Disentangling Diffusion: The Effects of Social Learning and Economic Competition on State Policy Innovation and Expansion

• Author: Richard Witmer,Frederick J. Boehmke
• DOI: 10.1177/106591290405700104
• Date: 01 March 2004
• Published date: 01 March 2004
• Subject Matter: Articles

Traditionally, political science studies of state policy

adoption have focused exclusively on explaining

policy innovation—the timing of when a state first

implements that policy. One problem with this approach is

that it does not consider the extent of policy adoption fol-

lowing this innovation.1Further, the focus on the initial

date of adoption has diverted our attention from studying or

theorizing about changes in the extent of policy adoption

within a state, which we refer to as policy expansion. In this

article we apply a method for combining the study of policy

adoption and policy expansion to an important and expand-

ing policy issue, casino-style Indian gaming.

To demonstrate the advantages of studying policy adop-

tion and subsequent changes in the extent of policy, we dis-

tinguish and test two theories of policy diffusion. Policy dif-

fusion is the process whereby a state is more likely to adopt

a policy if other states have already adopted that policy. In

general, researchers have focused on a regional diffusion

effect motivated by social learning theory. Social learning

theory posits that state officials tend to draw on the experi-

ence of nearby states when considering whether they should

adopt a policy. On the other hand, economic competition

may explain policy diffusion as a response to inter-state

pressures in the form of lost business, tax revenues and jobs.

We argue that these two types of diffusion may have differ-

ent effects for policy innovation and policy expansion and

that these effects may vary across policy areas.

To distinguish between these two types of diffusion we

employ an event count model that allows us to study not

just the timing of initial innovation, but also the extent of

policy adoption in that year and whether it is expanded in

subsequent years. Thus we are able to obtain a better meas-

ure of the extent of adoption than the dichotomous measure

c u r rently employed by most innovation re s e a r ch. We

believe that applying event count models to the many

appropriate types of state policy adoption will allow us to

test a more diverse set of theories than currently exists in the

l i t e r a t u re. Examples of such policy areas include the

number of licenses to operate commercial casinos, the

number of different lottery games offered, and the number

of charter schools.

Event count models are also a perfect fit for the specific

policy that we are interested in: Indian gaming. Indian

nations2that wish to establish casino-style gaming on tribal

lands must secure agreements with the state in which they

have land in trust. States that agree to additional compacts

are clearly agreeing to more gaming. This provides an excel-

lent opportunity for us to test whether social learning and

39

Disentangling Diffusion:

The Effects of Social Learning and Economic Competition

on State Policy Innovation and Expansion

FREDERICK J. BOEHMKE, UNIVERSITY OF IOWA

RICHARD WITMER, GRINNELL COLLEGE

When modeling regional policy diffusion effects, scholars have traditionally made appeals to both social learn-

ing and economic competition as causes of diffusion. In their empirical studies of policy adoption, however,

they do not attempt to determine which of these two processes are at work. In this article, we argue that these

two types of diffusion may have different implications for when a state first adopts a policy and subsequent

changes in the extent of that policy and that these effects vary by policy area. In the specific policy area that

we study, Indian gaming, we expect social learning diffusion to influence adoption but not expansion; eco-

nomic competition should influence both policy adoption and policy expansion. Our empirical results con-

firm these predictions. To study both policy adoption and innovation, we apply models for event counts to

state policy data, which allows us to model the extent of policy adoption over time, rather than just the timing

of first adoption as is common with event histor y models.

1Though see the literature on policy re-invention (Glick and Hays 1991;

Hays 1996) for studies of how later innovators may modify the policies

adopted by early innovators.

NOTE: The authors would like to thank participants at the 2001 Iowa

Conference of Political Scientists, the 2002 Midwest Political Sci-

ence Association Meetings, a faculty research workshop at The

University of Iowa, and seminar participants at the Harris School

of Public Policy at the University of Chicago for comments and

suggestions on earlier drafts of this article. Suggestions from Gary

King and comments from Gary Segura and Chuck Shipan are also

gratefully acknowledged. Richard Witmer would like to thank

Northern Arizona University for an Intramural Grant that helped

fund his participation in the project.

Political Research Quarterly, Vol. 57, No. 1 (March 2004): pp. 39-51

2We use American Indian, Indian, and Native American interchangeably

in this paper. We also use nation as it more accurately describes the sov-

ereign rights of Indian people. See Bordewich (1996) for additional dis-

cussion of appropriate terminology.