national government, Welch and Thompson (1980)demonstrated that innovative policies will diffuse through the states morerapidly when incentives are provided to states and that policies with posi-tive incentives will diffuse more rapidly than policies with negative incen-tives (see also Chubb 1985). Thus, the influence of national government in-centives on state policymaking has been empirically confirmed.

This discussion suggests several propositions about the conditions un-

fall 2004 / state politics and policy quarterly 321

der which national government activity should increase the likelihood thatstates will act:

P1: A state will be more likely to adopt a policy or program when thenational government mandates it to enact that policy.

P2: A state will be more likely to adopt a policy or program when thenational government provides funding for that policy or program.2

P3: A state will be more likely to adopt a policy when it is consistent withan unambiguous United States Supreme Court decision regardingthat policy.

Even beyond concrete incentives or sanctions, the national governmentmay send state policymakers signals concerning its preferences and the like-lihood of its future action. If strong and unambiguous, these signals may betaken as an endorsement by the national government of a specific policychoice, making it more likely that state policymakers will adopt that policy,all things being equal (Dubnick and Gitelson 1981). For example, when thenational government takes action on an issue, it sends a signal to state gov-ernments that the national government considers the issue to be importantand what its specific policy preference is (Dubnick and Gitelson 1981; Hamil-ton and Wells 1990). Such actions may be strong and clear, or they may beweak and ambiguous. Weak and ambiguous actions are more difficult tointerpret and, hence, they are less likely to influence state policy. For exam-ple, in 1994, President Clinton signed Executive Order 12898, instructingnational government agencies to consider the disproportionate environmen-tal impacts of their activities on racial, ethnic, and economic groups. How-ever, this order did not require any specific action, making it largely symbolic.Given this weak national signal in favor of environmental justice, subsequentstate action on the issue was limited. Indeed, only four states took similarsteps in the 1990s (Ringquist 2000). If the national government action on anissue is strong and clear, state leaders can use it to marshal support for theadoption of related policy proposals. Likewise, strong and clear national sig-nals may mobilize interest groups and advocates (Baumgartner and Jones1993; Haider-Markel 2001; Mintrom 2000). Weak and ambiguous nationalgovernment signals provide little ammunition for policy proponents and mayeven strengthen the arguments of opponents seeking to block state adoptionof specific policy proposals.

P4: When the national government sends a strong and clear policy sig-nal concerning its preferences on an issue, a state is more likely to aadopt a policy in line with those preferences.

322 allen, pettus, and haider-markel

On the other hand, when the national government cannot act, it alsosends the states a signal. For example, divided party government has beenthe norm for at least 30 years in the national government, with one partycontrolling the White House and the other party controlling at least onechamber of Congress. So if the majority party in one chamber of Congressis unable to convince the other chamber and the president to adopt a policy,state policymakers receive a signal that the national government is unlikelyto act on that issue and endorse a specific policy choice, at least until afterthe next election. Given this clear signal, state policymakers may choose toact themselves, based on factors unique to the state. And the further away thenext election that could unify national government, the more likely a stateshould be to act on its own.

P5: When the national government is unable to pass new policy becauseits institutions are divided in their preferences, a state is more likelyto adopt relevant new policies on its own.

Note that unlike our first set of propositions, P4 and P5 are about con-ditions where the national government makes no explicit attempt to directstate policymaking. Rather, these propositions suggest that national govern-ment actions can simply provide an endorsement of a policy position thatcan then be used by advocates at the state level.

Below, we do not fully address each of these propositions due to data lim-itations. Instead, we focus on P2, P4, and P5, leaving the examination of theother propositions to future research.

Additional External and Internal Influences on State Policy Adoption

Even if the national government gives clear and strong signals and incentivesfor states to act on an issue, these may conflict with the political and economicforces in a state’s environment. Indeed, the traditional model of state policyadoption focuses on forces representing states’ internal characteristics, suchas public opinion and party competition, and such external characteristicsas the adoption of a policy by neighboring or regional states (Berry and Berry1999). Thus, national government incentives and signals can be thought ofas just another external force in this model of policy adoption. Such modelshave been used to explore the adoption and diffusion of a diverse set of pol-icies in the states (Berry and Berry 1990, 1999; Gerber and Teske 2000; Gray1994; Haider-Markel 2001; Hwang and Gray 1991; Mintrom 1997; Mooney2001). We include independent variables in our models for each of our testcases to control for forces known to influence the adoption of many of thesepolicies, including partisan control of government, party competition, inter-

fall 2004 / state politics and policy quarterly 323

est group influence, public opinion, regional forces, and, where applicable,the extent of the problem, bureaucratic forces, and additional environmen-tal variables. All data and their sources are described in Appendix A, anddescriptive statistics for these variables are shown in Appendix B.

modeling national government influence onstate policymaking—three cases

To test our propositions about national government influence on state pol-icy adoption, we model this process for three policies. The only commonal-ity among the cases we have selected to test our hypotheses is that each in-volved both national and state action. These cases are not substantivelysimilar, but in each case, both state and regional forces and national govern-ment actions potentially influenced state policy adoption. Furthermore, fol-lowing Eyestone’s (1977) advice, the diffusion of each of these policies oc-curred within one political generation (30 years). Thus, our analyses speakto the validity of our propositions, but we do not interpret our results in termsof policy types or typology theories of state policy adoption.

Because state policy adoption occurs over both space and time, research-ers typically model it with event history analysis (EHA) (Berry and Berry1990, 1999). These EHA models predict the probability that a state will adopta policy in a given year, given that it has not already done so. In an EHAdataset, each state year is a case. For each case, the dependent variable, poli-cy adoption, is coded as zero unless that state adopts the policy that year. Atthe year of adoption, the case is coded as one, and no further cases are en-tered into the dataset for that state. States included in the dataset for a par-ticular year make up the risk set for that year, that is, those states that couldadopt the policy that year. The size of the risk set varies by year according tothe number of states that have previously adopted the policy. We use logis-tic regression to estimate each of our EHA models.

Truth-in-Sentencing Laws

Beliefs about the appropriateness and effectiveness of incarceration in crim-inal justice have long fluctuated between an emphasis on rehabilitation andone on punishment, but in the last two decades, the dominant arguments inthe United States have focused on increasing punitive measures (Beckett andSasson 2000). Public outcries for harsher punishments, greater media atten-tion to crime, and the increasing involvement of the federal government incriminal justice policy have all encouraged more state activity in this policyarea (Becket and Sasson 2000).

324 allen, pettus, and haider-markel

We examine the influence on state policy of the national Violent OffenderIncarceration and Truth-In-Sentencing Incentive Grant Program of the Vi-olent Crime Control and Law Enforcement Act of 1994 (PL 103–322). Thisprogram provided formula grants to states primarily for building or expand-ing correctional facilities (Office of Justice Programs 2001). One section ofthis program provides a good test of our hypothesis, P2, in that it allocated$11 billion to states for prison construction based on certain eligibility re-quirements. A key requirement is that to receive this money, a state must haveadopted a Truth-in-Sentencing (TIS) law requiring certain violent offend-ers to serve a minimum of 85 percent of their sentences. States that did notadopt these TIS laws were not eligible for these funds (Office of Justice Pro-grams 2001). Thus, the national government provided a specific financialincentive for states to adopt TIS laws. Was this incentive successful in in-fluencing state law here?

Dependent Variable. The dataset begins in the first year a state adopted a TISpolicy (1985) and ends in 2000. Data on state TIS law adoptions through 1998are from the United States General Accounting Office (1998) and for after1998, they were obtained through correspondence with state legislative re-search offices.3

Independent Variables: Vertical Diffusion Forces. The national governmentprovided a financial incentive for states to adopt TIS laws through the Vio-lent Crime Control and Law Enforcement Act of 1994. The passage of thatfederal law is coded as a binary variable with state-years prior to its adop-tion in 1994 coded as zero and state-years for 1994 and later coded as one.Proposition P2 suggests that the financial incentive provided by the federallaw will increase the probability that a state will adopt a TIS policy.

Independent Variables: Internal Determinants. Studies of state policy diffu-sion have specified a number of state characteristics that may influence pol-icy adoption, and we use them to model state adoption of TIS fully. Somestates have more revenue than others and, therefore, have greater discretionwhen considering the adoption of new policy (Taggart and Winn 1993;Tompkins 1975). Thus, we expect that greater fiscal capacity will increase thelikelihood of adopting a TIS law. We measure state fiscal capacity as gross stateproduct per capita.

A state may show a propensity or need for tougher criminal laws throughits expenditure on its corrections system (Turner et al. 1999). Such a state mayalready have overextended itself in spending on corrections. As such, it may

fall 2004 / state politics and policy quarterly 325

be less likely to adopt new laws, such as TIS laws, that would place greaterdemands on its corrections system. Thus, we expect higher per capita spend-ing on corrections in a state to reduce the likelihood of it adopting a TIS law.

In criminal justice policy, Democrats tend to support less punitive poli-cy (Meier 1992). Therefore, states where Democrats control the state legis-lature and the governorship should be less likely to adopt TIS laws.4 Parti-san competition may also affect policy adoption (Haider-Markel 1998;Holbrook and Van Dunk 1993). Berry and Berry (1999) argue that policyinnovation will vary with the electoral security of state officials—the moresecure they feel, the more likely they are to adopt new policies. Because noelected official wants to be depicted as “soft on crime” (Beckett and Sasson2000), greater party competition in a state may lead legislators to supporttougher criminal penalties, including TIS.5

Severe problems are more likely to be addressed by government actionas demand for a solution increases (Lowry 1992). Criminal justice researchhas sometimes demonstrated a positive relationship between crime rate andthe adoption of harsher penalties (Meier 1992; Taggart and Winn 1993), butother studies have found no such relationship (Haider-Markel 1998). But tocontrol for any potential influence, we include a state’s per capita crime ratein our model of TIS policy adoption.

Hero and Tolbert (1996) suggest that the racial and ethnic compositionof a state’s population can influence the policies it adopts. In particular, thepercentage of racial minorities in a state has been found to be associated withthe adoption of tougher criminal justice laws and the stronger implementa-tion of these laws (Meier 1992; Nice 1992). Thus, we hypothesize that thegreater the percentage of African Americans in a state, the greater the chanceof its adoption of a TIS law.

Public opinion ought to influence policy adoption (Berry and Berry 1999;Erickson, Wright, and McIver 1993). Public support for a policy should makeelected officials more inclined to adopt the policy, but more general measuresof public attitudes and ideology may also influence policy adoption (Berryet al. 1998). For example, strict criminal justice policies should be easier topass in a conservative state than in a liberal state. We measure public opin-ion with Berry et al.’s (1998) measure of liberal/conservative citizen ideolo-gy, with the expectation that conservatism will increase the likelihood of TISpolicy adoption.

Although interest groups often play a significant role in the policy pro-cess (Gerber and Teske 2000; Haider-Markel 2001; Mintrom 2000), on TIS,it is not obvious that there were active organized interests lobbying on eitherside of the issue. However, both the American Civil Liberties Union and

326 allen, pettus, and haider-markel

Amnesty International have opposed measures such as TIS in principle(Turner et al. 1999). We control for the potential influence of these groupsby including in our model a variable for the number of Amnesty Interna-tional chapters per 1,000 state population. We expect this interest groupmeasure to be negatively related to the adoption of TIS laws.

Independent Variables: External Determinants. Although many studies havetested whether state policy adoption is influenced by the policies of neigh-boring or regional states (Berry and Berry 1990; Haider-Markel 2001; Moo-ney 2001), there is no consensus as to this diffusion effect (Berry and Berry1999; Mooney 2001). But we control for any potential effect by including aregional variable coded as the average number of states in its United StatesCensus region that have previously passed TIS laws for each state-year. Weexpect that as more states in its region adopt TIS laws, the likelihood of a stateadopting the law will increase.

Results and Discussion. We estimated full and reduced EHA models of TISadoption using logistic regression. For parsimony, the reduced model in-cludes variables based on their theoretical importance, ability to improve themodel’s goodness-of-fit statistics, statistical significance, and the degree ofcollinearity with other variables in the model. The results are displayed inTable 1. The full model does a reasonable job of predicting state adoption ofTIS laws, even though few of the variables are statistically significant. Thereduced model provides only a marginal improvement over the full model.Although many of the hypothesized relationships do not achieve statisticalsignificance, most of the coefficient signs are in the hypothesized directionand most of the standard errors are smaller than the coefficients. As to thecontrol variables in the model, only the coefficients for corrections spend-ing, party competition, and African-American population (in the reducedmodel) were statistically significant.

In both models, national passage of the Violent Crime Control and LawEnforcement Act (VCCLA) of 1994 providing financial incentives for TIS hasa positive and statistically significant influence on state adoption of a TIS law.This supports Proposition P2. The financial incentives provided by VCCLAseem to have helped policymakers in at least some states decide to adopt TISlaws, even beyond the influence of other factors in the model.

Partial Birth Abortion Bans

Of course, financial incentives are perhaps the easiest and most direct wayfor the national government to influence state policymaking. But does the

fall 2004 / state politics and policy quarterly 327

national government have influence in the absence of such incentives? Wenow consider such a policy situation.

In 1995, Ohio became the first state to adopt legislation banning so-calledpartial birth abortions, known medically as intact dilation and extractionabortion procedures. Over the next five years, 30 other states followed Ohio’slead and adopted such bans. Congressional attempts to enact a similar banwere unsuccessful before 2003, with Congress twice passing, and PresidentClinton twice vetoing, partial birth abortion bans in the 1990s. These presi-dential vetoes effectively left this debate to be decided in the states until elec-tions significantly changed the composition of Congress or a new presidentwas elected. Congress’s attempt to pass a partial birth abortion ban received

Table 1. Determinants of State Adoption of Truth-in-SentencingLaws, 1985–2000

Independent Variables Full Model Reduced Model

National TIS incentives (1994) 3.262** 2.915**(.763) (.610)

African-American population .001 .001*(.000) (.000)

Crime rate .000 .000(.000) (.000)

Corrections spending –.010# –.010#(.006) (.006)

Democratic control –.163 –.184(.256) (.254)

Gross state product .000 .000(.000) (.000)

Amnesty International chapters .001 .—(.011)

Party competition .056* .060*(.028) (.026)

Citizen ideology .025 –.023(.019) (.019)

Regional adoption –.074 .—(.104)

Constant –6.081** –6.393**(1.689) (1.645)

Log Likelihood 183.546 184.138Chi-Square 47.547 46.955Prob. Chi-Square .000 .000df 10 8PRE .18 .18% predicted correctly 95.1% 95.1%Number of cases 588 588

Notes: Coefficients are from an event history analysis using logistic regression; — indicatesan omitted variable; statistical significance levels: ** < .01; * < .05; # < .10.

328 allen, pettus, and haider-markel

much media attention, sometimes obscuring the debate raging in the states(Christianity Today 1997). Following Congress’s failure to enact a ban withClinton’s first veto in 1996, “the issue swept the country in 1997, receivingserious legislative consideration in half the states, and accounting for over one-third of all state abortion-related legislative activity in 1997” (Saul 1998).6

Given the national government’s failure to act on a partial birth abortionban, the intense state political activity on this issue, and the rapid adoptionof these bans by the states, this issue provides an excellent case with whichto test Proposition P5. On the face of it, national government action appearsto have spurred considerable state activity, especially once it became clear thatCongress would be unable to override Clinton’s veto. These events appearto have sent a signal to the states that national action would not occur any-time soon because national institutions were divided on the issue. We nowtest P5 and this journalistic assessment of this situation with a multivariateEHA model.

Dependent Variable. Our dependent variable is coded zero for each state-yeara ban is not adopted and one in the year a state adopts one. Our dataset be-gins in 1995 because that was the first year a state adopted a partial birthabortion ban. For states not adopting by the end of 2000, the full series ofstate years (1995–2000) is included in the dataset.

Independent Variables: Vertical Diffusion Forces. Proposition P5 suggests thatcertain kinds of inaction by the national government may motivate statepolicymaking activity. In the case of partial birth abortion, congressionalpassage and President Clinton’s veto of a ban received a good deal of mediaattention. Some observers have suggested that Clinton’s 1996 veto prompt-ed states that had been waiting for national action to adopt their own bans(Moore 1997, 1999) and, indeed, the bulk of state bans were adopted in 1997and 1998. To test the influence of the national-level stalemate on state poli-cy adoption, we include a dummy variable coded zero for all state-years pri-or to 1997 and one for all 1997–2000 state-years in our EHA models.7 Wecoded one in 1997 because the 1996 veto occurred in April, which was afterthe bill introduction deadline in most state legislatures and even after somestate legislative sessions had ended for the year (Council of State Govern-ments 1999). We expect the 1996 veto variable to increase the likelihood ofa state adopting a partial birth abortion ban.

Independent Variables: Internal and External Determinants. We control forthe same internal and external forces in the partial birth abortion model as

fall 2004 / state politics and policy quarterly 329

we did in the TIS model (Table 1), with a few changes.8 First, although abor-tion is clearly a partisan issue (Adams 1997), support for partial birth abor-tion bans in Congress and state legislatures has been more bipartisan (Con-klin 1998). As such, we expect party control of state government to play asmaller role in this debate. Furthermore, while party competition is typical-ly associated with more liberal policies (Holbrook and Van Dunk 1993), thebipartisan nature of support for these conservative bans might cause partycompetition to be positively related to the adoption of these policies.

Mooney and Lee (1995, 615) proposed that on morality policy, which ishigh in public salience and low in technical complexity, elected officials wouldattend especially closely to the views of their constituents. To control forpublic opinion’s effect on partial birth abortion ban adoption, we use ameasure of public support in a state for legal abortion (Norrander 2001), withthe expectation that greater support for legal abortion should decrease thelikelihood of adopting partial birth abortion bans. We control for abortiondemand by including a state’s abortion rate per 1,000 women in the model.We posit that the likelihood of a state adopting a partial birth abortion banwill decrease as the abortion rate increases.

Unlike on crime policy, political elites have taken strong and polarizedpositions on the abortion issue (Adams 1997). As such, the preferences ofthese elites may well influence the likelihood of a state banning partial birthabortions. We measure the preferences of political elites with an average ofthe anti-abortion preference scores of each state’s United States Senate del-egation. We expect that as the anti-abortion preference scores of a state’sSenate delegation increases, the likelihood of it banning partial birth abor-tions will also increase.

Unlike on TIS policy, interest groups are very active on abortion policy(Risen and Thomas 1998), so we include measures of interest group strengthon both sides of the abortion debate. For pro-choice interest group strength,we use the number of members of the National Abortion Rights ActionLeague (NARAL) per 1,000 state population. Because we do not have a di-rect measure of anti-abortion interest group strength, we include the percent-age of a state’s population who are Protestant fundamentalists in the mod-el. Protestants and evangelicals have been very active in the anti-abortionmovement, surpassing Catholic involvement in the 1980s (Risen and Thomas1998). We hypothesize that the likelihood of a state adopting a partial birthabortion ban will increase as the percentage of its population that is Protes-tant fundamentalist increases, but it will decrease as NARAL membershipincreases.

330 allen, pettus, and haider-markel

Results and Discussion. The results of our full and reduced EHA models ofstate partial birth abortion bans are shown in Table 2. As in the TIS models,both the full and reduced models predict policy adoption fairly well, with themodel fit improving slightly in the reduced model. Again, only a few of thevariables have a statistically significant influence on the likelihood of policyadoption, including Clinton’s 1996 veto of a national partial birth abortionban, elite abortion preferences, and the strength of the pro-choice interestgroup (NARAL). Consistent with Mooney and Lee’s (1995) study of abor-tion policy adoption, none of the economic variables appear to influence theadoption of partial birth abortion bans.

Table 2. Determinants of State Adoption of Partial Birth AbortionBans, 1995–2000

Independent Variables Full Model Reduced Model

National partial birth 2.245** 2.123**abortion ban veto (1996) (.942) (.910)

Abortion rate –.042 –.040(.038) (.031)

Democratic control –.296 –.397(.287) (.251)

Elite abortion preferences –.017# –.017#(.010) (.010)

Gross state product .000 .000(.000) (.000)

NARAL membership –1.548* –1.436**(.748) (.585)

Protestant fundamentalists .034 .024(.024) (.023)

Party competition .028 .—(.030)

Public opinion on abortion .057 .—(1.827)

Regional adoption .143 .143(.097) (.093)

Constant –1.905 –.603(6.060) (1.055)

Log Likelihood 118.825 119.701Chi-Square 55.246 54.370Prob. Chi-Square .000 .000

df 10 8PRE .13 .16% predicted correctly 88.0% 88.4%Number of cases 216 216

Notes: Coefficients are from an event history analysis using logistic regression; — indicatesan omitted variable; statistical significance levels: ** < .01; * < .05; # < .10.

fall 2004 / state politics and policy quarterly 331

We argued that Clinton’s 1996 veto of a national partial birth abortionban would have a positive influence on state adoptions because it sent a clearmessage that no further action would be taken on this issue until the com-position of the national government changed. In our model, the 1996 vetovariable had a statistically significant, positive impact on state adoptions. Thissupports Proposition P5, suggesting that Clinton’s veto of a national banstimulated state policymakers to act. State policymakers wanting to prohib-it partial birth abortions appear to have viewed the 1996 veto as a signal thatfurther national action was unlikely in the short term, and, therefore, theydecided to take matters into their own hands by banning the procedure them-selves (Moore 1999).9

Hate Crime Policies

Finally, we assess the national government’s influence on state adoption ofhate crime policies. Hate crime laws typically have one or more of three ele-ments. First, some laws only call for police to identify crimes motivated bybias toward a particular group and to collect statistics on those crimes (Haid-er-Markel 1998). Second, some state and local hate crime laws provide apenalty enhancement for crimes motivated by bias. For example, an assaultsuspected of being motivated by bias toward a particular group might causethe perpetrator to receive a sentence for the assault and an additional sen-tence based on the hate crime penalty enhancement. Third, some hate crimelaws allow the victims of bias-motivated crimes to file civil lawsuits againstthe alleged perpetrator.

By 2000, 25 states required the collection of statistics on certain hatecrimes, 42 states allowed a criminal penalty enhancement for bias-motivat-ed crimes, and 28 states had laws enabling hate crime victims to file civil law-suits (Anti-Defamation League 2000). We focus on laws that allow for civilsuits or enhance penalties as these are the most stringent. In all hate crimelaws, some victim groups are covered while others are not. For example, theFederal Bureau of Investigation (FBI) defines hate crimes as crimes that arecommitted, wholly or in part, because of the victim’s race, ethnicity, religion,or sexual orientation (United States Department of Justice 1993, 1). Moststate and local hate crime laws cover race, ethnicity, and religion, but only23 states cover sexual orientation (Haider-Markel 2000).

Dependent Variable. As with the dependent variables in our first two cases,state-years are coded zero until the year a state adopts any hate crime law thatenhances penalties for bias-motivated crimes or allows for civil suits for suchcrimes, when the state-year is coded as one and the state then leaves the

332 allen, pettus, and haider-markel

dataset. The dataset begins in 1972, the first year a state adopted a hate crimelaw, and runs through 2000.

Independent Variables: Vertical Diffusion Forces. Proposition P4 holds thatwhen the national government sends a strong, clear signal as to its prefer-ences, it may motivate state policymaking activity. But if the national gov-ernment’s message is ambiguous, state policymaking should be determinedonly by state internal characteristics and regional patterns. The national gov-ernment has sent some messages on hate crime policy, but these messageshave been largely ambiguous. In 1990, Congress passed the Hate Crime Sta-tistics Act (HCSA) requiring the FBI to collect data on crimes that are com-mitted, wholly or in part, because of the victim’s race, ethnicity, religion, orsexual orientation. However, while the FBI must collect this information fromlocal and state law enforcement agencies, state and local law enforcementagency participation in the program is voluntary. As a result, critics of theHCSA suggest that it is largely symbolic, highlighting the fact that 40 to 55percent of law enforcement agencies did not participate in the programthroughout most of the 1990s (Haider-Markel 1998). That is, Congresspassed the law but impeded its implementation, giving the states an ambig-uous message concerning the national government’s position on hate crimes.Thus, unlike our first two cases, we expect the national government’s mes-sage on hate crime to be too unclear and weak to influence state policymak-ing. To test this hypothesis, we included in our EHA model a dummy vari-able coded as zero for all state-years prior to passage of the HCSA in 1990and coded one for all 1990–2000 state-years. We expect that this variable willhave no influence on the likelihood of a state adopting a hate crime law.

Independent Variables: Internal and External Determinants. Our hate crimeEHA models include a similar set of control variables as those for our firsttwo cases, including those for the extent of the problem, partisanship, partycompetition, public opinion, interest group strength, African-Americanpopulation, gross state product, regional adoption, and bureaucratic forces.Each of these variables is measured the same as for the previous two casesand the same relationships are expected, with the following exceptions.

First, we measure the extent of the problem with the state crime rate.Second, we measure law enforcement bureaucracy with the number of po-lice per capita. Because higher crime rates can increase demand for tougherlaws (Beckett and Sasson 2000; Haider-Markel 1998), we expect that a high-er crime rate will make it more likely that a state will adopt a hate crime law.Furthermore, we expect that more police per capita will also increase the like-

fall 2004 / state politics and policy quarterly 333

lihood of hate crime law adoption because the expansion of criminal lawenhances police authority, which is preferred by the law enforcement bureau-cracy (Haider-Markel 1998; Meier 1992). Third, because there are no com-parable state polls on attitudes about criminal justice policy, we use the samegeneral measure of state citizen ideology we used in the TIS case. We expectthat conservatism will decrease the likelihood of adopting a hate crime lawsince conservatives, although traditionally supportive of tougher criminalpenalties, have generally opposed hate crime laws, especially those pertain-ing to sexual orientation (Haider-Markel 1998).

Finally, like the partial birth abortion ban, interest groups are activelyinvolved in the hate crime policy debate, especially groups representingJews, African Americans, women, and lesbians and gays (Haider-Markel1998). Indeed, many states have passed hate crime laws that are virtuallyidentical to model legislation drafted by the Anti-Defamation League(ADL) (Anti-Defamation League 2000), suggesting that the ADL and itsallies played an active role in state adoption of these laws. We include twomeasures of potential interest group resources in our EHA models: mem-bership in the two largest gay groups (the National Gay and Lesbian TaskForce [NGLTF] and the Human Rights Campaign [HRC]) per 1,000 statepopulation and ADL chapters per 1,000 state population. We expect theseinterest group resources to be positively related to the likelihood of a statepassing a hate crime law.

Results and Discussion. The results of our full and reduced EHA models forhate crime law adoption are shown in Table 3. Our reduced model does amuch better job of predicting adoption, but the results of the full model aremost telling for testing our hypothesis. The coefficients of most variables inthe full model are not statistically significant, including our measure of na-tional government activity, the passage of the HCSA of 1990. This resultsupports Proposition P4, which holds that even if the national governmenttakes action on an issue, if that action is unclear or ambiguous, state policy-making is not likely to be influenced. Thus, not all national government ac-tion will influence state policymaking.

Our reduced model suggests that state hate crime policymaking is large-ly driven by internal state characteristics, including the strength of interestgroups, competition between political parties, and the size of the law enforce-ment bureaucracy. Unlike our other cases, the diffusion of hate crime policyalso appears to be influenced by the regional pattern of adoptions. As morestates in a region adopt hate crime laws, states within that region have a great-er likelihood of adopting them.

334 allen, pettus, and haider-markel

Alternative Explanations and Further Evidence

Because of the relatively crude manner in which we measured national gov-ernment action on these policies (dichotomous variables based on the yearof the action), one might question whether we are capturing the influenceof these national government actions or the influence of some other unob-served, but related, factor (Eyestone 1977). To assess this alternative expla-nation, we examine the timing of national government actions relative to statepolicy adoptions. If national government actions did have an influence, mostadopting states should have acted during or after the period when the na-tional government acted. Figure 1 shows the cumulative distribution of state

Table 3. Determinants of State Adoption of Hate Crime Policy, 1972–2000

Independent Variables Full Model Reduced Model

National HCSA adoption (1990) –.681 .—(.675)

African-American population .016 .—(.021)

Crime rate .000 .—(.000)

Democratic control .194 .—(.214)

Gross state product .000 .—(.000)

Anti-Defamation League chapters –.898 .—(1.517)

Gay group membership .003# .003**(.002) (.001)

Party competition .043# .028#(.022) (.016)

Police per capita .000 .000#(.000) (.000)

Citizen ideology –.004 .—(.015)

Regional adoption .053** .043**(.011) (.006)

Constant –6.541** –6.336**(1.360) (.850)

Log Likelihood 277.106 288.979Chi-Square 57.942 54.480Prob. Chi-Square .000 .000df 11 4PRE .12 .12% predicted correctly 95.5% 95.5%Number of cases 918 943

Notes: Coefficients are from an event history analysis using logistic regression; — indicates an omitted variable;statistical significance levels: ** < .01; * < .05; # < .10.

fall 2004 / state politics and policy quarterly 335

adoptions of each of our policies and the associated national governmentactions. The figure shows the S-shaped curves typical in the diffusion of in-novative policies (Gray 1973). But more important, the pattern of adoptionsin the TIS and partial birth abortion ban cases support our conclusion fromthe EHA analysis that the national government actions influenced state lawadoption. In these cases, most adopting states adopted the law during thesame year or in the years after national government action. On the otherhand, in the hate crime law case, where we found no influence of the nationalgovernment in our EHA models, most adopting states had acted prior to thenational government’s action. As such, Figure 1 supports the findings of ourEHA models.

In addition, we estimated a variety of models for each policy area, includ-ing other socioeconomic and demographic variables, such as alternativemeasures of racial diversity, democratic control of government, urbanizationand population density, neighboring state policy adoption, crime rate, pub-

Figure 1. Cumulative Frequency Distribution of State Policy Adoption andFederal Action

336 allen, pettus, and haider-markel

lic opinion, and state economic conditions, and including variables to cap-ture time trends similar to those used by a variety of authors (Mooney andLee 1995; Hero and Tolbert 1996; Mintrom 1997, 2000; Haider-Markel 2001;Mooney 2001). In none of our cases did these alternative measures signifi-cantly improve the EHA model prediction beyond what is presented in Ta-bles 1–3.10 Thus, we are confident that our reported models include the bestmeasures of our concepts and provide the most thorough explanation ofpolicy adoption in these issue areas.

conclusion

Scholars of public policy have noted frequently that within our federalist sys-tem of shared power, the national government might be able to take certainactions that influence state policymaking (Eyestone 1977; Gray 1973, 1994;Lowry 1992). We have explored this relationship by expanding the notion ofwhat types of national government action might have an impact in the states.We developed five propositions specifying conditions under which nationalgovernment action might influence state policymaking, and we tested threeof these propositions with case studies of state policy adoption. Our resultssupported these three propositions, suggesting several conclusions.

We have shown that vertical policy diffusion can occur even without di-rect incentives and punishments from the national government, and evencontrolling for the internal and external forces that are widely understoodto influence state policymaking. Consistent with a considerable body ofscholarship (Eyestone 1977; Gray 1973, 1994; Hedge 1983; Krane 1993; Lowry1992; Soss et al. 2001; Volden 1999; Welch and Thompson 1980), we foundthat national government financial incentives can have a strong influence onstate policymaking. But we also found that state policy adoption is more likelywhen national government action on an issue is clearly not forthcoming, suchas in the case of divided government and Clinton’s 1996 veto of the federalban on partial birth abortion. On the other hand, we found that when thenational government sends states an ambiguous message, its influence onstate policymaking is weaker or nonexistent.

Although this is an early effort to expand our understanding of the dif-ferent ways the national government can influence state policymaking andvertical diffusion, we believe it is an important step. Of course, these threecases are more suggestive than definitive. But future research needs to movebeyond the simple idea that only money and sanctions from the nationalgovernment can cause state policymakers to act. National government in-fluence is clearly both more complex and more subtle than this, and research-

fall 2004 / state politics and policy quarterly 337

ers should build on ideas outlined here. For example, might there even beunintentional effects of national government action on state policymaking?How do interest groups in the states respond to national government actions(or failures to act)? Baumgartner and Jones (1993) and Mintrom (2000),among others, have suggested that interest group activity within and acrossstates is influenced by national forces. Our findings support the notion thatinterest groups play an important role, but additional research is necessary.Finally, greater effort should be made to identify more precise measures ofnational government action than simple dichotomous variables used in thisstudy. This is important both conceptually and methodologically. Likewise,a more complete understanding of national government influence on statepolicymaking can only occur if we also develop more refined measures ofstate policymaking than the simple dichotomous variables measuring poli-cy adoption that are typically used.

appendix a: variable measurement and data sources

Variable Measurement Source

Dependent Variables

State TIS policy

State partial birthabortion ban

State hate crime law

Coded 1 for a state’s adoptionof a truth-in-sentencing lawin a given year and 0 other-wise

Coded 1 for a state’s adoptionof a partial birth abortionban in a given year and 0otherwise

Coded 1 for a state’s adoptionof any hate crime policy in agiven year and 0 otherwise

United States General Ac-counting Office (1998)and a survey conductedby the authors

Staff at the Alan Guttma-cher Institute, the Na-tional Abortion RightsLeague, and the Centerfor Reproductive Lawand Policy

Anti-Defamation League(2000)

National partial birthabortion ban veto(1996)

National HCSA adop-tion (1990)

Coded 0 for all years prior to1997 and 1 thereafter

Passage of the 1990 Hate CrimeStatistics Act; coded 0 for allyears prior to 1990 and 1thereafter

Moore (1997)

Haider-Markel (1998)

Independent Variables: Vertical Diffusion Forces

338 allen, pettus, and haider-markel

Variable Measurement Source

National TIS incentives(1994)

Passage of the Violent CrimeControl and Law Enforce-ment Act of 1994; coded 0for all years prior to 1994and 1 thereafter

Office of Justice Programs(2001)

Number of abortions per 1,000women

Percent of a state’s populationthat is African-American

Offenses reported per 10,000population

Per capita spending on correc-tions system

0 to 3 index of Democraticcontrol of lower legislativechamber, upper legislativechamber and governor’soffice

Average pro-choice score of astate’s United States Senatedelegation

Gross state product per capita

State chapters per 1,000 statepopulation

State chapters per 1,000 statepopulation

State members of the NationalGay and Lesbian Task Forceand Human Rights Cam-paign per 1,000 state popula-tion

State members of the NationalAbortion Rights ActionLeague per 1,000 state popu-lation

Abortion rate

African-Americanpopulation

Crime rate

Corrections spending

Democratic control

Elite abortion prefer-ences

Gross state product

Anti-DefamationLeague chapters

Amnesty Internationalchapters

Gay group member-ship

NARAL membership

Independent Variables: Control Variables

Henshaw and Van Vort(1994) and Alan Gutt-macher Institute staff

United States Bureau ofthe Census (variousyears)

United States Departmentof Justice (variousyears)

United States Departmentof Justice (variousyears)

United States Bureau ofthe Census (variousyears)

National Abortion RightsAction League (variousyears)

United States Bureau ofthe Census (variousyears)

Obtained from the Anti-Defamation League bythe authors

Obtained from AmnestyInternational by theauthors

Obtained from thesegroups by the authors

Obtained from NARAL bythe authors

fall 2004 / state politics and policy quarterly 339

appendix b: descriptive statistics for all variables

Std.Variable Mean Median Deviation

Abortion rate 21.75 21.20 10.68African-American population 9.34 5.95 9.60Crime rate 4909.34 4799.00 1295.47Corrections spending 84.96 73.00 46.81Democratic control 1.83 2.00 .98Elite abortion preferences 43.15 50.00 32.81Gross state product 22772.50 21623.00 9140.31Hate crime law .05 .00 .21Anti-Defamation League chapters .10 .00 .16Amnesty International chapters .074 .00 .22Gay group membership 208.97 165.52 155.46NARAL membership 1.64 1.01 .71Protestant fundamentalists 12.83 8.73 13.20

Protestant fundamen-talists

Party competition

Police per capita

Public opinion onabortion

Citizen ideology

Regional adoption

Variable Measurement Source

Percent of the state’s popula-tion belonging to fundamen-talist denominations, asdefined in Haider-Markeland Meier (1996)

State legislative district-levelmeasure of party competi-tion

Number of full-time swornpolice officers per 10,000population

Citizen opposition to legalabortion, based on the Sen-ate National Election Studies

State citizens’ liberal-to-conser-vative ideology score, withhigher scores indicatinggreater liberalism

Average number of statesadopting the relevant policy(TIS, partial birth abortionban, or hate crime law), andexcluding the state in ques-tion, in a state’s UnitedStates Census Bureau region

Bradley et al. (1992)

Holbrook and Van Dunk(1993)

United States Bureau ofthe Census (variousyears)

Norrander (2001)

Calculated by Berry et al.(1998) and updated onInter-university Con-sortium for Politicaland Social Researchwebsite

Survey of legislation con-ducted by the authors

340 allen, pettus, and haider-markel

Std.Variable Mean Median Deviation

National partial birth abortion ban veto (1996) .55 1.00 .50National HCSA adoption 1990 .17 .00 .38National TIS incentives (1994) .31 .00 .46State partial birth abortion ban .14 .00 .35Party competition 37.12 39.19 11.06Police per capita 24.50 24.00 5.07Public opinion on abortion 2.59 2.59 .26Citizen ideology 44.96 44.16 15.34State TIS policy .05 .00 .22

Note: Where a variable was included in multiple models, the descriptive statistics are shownfor the dataset for the model with the largest number of cases.

endnotes

An earlier version of this article was presented at the 2002 meeting of the Midwest Polit-ical Science Association. We thank Michael Mintrom, Dorothy Daley, and Jim Garand fortheir comments on earlier drafts of this article.

1. Bohte and Meier (2000) argue that growth in state bureaucracy and expenditures isassociated with growth in federal bureaucracy and expenditures but that this relation-ship is bottom-up, not top-down, as is commonly believed.

2. Although we do not test for the impact of the magnitude of penalties and incentiveson state policy adoption, P1 implies that the greater the penalty, the more likely states willbe to adopt relevant policies, and P2 implies that the greater the national funding, the morelikely states will be to adopt relevant policies.

3. Pennsylvania was not included in the dataset because it adopted a TIS policy in 1911,74 years before the next state adopted one.

4. Because the state legislature in Nebraska is nonpartisan, Nebraska was coded as miss-ing for this variable. Therefore, in models using the party control variable, Nebraska isnot included.

5. We use Holbrook and Van Dunk’s (1993) measure of party competition. BecauseLouisiana is missing from their data, we used other measures of partisanship and com-petition to estimate Louisiana’s score on their scale as 17.07.

6. In 2003, large majorities in the United States House and Senate passed a bill ban-ning partial birth abortion procedures, and on November 4, 2003, President Bush signedthe bill. As of this writing, the new law was being challenged in federal court.

7. The federal Partial Birth Abortion Ban Act of 1995 would have prohibited partialbirth abortions unless the procedure was necessary to save the woman’s life (Sollom 1997).President Clinton, demanding a broader exception for medical circumstances that werenot life threatening to the woman, vetoed the bill in April 1996 (Sollom 1997). Althoughthe House voted to override Clinton’s veto, proponents failed to garner enough votes todo so in the Senate. In 1997, Congress again passed a similar bill, Clinton again vetoedthe bill and Congress again failed to override the veto (Saul 1998). In our preliminaryanalysis, we included variables for both the 1996 and 1997 vetoes in our model. The 1997

fall 2004 / state politics and policy quarterly 341

veto was coded zero for all state-years prior to 1998 and one for all 1998–2000 state-years.We began coding one in 1998 rather than in 1997 because the 1997 veto occurred inOctober, well after nearly all state legislatures had ended their sessions. We expected thatthe 1996 veto would have had the greatest impact on the likelihood of a state adoptingthe bans because this veto was further from the 1998 midterm elections and the 2000presidential election. Our analysis confirmed our expectations. Indeed, the 1997 veto didnot even have a statistically significant relationship with state policy adoption in a bivariatemodel. Because of the collinearity problems created by including two annual dummyvariables in the model, the results presented here display the results for the model withonly the 1996 veto variable.

8. While a state’s socioeconomic characteristics may influence the likelihood of it adopt-ing many types of policy (Berry and Berry 1999; Lowry 1992), Mooney and Lee (1995)found that most such characteristics had no influence on the diffusion of abortion regu-lation reforms before Roe v. Wade. They hypothesized that because morality policies re-late to people’s moral, rather than their economic, condition, economic indicators shouldhave little or no influence on morality policy (Mooney and Lee 1995, 611).

9. The state bans of partial birth abortion have not gone unchallenged. Of the 28 statebans enacted by the end of 1998, courts had permanently enjoined 10 and temporarilyenjoined another seven (Cohen and Saul 1998).

10. These results are not shown, but they are available from the authors.

references

Adams, Greg D. 1997. “Abortion: Evidence of an Issue Evolution.” American Journal ofPolitical Science 41:718–37.

Anti-Defamation League of B’nai B’rith. 2000. 2000 Hate Crime Laws. New York: Anti-Defamation League of B’nai B’rith.

Baumgartner, Frank R., and Bryan D. Jones. 1993. Agendas and Instability in AmericanPolitics. Chicago: University of Chicago Press.

Beckett, Katherine, and Theodore Sasson. 2000. The Politics of Injustice: Crime and Pun-ishment in America. Thousand Oaks, CA: Pine Forge Press.

Berry, Frances, and William D. Berry. 1990. “State Lottery Adoptions as Policy Innova-tions: An Event History Analysis.” American Political Science Review 84:395–415.

Berry, Frances Stokes, and William D. Berry. 1999. “Innovation and Diffusion Models inPolicy Research.” In Theories of the Policy Process, ed. Paul E. Sabatier. Boulder, CO:Westview Press.

Berry, William D., Evan J. Ringquist, Richard C. Fording, and Russell L. Hanson. 1998.“Measuring Citizen and Government Ideology in the American States, 1960–93.”American Journal of Political Science 42:327–48.

Bohte, John, and Kenneth J. Meier. 2000. “The Marble Cake: Introducing Federalism tothe Government Growth Equation.” Publius 30:35–46.

Bradley, Martin B., Norman M. Green, Dale E. Johnson, Mac Lynn, and Lou McNeil. 1992.Churches and Church Membership in the United States, 1990. Atlanta, GA: GlenmaryResearch Center.

Christianity Today. 1997. “States Approving Bans on Partial-Birth Abortion.” Christian-ity Today, October, A2.

342 allen, pettus, and haider-markel

Chubb, John. 1985. “The Political Economy of Federalism.” American Political ScienceReview 79:994–1015.

Cohen, Susan A., and Rebekah Saul. 1998. “The Campaign against ‘Partial-Birth’ Abor-tion: Status and Fallout.” The Guttmacher Report on Public Policy 1:305–21.

Conklin, Melanie. 1998. “Lights Out on Abortion.” Progressive 62(7):15–7.Council of State Governments. 1999. The Book of the States, 1998–99. Lexington, KY: The

Council of State Governments.Dubnick, Mel, and Alan Gitelson. 1981. “Nationalizing State Policies.” In The National-

ization of State Government, ed. Jerome J. Hanus. Lexington, MA: Lexington Books.Erickson, Robert S., Gerald C. Wright, and John P. McIver. 1993. Statehouse Democracy:

Public Opinion and Policy in the American States. New York: Cambridge University Press.Eyestone, Robert. 1977. “Confusion, Diffusion, and Innovation.” American Political Sci-

ence Review 71:441–7.Gerber, Brian J., and Paul Teske. 2000. “Regulatory Policymaking in the American States:

A Review of Theories and Evidence.” Political Research Quarterly 53:849–86.Goggin, Malcolm L., Ann O. Bowman, James P. Lester, and Laurence J. O’Toole, Jr. 1990.

Implementation Theory and Practice: Toward a Third Generation. Glenview, IL: Scott,Foresman and Little, Brown.

Gray, Virginia. 1973. “Innovation in the States: A Diffusion Study.” American PoliticalScience Review 67:1174–85.

Gray, Virginia. 1994. “Competition, Emulation, and Policy Innovation.” In New Perspec-tives on American Politics, eds. Lawrence C. Dodd and Calvin Jillson. Washington, DC:CQ Press.

Grogan, Collen M. 1999. “The Influence of Federal Mandates on State Medicaid andAFDC Decision-Making.” Publius 29:1–30.

Haider-Markel, Donald P. 1998. “The Politics of Social Regulatory Policy: State and Fed-eral Hate Crime Policy and Implementation Effort.” Political Research Quarterly 51:69–88.

Haider-Markel, Donald P. 2000. “Lesbian and Gay Politics in the States: Interest Groups,Electoral Politics, and Public Policy.” In The Politics of Gay Rights, eds. Craig A. Rim-merman, Kenneth D. Wald, and Clyde Wilcox. Chicago: University of Chicago Press.

Haider-Markel, Donald P. 2001. “Policy Diffusion as a Geographical Expansion of theScope of Political Conflict: Same-Sex Marriage Bans in the 1990s.” State Politics andPolicy Quarterly 1:5–26.

Hamilton, Christopher, and Donald T. Wells. 1990. Federalism, Power, and Political Econ-omy. Englewood Cliffs, NJ: Prentice Hall.

Hedge, David M. 1983. “Fiscal Dependency and the State Budget Process.” Journal ofPolitics 45:198–208.

Henshaw, Stanley K., and Jennifer Van Vort. 1994. “Abortion Services in the United States,1991 and 1992.” Family Planning Perspectives 26:100–6, 112.

Hero, Rodney E., and Caroline J. Tolbert. 1996. “A Racial/Ethnic Diversity Interpretationof Politics and Policy in the States of the U.S.” American Journal of Political Science40:851–71.

Holbrook, Thomas M., and Emily Van Dunk. 1993. “Electoral Competition in the Amer-ican States.” American Political Science Review 87:955–62.

Hwang, Sung-Don, and Virginia Gray. 1991. “External Limits and Internal Determinantsof State Public Policy.” Western Political Quarterly 44:277–99.

fall 2004 / state politics and policy quarterly 343

Kettl, Donald F. 1983. The Regulation of American Federalism. Baton Rouge, LA: Louisi-ana State University Press.

Krane, Dale. 1993. “American Federalism, State Governments, and Public Policy: Weav-ing Together Loose Theoretical Threads.” PS: Political Science and Politics 26:186–91.

Lowry, William. 1992. The Dimensions of Federalism: State Governments and PollutionControl Policies. Durham, NC: Duke University Press.

Madison, James. [1788] 1999. “The Federalist, No. 39.” In American Politics: Classic andContemporary Readings, eds. Allan J. Cigler and Burdett A. Loomis. Boston, MA:Houghton Mifflin Company.

Meier, Kenneth J. 1992. “The Politics of Drug Abuse: Laws, Implementation, and Conse-quences.” Western Political Quarterly 45:41–69.

Mintrom, Michael. 1997. “Policy Entrepreneurs and the Diffusion of Innovation.” Amer-ican Journal of Political Science 41:738–70.

Mintrom, Michael. 2000. Policy Entrepreneurs and School Choice. Washington, DC: Geor-getown University Press.

Mooney, Christopher Z. 2000. “The Decline of Federalism and the Rise of Morality-Pol-icy Conflict in the United States.” Publius 30:171–88.

Mooney, Christopher Z. 2001. “Modeling Regional Effects on State Policy Diffusion.”Political Research Quarterly 54:103–24.

Mooney, Christopher Z., and Mei-Hsien Lee. 1995. “Legislating Morality in the Ameri-can States: The Case of Pre-Roe Abortion Regulation Reform.” American Journal ofPolitical Science 39:599–627.

Moore, Art. 1997. “States Approving Bans on Partial-Birth Abortion: Lawmakers Are NoLonger Waiting On Federal Bill.” Christianity Today, December, 97.

Moore, Art. 1999. “Partial-Birth Bans Make Little Headway in States.” Christianity To-day, April, 18.

Mossberger, Karen. 1999. “State-Federal Diffusion and Policy Learning: From EnterpriseZones to Empowerment Zones.” Publius 29:31–56.

Mossberger, Karen. 2000. The Politics of Ideas and the Spread of Enterprise Zones. Wash-ington, DC: Georgetown University Press.

National Abortion Rights Action League. Various years. Congressional Record on Choice.Washington, DC: National Abortion Rights Action League.

Nice, David C. 1992. “The States and the Death Penalty.” Western Political Quarterly45:1037–48.

Norrander, Barbara. 2001. “Measuring State Public Opinion with the Senate NationalElection Study.” State Politics and Policy Quarterly 1:111–25.

Office of Justice Programs. 2001. “Violent Offender Incarceration and Truth-in-Sentenc-ing Incentive Program.” Office of Justice Grant Program Description. Washington, DC:United States Printing Office.

Peterson, Paul E., Barry G. Rabe, and Kenneth K. Wong. 1986. When Federalism Works.Washington, DC: Brookings Institution.

Ringquist, Evan J. 2000. “Environmental Justice: Normative Concerns and EmpiricalEvidence.” In Environmental Policy, eds. Norman J. Vig and Michael E. Kraft. 4th ed.Washington, DC: CQ Press.

Risen, James, and Judy L. Thomas. 1998. Wrath of Angels. New York: BasicBooks.Saul, Rebekah. 1998. “Major Developments in the States: 1997.” The Guttmacher Report

on Public Policy 1:22–31.

344 allen, pettus, and haider-markel

Sollom, Terry. 1997. “State Actions on Reproductive Health Issues in 1996.” Family Plan-ning Perspectives 29:8–14.

Soss, Joe, Sanford F. Schram, Thomas P. Vartanian, and Erin O’Brien. 2001. “Setting theTerms of Relief: Explaining State Policy Choices in the Devolution Revolution.” Amer-ican Journal of Political Science 45:378–95.

Stream, Christopher. 1999. “Health Reform in the States: A Model of State Small GroupHealth Insurance Market Reforms.” Political Research Quarterly 52:499–526.

Taggart, William A., and Russell G. Winn. 1993. “Imprisonment in the American States.”Social Science Quarterly 74:736–49.

Tompkins, Gary L. 1975. “A Causal Model of State Welfare Expenditures.” Journal ofPolitics 37:392–416.

Turner, Susan, Peter W. Greenwood, Elsa Chen, and Terry Fain. 1999. “The Impact ofTruth-in-Sentencing and Three Strikes Legislation: Prisons, State Budgets, and CrimeRates.” Stanford Law and Policy Review 11:75–91.

United States Bureau of the Census. Various years. Statistical Abstract of the United States.Washington, DC: United States Bureau of the Census.

United States Department of Justice. 1993. Hate Crime Statistics, 1993. Washington, DC:United States Department of Justice, Criminal Justice Information Services Division.

United States Department of Justice. Various years. Sourcebook of Criminal Justice Statis-tics. Washington, DC: Office of Justice Programs, Bureau of Justice Statistics.

United States General Accounting Office. 1998. Truth in Sentencing: Availability of Feder-al Influenced Laws in Some States. Washington, DC: United States General AccountingOffice.

Volden, Craig. 1999. “Asymmetric Effects of Intergovernmental Grants: Analysis andImplications for U.S. Welfare Policy.” Publius 29:51–74.

Welch, Susan, and Kay Thompson. 1980. “The Impact of Federal Incentives on State Pol-icy Innovation.” American Journal of Political Science 24:715–29.

Wood, B. Dan. 1992. “Modeling Federal Implementation as a System.” American Journalof Political Science 36:40–67.