The Diffusion of Local Antismoking Policies

The Diffusion of Local Antismoking Policies
Charles R. Shipan*
Department of Political Science
University of Iowa
Craig Volden
Department of Political Science
The Ohio State University
Abstract
This paper examines the decisions of the 675 largest U.S. cities about
when and whether to adopt three types of antismoking policies between
1975 and 2000. Placed in the context of horizontal and vertical policy
diffusion, cities were more likely to adopt government building and youth
access restrictions if their nearest neighbors and other cities in the state
had already adopted such measures. City governments were likely to adopt
restaurant restrictions, however, only if their nearest larger neighbor had
adopted such a restriction first, without regard to more distant localities.
For all types of policies, state laws reduced the likelihood of local
adoptions, dramatically so when the state law included a preemption
clause. All of these diffusion results were strongest for smaller cities, with
major cities acting fairly independently of other cities or of the state
government. Finally, we find strong evidence of venue shopping, with
cities adopting laws when state legislature is unlikely to do so – when, at
the state level, tobacco lobbyists are strong, health organizations are weak,
and the government leans in a conservative direction.
*
Prepared for presentation at the 2005 Midwest Political Science Association conference, Chicago, IL, April 7-10.
The authors would like to thank Jacob Nelson, Tracy Finlayson, and Chad Diefenderfer for valuable research
assistance, The Robert Wood Johnson Foundation for financial support, and Jamie Chriqui for providing us with the
updated version of the National Cancer Institute’s State Cancer Legislative Database. In addition, local tobacco
control ordinance data was provided by the American Nonsmokers’ Rights Foundation Local Tobacco Control
©
Ordinance Database .
The Diffusion of Local Antismoking Policies
Cities are all too often neglected in American conceptions of federalism. In his famous
dissent in New State Ice Co. v. Liebmann (1932), Justice Louis Brandeis referred to states, not
localities, as serving as laboratories of democracy. And the U.S. Constitution grants no political
sovereignty to cities – which are, indeed, creatures of the states. Yet, in many aspects of
American federalism, dismissing localities would result in a poor portrait of public governance.
For example, although his work is most often applied to competitive federalism at the state level,
Charles Tiebout’s (1956) theory was advanced to describe politics at the local level. Where
better to reflect heterogeneous tastes and to carry out extensive policy experiments than in the
hundreds and thousands of communities across the country?
Yet, following these conceptions of American federalism, most scholars studying such
experimentation and policy diffusion have focused on the state level. Perhaps due to data
availability issues or to a recently resurgent state politics subfield, most large-scale studies of the
politics behind the spread of policies have been done with a state-centered approach, often
lending great insight into American politics, especially given the extensive variance found across
the U.S. states. However, when looking at issues of vertical federalism – in particular, the
influence of the national government on the states – that diversity of political conditions from a
state-based approach is all too often lost. States are indeed influenced by national pressures, but
there is insufficient variance at the national level to fully examine considerations such as how
higher level policies affect lower level ones or how interest groups that are unsuccessful at one
level of government might target another.
1
This paper seeks to overcome these previous limitations by analyzing both horizontal and
vertical policy diffusion, with a primary focus on local policies in the area of antismoking
regulations. In so doing, we propose to answer the following questions. What effect do the
actions of other governments have on a city’s decision about whether to enact a law? More
specifically, if neighboring cities adopt a law, do those actions influence the likelihood that a city
will adopt a law? And will the city’s decision about whether to adopt a law be affected by the
state’s adoption of a similar law and by politics at the state level? These questions fit squarely
within the literature on diffusion, but with a twist. Most previous studies focus on one of two
types of diffusion – from state to state, or from country to country.1 In this paper we make two
key contributions. First, we examine diffusion from city to city, a type of horizontal diffusion
that is rarely studied. Second, at the same time we also focus on vertical diffusion from states to
cities in order to investigate the effect that a state-level adoption has on the likelihood that a city
will adopt a law. We examine these vertical and horizontal diffusion effects within the specific
context of antismoking laws, an area that, as we shall see, provides a unique opportunity for
studying several types of diffusion simultaneously.
Our paper proceeds as follows. First, we briefly survey the literatures on diffusion and
on antismoking regulations, in order to place this study in context. Second, we present our main
theoretical arguments and testable hypotheses. We then introduce our data and discuss how we
test our hypotheses. We finally highlight our results before concluding with a discussion of our
main substantive findings and future research directions.
1
As we discuss in the next section, there are, of course, exceptions to this general statement. Still, the majority of
diffusion studies focus on horizontal diffusion, and within that category, the majority examine state-to-state
diffusion within the U.S.
2
Policy Diffusion and Cities
Policy innovation may be thought to occur whenever a government – a national
government, a state legislature, a city – adopts a new policy (Walker 1969, Mintrom 1997a).
The impetus for this policy innovation can come from within the polity, such as when interest
groups within a state push for (or against) the adoption of a new policy, or when electoral and
institutional forces within a legislature affect the likelihood of adoption. At the same time,
pressure for policy innovation can come from outside the polity, with the spread of innovations
from one government to another, a process that political scientists and scholars in other
disciplines have labeled policy diffusion.
The literature on policy diffusion is vast and expanding rapidly. Building on a series of
classic early studies (e.g., Crain 1966, Walker 1969, Gray 1973), as well as more recent
significant theoretical and methodological advances (e.g., Berry and Berry 1990), scholars have
conducted a number of studies of diffusion during the past decade. These studies have focused
on the diffusion of a range of policies, including gambling on Indian reservations (Boehmke and
Witmer 2004), same-sex marriage bans (Haider-Markel 2001), educational reform (Mintrom
1997a), abortion policies (Mooney and Lee 1995), death penalty legislation (Mooney and Lee
1999), and HMO reforms (Balla 2001), among many others. In addition, these and other studies
have shed light on the processes by which diffusion takes place, focusing on factors that enable
or hinder diffusion, including the role of the policy’s success (Volden 2005), policy
entrepreneurs (Mintrom 1997a, 1997b, Balla 2001), and the initiative process (Boehmke
forthcoming).
What is striking about these studies is that they focus overwhelmingly on state-to-state
diffusion within the U.S. There are, to be sure, exceptions to this pattern. Some studies, for
3
example, concentrate explicitly on city-to-city diffusion; examples of these sorts of studies
include Crain’s (1966) study of fluoridation, Knoke’s (1982) analysis of the adoption of
municipal reforms, Godwin and Schroedel’s (2000) investigation of local gun control ordinances
in California, and Martin’s (2001) examination of living wage laws. And other studies, such as
Shipan and Volden’s (2005) investigation of antismoking laws, Mintrom’s (1997b) analysis of
school choice plans, and Allen, Pettus, and Haider-Markel’s (2001) study of a range of morality
laws, examine the relationship across different levels of government, such between cities and
states or between the federal government and the states, a process of vertical diffusion. Still, the
vast majority of studies on diffusion within the U.S. have focused on horizontal diffusion at the
state level.
Although most empirical tests of diffusion have taken place at the state level, there are
strong reasons to believe that diffusion is not just a state-level occurrence, but rather is a general
political phenomenon.2 One mechanism by which policies diffuse, for example, is social
learning – when one government sees that another government has adopted a new policy, it
learns from the leader’s innovation, thereby increasing the likelihood that it, too, will adopt such
a policy. Similarly, economic competition between governments can lead to diffusion. A
government may, for example, adopt a policy because the recent adoption of a similar policy by
a neighboring government provides it with political cover (e.g., Berry and Berry 1992). Or it
may adopt a policy, especially one that is popular, because it wants to keep pace with its
neighbors (e.g., Berry and Berry 1990).
Both social learning and economic competition can take place between states.3 These
processes, of course, can also exist between cities. One city can learn from the innovations of
2
3
For an example of a study that analyzes diffusion cross-nationally, see Gilardi (2005).
See Boehmke and Witmer (2004) for an attempt to disentangle these two effects.
4
another, for example. And if one city adopts a policy, then economic pressures may increase the
likelihood that another city will adopt a similar policy. Still, although scholars have uncovered
some evidence of city-to-city diffusion, previous studies have tended to suffer from one
limitation or another, with the tests in these analyses occurring only within one state (e.g.,
Godwin and Schroedel 2000), not using multivariate analyses (e.g., Crain 1966), or looking at
structural innovations rather than public policies (e.g., Knoke 1982). In our empirical analysis,
we will investigate whether policies diffuse across cities in all states, and whether this effect can
be found when we control for a variety of other potential influences. Thus, our first hypothesis
centers on the role of policy innovations across cities.
City-to-City Diffusion Hypothesis: The likelihood of a city adopting a policy increases when
neighboring cities adopt the same policy.
Our starting point is that policies can diffuse from one city to another. At the same time,
cities often share jurisdiction over policy areas with states, with each level of government vying
to put its own stamp on policy. Hence, when examining city-level policy innovations, it is
essential to take the state context into account. Viewed somewhat differently, policies can
diffuse from states to cities, with the likelihood of city-level adoption being influenced by
whether or not the state has adopted a similar policy. Again, earlier studies of city-level
diffusion have not been able to control for this possibility, either because of the bivariate nature
of the analysis (Crain 1966), the focus on a single state (Godwin and Schroedel 2000), or the
lack of shared jurisdiction over the policy choice in question, as is the case with municipal
reforms (Knoke 1982).
How does state-level action influence city-level innovations? On one hand, it is possible
that the state’s adoption of a policy would spur local governments to adopt laws. Looking at
5
national-state vertical diffusion patterns, there is significant evidence that higher levels of
government can provide incentives and information that make lower level adoption more likely
(Walker 1973; Welch and Thompson 1980; Allen, Pettus, and Haider-Markel 2004; and Shipan
and Volden 2005). On the other hand, state-level adoptions can remove the incentive for local
governments to act at all. If the state law, for example, does exactly what the city law would do,
then once the state adopts the policy there is no more reason for the city to act. And even if the
state law is weaker than what the city would prefer, the passage of a law at the state level – a law
that would encompass the city – would decrease pressure at the city level for an additional law.
On the whole, then, although certain types of state laws – namely, those that provide some
specific incentives for cities to adopt laws – may lead to a positive relationship between the
adoption of a state law and future innovations by cities, in most cases the adoption of a state law
will lead to fewer city laws in the future. We focus here on this latter case, where the state
adopts a law covering the locality, rather than providing incentives for the city to act on its own.
Therefore, we expect the following.
State-to-City Diffusion Hypothesis: The likelihood of a city adopting a policy decreases when
the state adopts a similar policy.
Although we expect that state adoptions will decrease the likelihood of cities within that
state adopting a law, there is a specific type of state action that reduces the likelihood of city
action even further. The state can, when writing its law, explicitly forbid cities within the state’s
boundaries to pass other laws (or other stronger laws) in the policy area, a tactic known as
preemption. When a state passes a preemptive law, this should have an effect above and beyond
other kinds of laws. While a locality could still pass weaker laws or laws that ensure the
6
continuance of the policy if the state were to reverse its stance, their usefulness and likelihood of
passage are greatly diminished.
Preemption Hypothesis: When a state passes a law that preempts future local laws on the same
policy, or future stronger laws, a city will be less likely to adopt a policy.
Antismoking Policies at the State and Local Level
To test the three primary hypotheses that we spelled out in the previous section requires,
first, a policy area in which states and cities share jurisdiction, and second, data on policy
adoptions at both levels of government. Antismoking policies meet both requirements. To begin
with, states and cities are active policymakers in this policy area, passing a large volume of laws
that regulate a wide range of activities (e.g., Schroeder 2004, Shipan and Volden 2005). We
focus on three types of antismoking policies in this paper: restrictions on smoking in government
buildings, restrictions on smoking in restaurants, and youth access restrictions. Our choice of
these three policies reflects several considerations. At one extreme, we could simply look to see
whether a city (or state) has adopted any sort of antismoking law. This, however, would take
place at a level of analysis that is far too aggregated; it ignores variance among specific areas of
antismoking policies; and it would make uncovering patterns difficult. At the other extreme, we
could focus on very specific differences in types of laws, assigning each law to a category based
on detailed criteria contained within the law. The problem with this approach is that the data
would then be far too disaggregated. Laws that place restrictions on smoking in government
buildings, for example, can contain outright prohibitions on smoking. They can restrict smoking
in common areas, or provide for specific areas where smoking is allowed, such as individual
offices. They can set limits on smoking near doorways, perhaps mandating a non-smoking
7
radius near the entrance. The list of specific modifications is nearly endless, even to the point of
finding laws that are only adopted in the same exact form by a handful of cities at most. Thus,
we have chosen to strike a middle ground, neither aggregating all laws together, nor
disaggregating them by their components, but rather grouping them into three fairly broad, yet
distinct, categories.
Grouping laws into these three categories also has two other beneficial effects in terms of
studying variations across types of laws. First, two of our policy areas – government buildings
and restaurants – are generally classified by public health scholars as clean indoor air laws, since
they are generally spurred by concerns over the health effects suffered by nonsmokers who are
exposed to secondhand smoke. The other policy area – youth access laws – is instead designed
to make it more difficult for young people, especially teenagers, to obtain cigarettes. Policies in
this area range from the placement of vending machines, to fines for selling cigarettes to minors,
to restrictions on the sale of cigarettes out of their original packaging. Second, some of these
laws may diffuse primarily, or even exclusively, through social learning, while others have
economic consequences. In the cases we have chosen to examine, laws regarding both
government buildings and youth access restrictions would spread primarily through social
learning, while restaurants laws would entail a mixture of social learning and economic
consequences. Specifically, restaurant bans have been opposed by restaurateurs and bar owners
who fear negative economic consequences, especially if neighboring communities do not have
similar restrictions. By studying three policies that vary on these dimensions, we can determine
whether diffusion patterns are robust to many different political conditions.
As mentioned above, in order to test our hypotheses we need to have data on these three
types of laws at both the state and the city level. Fortunately, such comparable data exist for
8
antismoking laws. We obtained data on state-level laws from the National Cancer Institute’s
State Cancer Legislative Database (SCLD), which is maintained by the MayaTech Corporation.
For city level laws we use the American Nonsmokers’ Rights (ANR) Foundation’s Local
Tobacco Control Ordinance Database. Each of these comprehensive databases contains a variety
of pieces of information about each law passed by the state or city, including the topic of the law,
the specific action taken in the law, and the date when the law was adopted. In addition, the
SCLD dataset also contains a list of state-level laws that preempted future city-level laws in the
areas of clean indoor air and youth access restrictions. Thus, these two databases allow us to test
each of the hypotheses that we have developed.4
Before turning to the data analysis, it is helpful to consider the face validity of the three
hypotheses suggested above in the context of antismoking policies. The public health literature
has provided such an assessment through a variety of case study and single-state analyses, which
suggest initial support for our main hypotheses. Skeer et al. (2004), for example, note that
neighboring towns in Massachusetts, and especially those with high levels of income and
education, were more likely to adopt restaurant restrictions. Jacobson and Wasserman’s (1997)
case studies in seven states show reduced local adoption activities following the enactment of
state laws.5 And numerous studies point to the tobacco industry’s preemption strategy at the
state level as a way to win otherwise lost battles in localities (Siegel et al. 1997, Givel and Glantz
2001, Henson et al. 2002).6
4
For additional discussions of these datasets, see Shipan and Volden (2005) and Chriqui (2000).
Andersen, Begay, and Lawson (2003) illustrate the more classical pattern of positive vertical policy diffusion, by
highlighting how state-level funding of local tobacco control initiatives helped explain their adoption in
Massachusetts. For more on local adoptions in Massachusetts, see Bartosch and Pope (2002).
6
Conlisk et al. (1995) present an interesting counter-point to this wave of research. They note how North Carolina
attached a three-month window to their preemption law, during which time localities were allowed to adopt and
5
9
Empirical Approach
Our primary dependent variables of interest capture whether a city adopts a law in each
of the areas we examine. For each of our three categories of antismoking laws, we construct a
dependent variable that is initially set equal to 0. In the year the city passes a law, this variable is
set equal to 1; and in following years the dependent variable is assigned a missing value, thus
removing the city from the dataset. This allows us to use a standard event history analysis
(EHA) to predict the probability that an event will occur given that it has not already occurred.
Our analyses include all 675 cities in the United States with populations of 50,000 or greater in
the year 2000. We do not include smaller cities, in large part because the ANR’s data is less
accurate for small cities than it is for large cities. The period that we examine begins in 1975 and
ends in 2000. We focus on this twenty-five year period starting in 1975 due to data availability
and because very few city-level antismoking laws were passed prior to this time.
To test our first hypothesis, the City-to-City Diffusion Hypothesis, we construct two
variables. First, we look within the state to see whether the nearest city with a larger population
than the city in question has previously adopted the policy (i.e., restrictions for government
buildings, restrictions on restaurants, and youth access restrictions, respectively). If it has, the
variable Nearest Bigger City takes on a value of 1; and if it has not, this variable is set equal to
0.7 For example, the nearest city to Oakland that has a larger population is San Francisco; if San
Francisco has adopted an antismoking law in the category of interest by the start of the year, then
this variable is set equal to 1. We expect this variable to have a positive influence on the
dependent variable.
grandfather in their own laws. Within that window, 89 new local regulations were passed, compared to a total of
sixteen previously.
7
To find the nearest city, we look only within state borders. Thus, the nearest bigger city for Council Bluffs, Iowa,
is Des Moines, rather than Omaha, Nebraska, even though Omaha is closer.
10
This focus on larger neighbors is consistent with the idea that cities may look to leaders
elsewhere based on size or “esteem” (Crain 1966, Walker 1969, Grupp and Richards 1975).
While this approach makes sense – cities seem more likely to be influenced by the actions of
larger neighbors than smaller neighbors – it creates a coding problem for the largest city in each
state. Since there is no “nearest bigger city” for the largest city in a state, for these cities we look
to see whether the second largest city has adopted a policy. We should note that when we
instead simply drop these largest cities from our analysis, our results remain substantively
unchanged. Similarly, looking at the nearest city, rather than the nearest bigger city, does not
significantly change our results, although there is a weaker statistical relationship among nearest
neighbors than among nearest larger neighbors, thus indicating additional support for this chosen
specification.
A second variable captures the idea that a city may be influenced not only by its nearest
neighbors, but also more generally by what other cities within the state have chosen to do. To
test for this effect, we constructed a variable called Proportion of State Population with Local
Restriction. This variable is calculated by identifying the cities that have each type of
antismoking law at the beginning of the calendar year, summing up the populations of those
cities, and dividing by the overall population within the state. Our expectation is that as this
proportion increases, so will the likelihood that the city will adopt a law.
Our second hypothesis turns our attention from the effect of other cities to the effect of
the state in which the city is situated. The State-to-City Diffusion Hypothesis suggests that the
adoption of a policy by the state will decrease the likelihood that a city will adopt a similar law.
To test this hypothesis, we create a variable, State Law, that is set equal to 1 in every year after
the state has adopted a law and is otherwise set to 0. For example, Iowa adopted a law restricting
11
smoking in government buildings in 1977, so for the regressions dealing with government
buildings in cities in Iowa, State Law is equal to 0 in 1975, 1976, and 1977, and takes a value of
1 starting in 1978 and continuing through 2000.
Of course, while state laws may have an effect on city laws, this is especially true for
state laws that explicitly preempt action by city governments. It is important to recognize that
cities can still pass laws once a state has passed a preemptive law. They might pass a law that is
weaker than the state law, for example, if the state law permits such actions. They might also
pass a stronger law than the state has passed, hoping to spark a court challenge. Or they might
attempt to craft a law that manages to sidestep the provisions of the state law. Overall, though,
we expect preemptive laws to strongly discourage the passage of city laws. Consequently, we
expect our variable, State Preemption, which is constructed in much the same way as State Law,
to have a negative influence on the likelihood of city level adoptions. Often these preemptive
laws are more general than are the substantive state laws. For example, a state may pass a
restriction on smoking in public places coupled with the preemption of local laws for all clean
indoor air policies – meaning that the localities cannot act on restaurant restrictions even though
the specific state law did not address restaurants at that point in time.
Additional Influences
Our primary interest in this paper lies in uncovering how one level of government actors
is affected by other actors at the same level and at a higher level. Thus, our three main
hypotheses posit the effects of state and city laws on the possibility of adoption by other cities.
At the same time, we recognize that we need to control for other factors in our analysis, many of
which may help us understand these diffusion processes more completely.
12
First, we need to control for characteristics of the city itself. In our empirical analysis we
include two variables that influence whether cities are likely to adopt antismoking laws,
independent of the effect of other governments. One of these variables, City Population, is set
equal to the city’s population at the time of the nearest census (scaled in 100,000s of city
residents). Larger cities have greater capabilities to pass laws, so we expect this variable to have
a positive effect on the likelihood that the city will adopt a policy. The other variable is MayorCouncil. As Knoke (1982) notes, many cities have adopted forms of government other than a
combination of a mayor and a city council, expecting these other forms of government (e.g.,
commissions, council-managers) to possess more expertise and to be more effective at passing
and implementing legislation. Thus, we anticipate Mayor-Council to have a negative influence
on the probability that a city will adopt an antismoking law.8 On the other hand, because mayorcouncil governments are likely to be more electorally responsive than other forms, the direction
of the effect may depend on the sentiment in the city or on pressure from interested parties.9 For
example, restaurant owners may be able to exert more influence against smoking policies in
mayor-council cities than in those with more politically independent city managers.10 In future
drafts we plan to include other variables that more directly address a city’s propensity to adopt
antismoking laws; currently we are collecting and coding data that will allow us to do so.
A second set of factors relate to the characteristics of the state. To begin with, if a state is
heavily engaged in tobacco production, then we expect all levels of government within that state
to be less likely to pass antismoking laws. This is true for cities as well as for the state
8
Such findings would be consistent with Moon (2002) who uncovers positive effects of city size and councilmanager governments in facilitating municipal website adoptions.
9
Interestingly, Frederickson, Johnson, and Wood (2004) show how these city government types have emulated one
another’s features, becoming more similar and homogeneous over time.
10
Such a story would be consistent with Ritch and and Begay’s (2001) discussion of how the restaurant and tobacco
industries were working together in Massachusetts to defeat state and local restaurant smoking restrictions.
13
government, so we expect our variable, Production, which measures statewide tobacco
production in millions of tons, to have a negative relationship with our dependent variables.
Similarly, to the extent that smokers will oppose antismoking laws, and to the extent that public
opinion influences policymaking, we expect that the percentage of smokers (Percent Smokers)
within a state will be negatively associated with the passage of city laws.11
Organized interests at the state level will also play an important role in the city’s decision
to pass a law. The expectations here are subtle and bring in the notion of venue shopping, which
is the idea that political actors will push for policy adoption by the politicians who are most
favorably inclined to their views (e.g., Baumgartner and Jones 1993). We include two measures
of interest group influence, Tobacco Lobbyists and Health Organization Lobbyists, where the
former is the ratio of tobacco lobbyists registered at the state level to the total number of
lobbyists registered at the state level, and the latter is a similar ratio that uses health organization
lobbyists in the numerator instead of tobacco lobbyists.12 Our expectations are that Tobacco
Lobbyists should exert a positive influence, while Health Organizations Lobbyists should have a
negative effect.
At first, these expectations about the effects of our interest group variables might seem
backwards, since they seem to suggest that when the tobacco lobby is strong, governments pass
more laws, and when public health advocates are strong, governments pass fewer laws. But this
is indeed what we should expect, given that the influence of these groups is measured at the state
level. When the tobacco industry is strong at the state level, we would expect to see an increase
in the number of local laws that are adopted, since the state is less likely to pass antismoking
11
These expectations are in line with Rigotti and Pashos’s (1991) survey-based evidence from the 1980s of a lower
likelihood of local laws in smaller communities, in tobacco producing states, and in states with a greater percentage
of smokers.
12
These measures are based on the 1994 snapshot for each state, constructed by Goldstein and Bearman (1996).
14
laws. Conversely, when public health advocates are strong at the state level, they are more likely
to be able to achieve success with the state legislature, thereby decreasing the need for, and the
likelihood of, city laws. That these interest groups are successful at the state level has been
established anecdotally in California (Magzamen and Glantz 2001) and statistically across all
states (Shipan and Volden 2005).
We expect similar state vs. local government effects for ideology variables. Using the
Berry, Ringquist, Fording, and Hanson (1998) measures, we include both State Government
Ideology and Citizen Ideology, measures in which higher values represent more liberal views.
When the state government is liberal, then, we would expect the likelihood of a city law to
decrease, since the state itself is likely to pass antismoking laws.13 When citizens within a state
are liberal, on the other hand, we would expect the likelihood of a city law to increase. Ideally,
of course, we would have city-level measures for government and citizen ideology. Since no
parallels for the Berry, Ringquist, Fording, and Hanson measures exist at the city level, we rely
on the state-level measure of citizen ideology, recognizing the problems that this may entail.14
Finally, we include dummy variables for the years in our analysis, omitting the dummy
for 2000. Inclusion of these dummies allows us to look for patterns over time and to control for
temporal dependence (Beck, Katz, and Tucker 1998). To save space in the tables, we do not
report these variables, but note that they follow a general pattern: they tend to be negative in the
early years of our series, positive in the middle, and negative toward the end, and are often
(although not always) significant. Such a pattern is typical for S-shaped policy diffusions (Gray
13
Cohen et al. (2000) note how ideological positions help explain legislative support for tobacco control.
Including state-level variables to account for city-level criteria may be problematic for a number of reasons. For
example, the percent of smokers in some cities may be more or less than that found statewide. Such inaccurately
measured variables are likely to be found less statistically significant, which should not be a problem in and of itself.
But their poor measurement may be equivalent to a type of omitted variable bias, leading to other variables (such as
the neighboring diffusion variables) accounting for the variance that would have been picked up by more accurately
14
15
1973), with a few leaders, a few laggards, and many adopters in the middle. Our results are not
substantively altered if we use year and year-squared, rather than yearly dummies. All variables
and their descriptions are summarized in the appendix.
Results
To test our hypotheses, we ran three separate regressions: one for restrictions on smoking
in government buildings, one for restrictions on smoking in restaurants, and one for youth access
restrictions. We conducted our tests using logit analyses, although the results are very robust to
other functional forms, such as probit or the complementary log-log function (Buckley and
Westerland 2004). To account for heteroskadasticity and correlation across observations, we
cluster by city using the cluster procedure in Stata 8, which assumes the errors are independent
across cities but potentially dependent within the same cities over time, and relies on
Huber/White robust standard errors. Year dummies help account for potential patterns of
temporal dependence.
[Insert Table 1 about here]
Our initial results, which we report in Table 1, strongly support the City-to-City
Diffusion Hypothesis. Nearest Bigger City is always significant at either the p < 0.05 or p <
0.01 levels and is always positive, indicating that the likelihood of a city adopting a law increases
when the nearest city that is bigger has already adopted a law.15 Substantively, the effect of
measured independent variables. See Berry (1994) for a discussion of the problem of potentially omitted variables –
a problem that is always present in diffusion studies.
15
As noted earlier, the result is just as strong regardless of how we treat the largest city in the state. The results are
weaker, although still statistically significant, when we simply look at the nearest city regardless of size.
16
neighboring city diffusion is quite large. Compared to a city without a previous adoption by its
larger neighbor, a city whose nearest bigger city already has a restaurant restriction in place has a
134% greater odds ratio of adopting its own restriction in any given year. The odds increases for
government building and youth access restrictions based on previous adoptions by the nearest
bigger neighbor are 63% and 39%, respectively.
Additional support for the city-to-city diffusion hypothesis comes from a second variable,
Proportion of State Population with Local Restriction. For both government building and youth
access laws, the coefficient on this variable is positive and significant, indicating that a city is
more likely to adopt one of these kinds of antismoking laws when a greater proportion of people
in other cities within the state are covered by a similar law. Each percent increase in the state
population covered by a local government buildings restriction is associated with a 3% boost in
the odds of such an adoption in the city we are focused on; for youth access, that increase is
1.6%. Model 2 shows that restaurants are an exception to this trend; the coefficient in this model
is still positive, but statistically indistinguishable from zero. One explanation that is consistent
with this pattern of findings is that the category of restaurants is fundamentally different from the
other two categories due to the potential economic repercussions of antismoking laws that target
restaurants. While cities learn from the experiences of their nearest neighbors and other cities in
the state in terms of youth access and government building restrictions, for restaurant restrictions
the economic considerations weigh more heavily. Here, city-level policymakers place less
weight on what the average city in the state does and more on what their nearest neighbors are
doing, as these are the areas to which their diners are likely to flee if the city unilaterally enacts a
smoking ban (or limitation) in its restaurants.
17
These results, then, provide strong evidence that policy innovations diffuse from one city
to another. This diffusion can work based on individual policy experiments, with the emulation
of a leading neighbor; or it can work at the aggregate level, with a city being influenced by the
combined action of other cities within the state. To control for features specific to each city, we
also included two other variables, City Population and Mayor-Council. As we noted earlier in
this paper, in future work we plan to include other city-level control variables in our analysis.16
These two, however, provide a good start. As predicted, the size of a city is an important
determinant of whether it adopts a law, with larger cities being much more likely to adopt
antismoking laws in each category than smaller cities. We also find some support, although
limited to restaurants, for the influence of the structure of city government. In all three models
the Mayor-Council variable is negative, and it achieves statistical significance in Model 2,
indicating that cities with mayor-council governments are less likely to adopt restaurant smoking
restrictions than are those with council-manager and other types of governments. As discussed
above, this is consistent with the view that mayor-council governments are more responsive to
constituent pressures, including those of restaurateurs.
In addition to examining the diffusion of policies from one city, or cities, to another,
another central part of our analysis concerns the effect of state-level actions on city-level
adoptions. Our results provide evidence that state laws do affect city-level adoptions. First, in
two of our models, State Law is negative and significant. Thus, the adoption of a state law
restricting smoking in government buildings or restaurants decreases the odds that a city within
that state will adopt a similar law. The effect of the state law dealing with government buildings
16
Such inclusion will help lessen the omitted variables concerns for these types of diffusion studies, raised most
clearly by Berry (1994).
18
is a 30% decline in the odds ratio for a local adoption. For restaurant restrictions, a previous
state law reduces the odds of a local adoption by nearly 90%.
Second, in all three antismoking categories we find that the passage of a law that contains
a preemptive clause leads to a dramatic decrease in the likelihood that a city will adopt a law. In
two of models, support for the Preemption Hypothesis can be seen by observing the negative and
significant coefficients for the State Preemption variable. The odds of a youth access restriction
in any given city are cut in half by a preemptive state-level youth access law. For government
buildings, this effect is a 93% reduction in the odds of a local adoption upon a state-level clean
indoor air preemption. In the model for Restaurants, the importance of this variable is indicated
by its omission from our results: State Preemption cannot be included in this regression because
it perfectly predicts failure. In other words, during our entire time period, despite 2,342
opportunities, not one single city passed a law relating to smoking in restaurants after the state
passed a preemptive clean indoor air law!
In addition to support for our primary hypotheses, the results in Table 1 also contain a
number of other interesting findings for the other variables in our model. In particular, a number
of state-level variables are significant. Production and Percent Smokers each has a negative and
significant influence on city-level adoptions of restaurant-targeted antismoking laws, although
not others. Once again illustrating an exception for restaurant restrictions, cities within states
that produce a lot of tobacco, or where larger percentages of the population smoke, are less likely
to adopt laws that restrict smoking in restaurants. We find no similar statistically significant
effect for laws that cover smoking in government buildings or youth access to tobacco.
Even more interesting are the results concerning the influence of interest groups and
ideologies. To begin with, we find strong evidence for the hypothesized relationships between
19
the prevalence of interest groups and the likelihood of city-level adoption of antismoking laws.
Across the three equations, we find that cities are more likely to adopt antismoking laws if they
are located in states with prominent state-level tobacco lobbies, relative to other interest groups,
and that cities are less likely to adopt antismoking laws in states that have strong state-level
public health advocates. The effects are particularly large for tobacco lobbyists. Compared to a
state with no tobacco lobbyists, one in which the tobacco lobby makes up one percent of all
lobbyists before the state legislature displays a 30% larger odds ratio for a government buildings
restriction in each city in the state in each year and 20% greater odds for restaurant and youth
access restrictions.
At first blush these results might seem counterintuitive; but what they imply is that
advocates engage in venue shopping. When the tobacco lobby is strong at the state level,
advocates of antismoking laws realize that they will have little success at the state level and so
turn their attention to cities. The direction of causality, however, is unclear. The tobacco
industry, realizing it was facing local-level defeats, may have turned its attention to state-level
efforts in an attempt to stop the damage, potentially with preemptive state laws.17 The
coefficient sizes are smaller for health organization lobbyists (who are likely engaged on many
other issues), but the findings remain suggestive. When public health advocates are strong at the
state level, they see less need to pass city-level laws due to a greater likelihood of state-level
successes.
We find a similar pattern for ideology. A negative coefficient on State Government
Ideology would imply that as the elected state officials (i.e., members of the state legislature and
the government more broadly) become more liberal, fewer laws are adopted at the city level.
17
Such a possibility would be consistent with anecdotal evidence by Siegel et al. 1997, Givel and Glantz 2001, and
Henson et al. 2002.
20
And in two of our models, government buildings and restaurants, that is what we find. When the
state government is liberal, it is thought to be more likely to adopt antismoking laws for the
whole state; hence, there is less incentive for cities to push to adopt such laws. We do not,
however, find such an effect for youth access laws, perhaps indicating that the politics
surrounding such laws are less ideological than for clean indoor air laws. Finally, in two of our
models the coefficient for Citizen Ideology is positive and significant, which implies that that as
the citizens of a state are more liberal, cities within that state are more likely to adopt
antismoking laws.
Interactive Effects
Although the above models have controlled for city and state characteristics, they have
fundamentally treated the diffusion process as the same for all cities. Each city was found to be
more likely to adopt policies found in other cities and less likely where the state government had
already acted (especially in a preemptive fashion). There is no reason to believe, however, that
diffusion works the same way for every government. Dealing with state-level antismoking
adoptions, for example, Shipan and Volden (2005) find different diffusion patterns for states
with more professional legislatures than those with less professional legislatures. Based on a
similar logic, we are here concerned with whether the diffusion hypotheses advanced above are
robust across different city types. Perhaps some kinds of cities are more likely to imitate others,
or to be influenced by the state, than are other cities. By including interactions between our
diffusion measures and city characteristics, we find no systematic evidence that the type of
governance (i.e., Mayor-Council) has a differential effect in policy diffusion. We do, however,
find some evidence of such an effect for the city’s population. Models 4, 5, and 6 in Table 2
21
show a significant result for the interaction of city size with our diffusion variables in two of our
categories.
[Insert Table 2 about here]
The inclusion of these new interacted variables has little effect on the substantive
findings for our other variables of interest. Compared to Table 1, the findings in Table 2 show
extremely similar, and in some cases slightly stronger, results for the main variables of concern.
We therefore turn our attention here to the interactive effects. First, the interaction of City
Population with Nearest Bigger City is negative and significant for government buildings and
youth access laws, which means that bigger cities are, all else equal, less likely to adopt policy
innovations just because one of their neighbors did so. The size of these coefficients suggests
that, for cities of about half a million residents, the actions of neighboring larger cities are
irrelevant, making the city no more or less likely to adopt its own policy in these areas. Smaller
cities are still reliant on the actions of their nearest bigger neighbors, and all cities – large or
small – seem to be focused on their neighbors regarding restaurant restrictions.
Second, the interaction of City Population with State Law is positive and significant for
government buildings and youth access laws, indicating that big cities are more likely to buck the
trend and pass a law even after the state has already done so. Taken together, these results
reinforce the image of leader and laggard cities, with the largest cities in a state acting on their
own, and the smaller cities taking their cues from their larger neighbors.
22
Discussion
We began this study with three expectations: (1) policy adoption in a federal system is
influenced by horizontal and vertical diffusion patterns, (2) horizontal diffusion is as likely at the
local level as at the state level, and (3) studying vertical diffusion through local-state relations is
more likely to be fruitful than studying state-national relations. On all grounds, we have found
support. In an important regulatory policy area, cities have been found to follow robust patterns
of policy diffusion over the past thirty years. For U.S. cities with populations over 50,000,
antismoking policies have spread based on the actions of other cities statewide and on the
behavior of each city’s nearest larger neighbor. This horizontal diffusion pattern was especially
strong for smaller cities. And, for smoking restrictions in restaurants, diffusion was much more
strongly determined by cities in close proximity than by the actions of distant cities elsewhere in
the state.
State-level influences on cities’ policy adoptions were manifold. The adoption of state
laws dealing with government buildings, with restaurants, and with youth access to tobacco were
frequently sufficient to discourage any further regulations at the local level. This was especially
true when the state legislation contained a preemptive component, limiting the ability of
localities to exceed state standards. Also intriguing are the politics surrounding such local-state
interactions. For example, greater tobacco industry lobbying pressure at the state level was
found to be associated with greater success in adopting local restrictions, whether because
antismoking activists turned to the local level in these cases or because the tobacco industry
sought statewide successes where their local strategies were inadequate. The reverse effect was
found for state-level health organization lobbyists, cementing the evidence of a systematic
23
pattern. Moreover, more liberal state governments were associated with fewer local restrictions,
with localities again potentially relying on the state for action in these cases.
Such support for the venue shopping aspects of vertical policy diffusion are intriguing,
and would have been nearly impossible to establish systematically in state-national studies,
which lack significant variation at the upper level of government. This does not mean that
similar relationships are not important for federal-state interactions. On the contrary, we believe
that the present study opens a window to understanding vertical policy diffusion much more
generally. We expect interest groups and political entrepreneurs to pursue their policy goals
wherever they deem most likely to be successful – be that at the local, state, or national level.
Finally, it is enlightening to view this study in the context of work that has come before
it. In particular, Shipan and Volden (2005) study the other side of this state-local antismoking
relationship. They find a “snowball effect” of local laws being more likely to lead to state laws
only for states with highly professional legislatures and those with strong health organization
lobbyists at the state level. For less professional states and those without such antismoking
activists, widespread local laws were associated with fewer subsequent state laws in the same
three areas studied here, through a “pressure valve effect.” Taken together, these two studies
provide a number of important and surprising results that uncover diffusion pathways and the
politics behind them. Due to space concerns, we note just one, a finding that points out the
Catch-22 situation for health advocates in states with non-professional legislatures. In these
states, a focus on local action by public health groups will undermine their ability to achieve
success at the state level. But if they concentrate on the state level, then if they are successful
they will undermine the possibility of future (and possibly stronger) local actions, while if they
are unsuccessful they will have used up the resources they could have used at the local level.
24
These considerations are made even more complex by the actions of the tobacco industry, which
has had much greater political strength at the state than the local level. As we found here, when
the tobacco industry was strong at the state level, antismoking advocates adopted a local-first
strategy, which was quite successful. But those groups that did not parlay local successes into
state action soon enough found state laws to be relatively hollow, often preempting their local
successes.
Ultimately, policy diffusion is the result of thousands of such calculations by citizens,
policymakers, and interest groups. A focus on their actions before Congress alone, or in only
one city or one state, will tell an incomplete story. Even analyzing policy diffusion across
multiple states is likely limited by its horizontal approach. Decisions in American federalism are
truly interrelated. As Justice Brandeis suggested, the actions of “a single courageous state” – or
city – may serve as an experiment that under the right circumstances could be emulated across
the country.
25
Appendix: Variable Descriptions, Summary Statistics, Sources
Variable
a
Antismoking Policy Adoption
a,b
Nearest Bigger City
Proportion of Population with
a,b
Local Restrictions
State Law
c
State Preemption
City Population
c
b
d
Mayor-Council
Production (millions of tons)
e
f
Percent Smokers
g
Tobacco Lobbyists
Health Organization Lobbyists
h
State Government Ideology
h
Citizen Ideology
g
Description
Dependent variable = 1 if city adopts its first law
in this area in this year. Set = 0 if no adoption to
date. (Observation removed if already adopted.):
Government Buildings
Restaurants
Youth Access Restriction
Dummy = 1 if the nearest city that is larger than
the observation city adopts its law in this area
prior to the observation year:
Government Buildings
Restaurants
Youth Access Restriction
Proportion of state population living in localities
with restrictions in this area at start of the year:
Government Buildings
Restaurants
Youth Access Restriction
Dummy = 1 if state adopted restriction in this
area prior to this year:
Government Buildings
Restaurants
Out-of-Package Sales Restriction
Dummy = 1 if state adopted law prior to this year
that prohibits or limits city-level government laws
in this area:
Government Buildings (Clean Indoor Air)
Restaurants (Clean Indoor Air)
Youth Access Restrictions
City population (in 100,000s) at the time of the
nearest census
Dummy = 1 if the city has a Mayor-Council form
of government
Amount of tobacco production in state in millions
of tons
Percent of adults who smoke in the state
Proportion of lobbyists in the state working for
tobacco industry, based on 1994 snapshot
Proportion of lobbyists in the state working for
health organizations, based on 1994 snapshot
Ideology score for state government
Ideology score for state citizenry
Mean
St. Dev.
0.016
0.018
0.013
0.125
0.131
0.114
0.250
0.248
0.135
0.433
0.432
0.342
0.136
0.143
0.056
0.176
0.186
0.099
0.533
0.352
0.206
0.499
0.478
0.404
0.193
0.193
0.046
0.394
0.394
0.210
1.378
3.625
0.324
0.468
0.016
24.5
0.071
3.51
0.014
0.007
0.089
52.5
48.6
0.062
21.6
13.5
a
Data sources: Constructed by authors based on American Nonsmokers’ Rights Foundation Local Tobacco Control
©
Ordinance Database .
b
Constructed by authors based on U.S. Census data.
c
Constructed based on National Cancer Institute, State Cancer Legislative Database Program, Bethesda, MD: SCLD.
d
City and County Databook, various years.
e
U.S. Department of Agriculture website (www.nass.usda.gov:81/ipedb/).
f
Centers for Disease Control and Prevention website (www2.cdc.gov/nccdphp/osh/state/report_index.asp).
g
Constructed by authors based on Goldstein and Bearman (1996).
h
Berry, Ringquist, Fording, and Hanson (1998) data on ICPSR website.
26
References
Allen, Mahalley D., Carrie Pettus, and Donald P. Haider-Markel. 2004. Making the National
Local: Specifying the Conditions for National Government Influence on State Policymaking.
State Politics and Policy Quarterly 4(3): 318-344.
Andersen, B. S., M. E. Begay, and C. B. Lawson. 2003. Breaking the Alliance: Defeating the
Tobacco Industry’s Allies and Enacting Youth Access Restrictions in Massachusetts.
American Journal of Public Health 93(11): 1922-1928.
Balla, Steven J. 2001. Interstate Professional Associations and the Diffusion of Policy
Innovations. American Politics Research 29(3): 221-245.
Bartosch, W. J., and G. C. Pope. 2002. Local Enactment of Tobacco Control Policies in
Massachusetts. American Journal of Public Health 92(6): 941-943.
Baumgartner, Frank R., and Bryan D. Jones. 1993. Agenda and Instability in American Politics.
Chicago: University of Chicago Press.
Beck, Nathaniel, Jonathan N. Katz, and Richard Tucker. 1998. Taking Time Seriously: TimeSeries-Cross-Section Analysis with a Binary Dependent Variable. American Journal of
Political Science 42(4): 1260-1288.
Berry, Frances Stokes. 1994. Sizing Up State Policy Innovation Research. Policy Studies
Journal 22(3): 442-456.
Berry, Frances Stokes, and William D. Berry. 1990. State Lottery Adoptions as Policy
Innovations: An Event History Analysis. American Political Science Review 84(2): 395-415.
Berry, Frances Stokes, and William D. Berry. 1992. Tax Innovation in the States: Capitalizing
on Political Opportunity. American Journal of Political Science 36 (3): 715-742.
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(1): 327-348.
Boehmke, Frederick J. Forthcoming. The Indirect Effect of Direct Legislation: How Institutions
Shape Interest Group Systems. Columbus, OH: Ohio State University Press.
Boehmke, Frederick J., and Richard Witmer. 2004. Disentangling Diffusion: The Effects of
Social Learning and Economic Competition on State Policy Innovation and Expansion.
Political Research Quarterly 57(1): 39-51.
Brandeis, Louis Dembitz. 1932. Dissenting opinion. New State Ice Co. v. Liebmann, 285 U.S.
262, 311.
27
Buckley, Jack, and Chad Westerland. 2004. Duration Dependence, Functional Form, and
Correct Standard Errors: Improving EHA Models of State Policy Diffusion. State Politics
and Policy Quarterly 4 (1): 94-113.
Chriqui, Jamie F. 2000. Restricting Minors’ Access to Tobacco Products: An Examination of
State Legislation and Policy Innovation. Ph.D. Dissertation, University of Maryland.
Cohen, Joanna E., et al. 2000. Political Ideology and Tobacco Control. Tobacco Control 9:
263-267.
Conlisk, E., et al. 1995. The Status of Local Smoking Regulations in North Carolina Following
a State Preemption Bill. JAMA: Journal of the American Medical Association 273(10): 805807.
Crain, Robert L. 1966. Fluoridation - Diffusion of an Innovation among Cities. Social Forces
44(4): 467-476.
Frederickson, H.G., G.A. Johnson, and C. Wood. 2004. The Changing Structure of American
Cities: A Study of the Diffusion of Innovation. Public Administration Review 64(3): 320330.
Gilardi, Fabrizio. 2005. The Institutional Foundations of Regulatory Capitalism: The Diffusion
of Independent Regulatory Agencies in Western Europe. Annals of the American Academy
of Political and Social Science 598: 84-101.
Givel, Michael S., and Stanton A. Glantz. 2001. Tobacco Lobby Political Influence on U.S.
State Legislatures in the 1990s. Tobacco Control 10: 124-134.
Godwin, Marcia L., and Jean R. Schroedel. 2000. Policy Diffusion and Strategies for Promoting
Policy Change: Evidence from California Local Gun Control Ordinances. Policy Studies
Journal 28(4): 760-776.
Goldstein, Adam O., and Nathan S. Bearman. 1996. State Tobacco Lobbyists and Organizations
in the United States: Crossed Lines. American Journal of Public Health 86(8): 1137-1142.
Gray, Virginia. 1973. Innovation in the States: A Diffusion Study. American Political Science
Review 67: 1174-1185.
Grupp, Fred W., Jr., and Alan R. Richards. 1975. Variations in Elite Perceptions of American
States as Referents for Public Policy Making. American Political Science Review 69(3): 850858.
Haider-Markel, Donald P. 2001. Policy Diffusion as a Geographical Expansion of the Scope of
Political Conflict: Same-Sex Marriage Bans in the 1990s. State Politics and Policy
Quarterly 1: 5-26.
28
Henson, R., et al. 2002. Clean Indoor Air: Where, Why, and How. Journal of Law, Medicine,
and Ethics 30(3, Suppl. S): 75-82.
Jacobson, Peter D., and Jeffrey Wasserman. 1997. Tobacco Control Laws: Implementation and
Enforcement. Santa Monica, CA: RAND.
Knoke, David. 1982. The Spread of Municipal Reform: Temporal, Spatial, and Social
Dynamics. American Journal of Sociology 87(6): 1314-1339.
Magzamen, S., and Stanton A. Glantz. 2001. The New Battleground: California’s Experience
with Smoke-Free Bars. American Journal of Public Health 91(2): 245-252.
Martin, I. 2001. Dawn of the Living Wage: The Diffusion of a Redistributive Municipal Policy.
Urban Affairs Review 36(4): 470-496.
Mintrom, Michael. 1997a. Policy Entrepreneurs and the Diffusion of Innovation. American
Journal of Political Science 41(3): 738-770.
Mintrom, Michael. 1997b. The State-Local Nexus in Policy Innovation Diffusion: The Case of
School Choice. Publius: The Journal of Federalism 27(3): 41-59.
Moon, M. J. 2002. The Evolution of E-Government among Municipalities: Rhetoric or Reality?
Public Administration Review 62(4): 424-433.
Mooney, Christopher Z., and Mei-Hsien Lee. 1995. Legislative Morality in the American
States: The Case of Pre-Roe Abortion Regulation Reform. American Journal of Political
Science 39(3): 599-627.
Mooney, Christopher Z., and Mei-Hsien Lee. 1999. The Temporal Diffusion of Morality
Policy: The Case of Death Penalty Legislation in the American States. Policy Studies
Journal 27(4): 766-780.
Rigotti, Nancy A., and Chris L. Pashos. 1991. No-Smoking Laws in the United States: An
Analysis of State and City Actions to Limit Smoking in Public Places and Workplaces.
JAMA 266(22): 3162-3167.
Ritch, W. A., and M. E. Begay. 2001. Strange Bedfellows: The History of Collaboration
between the Massachusetts Restaurant Association and the Tobacco Industry. American
Journal of Public Health 91(4): 598-603.
Schroeder, Steven A. 2004. Tobacco Control in the Wake of the 1998 Master Settlement
Agreement. New England Journal of Medicine 350(3): 293-301.
Shipan, Charles R., and Craig Volden. 2005. Policy Diffusion from Cities to States:
Antismoking Laws in the U.S. Manuscript, University of Iowa and Ohio State University.
29
Siegel, Michael, et al. 1997. Preemption in Tobacco Control: Review of an Emerging Public
Health Problem. JAMA 278(10): 858-863.
Skeer, M., et al. 2004. Town-Level Characteristics and Smoking Policy Adoption in
Massachusetts: Are Local Restaurant Smoking Regulations Fostering Disparities in Health
Protection? American Journal of Public Health 94(2): 286-292.
Tiebout, Charles M. 1956. A Pure Theory of Local Expenditures. Journal of Political Economy
64(5): 416-24.
Volden, Craig. 2005. States as Policy Laboratories: Emulating Successes in the Children’s
Health Insurance Program. Manuscript, The Ohio State University.
Walker, Jack L. 1969. The Diffusion of Innovations among the American States. American
Political Science Review 63: 880-899.
Walker, Jack L. 1973. Problems in Research on Diffusion of Policy Innovations. American
Political Science Review 67(4): 1186-1191.
Welch, Susan, and Kay Thompson. 1980. The Impact of Federal Incentives on State Policy
Innovation. American Journal of Political Science 24(4): 715-726.
30
Table 1: City Level Adoption of Antismoking Policies
City-to-City Diffusion
Nearest Bigger City
Proportion of State Population
with Local Restriction
State-to-City Diffusion
State Law
State Preemption
City-level Controls
City Population (in 100,000s)
Mayor-Council
Citizen and Producer Pressures
Production (millions of tons)
Percent Smokers
State-Level Organized Interests
Tobacco Lobbyists
Health Organization Lobbyists
Ideology
State Government Ideology
Citizen Ideology
Wald χ 2
N
Model 1
Government Buildings
Model 2
Restaurants
Model 3
Youth Access
0.49***
(0.18)
0.85***
(0.18)
0.33**
(0.18)
2.97***
(0.53)
0.36
(0.59)
1.61**
(0.84)
-0.37**
(0.19)
-2.06***
(0.27)
-0.16
(0.21)
-2.67***
(0.71)
###
-0.78**
(0.47)
0.04***
(0.01)
0.04***
(0.01)
0.05***
(0.01)
-0.23
(0.19)
-0.52***
(0.19)
-0.23
(0.18)
-0.98
(1.47)
-3.46**
(2.09)
-15.14
(14.95)
-0.03
(0.03)
-0.08***
(0.03)
-0.004
(0.03)
27.13**
(13.37)
19.80**
(12.01)
19.43**
(11.09)
-2.10*
(1.39)
-4.16***
(1.70)
-0.81
(1.30)
-0.012**
(0.006)
-0.007*
(0.005)
0.002
(0.004)
0.004
(0.009)
0.011*
(0.009)
0.019**
(0.009)
292.35***
11,886
359.10***
10,704
186.33***
10,908
Robust standard errors in parentheses, clustered by city. All models include yearly dummy variables and a constant.
### indicates predicts failure perfectly, thus variable omitted. *** p < 0.01, ** p < 0.05, * p < 0.1 (one-tailed tests).
31
Table 2: The Effect of Population on Diffusion to Cities
Model 4
Government Buildings
Model 5
Restaurants
Model 6
Youth Access
0.55***
(0.19)
0.74***
(0.21)
0.45**
(0.19)
Nearest Bigger City ×
City Population
-0.09***
(0.03)
0.08
(0.07)
-0.08***
(0.03)
Proportion of State Population
with Local Restriction
3.02***
(0.54)
0.38
(0.60)
1.55**
(0.84)
-0.54***
(0.22)
-2.04***
(0.32)
-0.21
(0.21)
0.15**
(0.08)
-0.02
(0.16)
0.03*
(0.02)
-2.70***
(0.72)
###
-0.79**
(0.47)
0.12***
(0.03)
0.04***
(0.01)
0.10***
(0.03)
-0.35**
(0.19)
-0.53***
(0.19)
-0.29*
(0.19)
-1.02
(1.48)
-3.45*
(2.11)
-15.31
(15.07)
-0.03
(0.03)
-0.08***
(0.03)
-0.001
(0.03)
29.60**
(13.44)
19.35*
(12.07)
19.03**
(11.04)
-2.95**
(1.56)
-4.38***
(1.76)
0.72
(1.32)
-0.012**
(0.006)
-0.007*
(0.005)
0.003
(0.004)
0.006
(0.009)
324.70***
11,886
0.012*
(0.009)
359.80***
10,704
0.021**
(0.009)
185.56***
10,908
City-to-City Diffusion
Nearest Bigger City
State-to-City Diffusion
State Law
State Law × City Population
State Preemption
City-level Controls
City Population (in 100,000s)
Mayor-Council
Citizen and Producer Pressures
Production (millions of tons)
Percent Smokers
State-Level Organized Interests
Tobacco Lobbyists
Health Organization Lobbyists
Ideology
State Government Ideology
Citizen Ideology
Wald χ 2
N
Robust standard errors in parentheses, clustered by city. All models include yearly dummy variables and a constant.
### indicates predicts failure perfectly, thus variable omitted. *** p < 0.01, ** p < 0.05, * p < 0.1 (one-tailed tests).