Berry and Berry: State lottery adoptions as policy innovations

Berry and Berry. 1990. State lottery adoptions as policy innovations: An event history analysis. American Political Science Review 84:395-416.

MAIN POINT

Studies of innovation/diffusion typically argue either that (1) internal characteristics determine when states adopt new policies or (2) regional diffusion does. The authors argue that these two views are wholly compatible. They also argue that political scientists should use Event History Analysis (EHA) more frequently when trying to explain rare outcomes. The authors test their theories with EHA by looking at adoption of state lotteries.

MOHR'S THEORY OF ORGANIZATIONAL INNOVATION

Mohr argued that innovation occurs when there is (1) motivation to innovate, assuming there are (2) resources available to help you overcome (3) obstacles to innovation. As applied the the present topic, the authors argue that internal characteristics and regional diffusion theories map nicely onto these three concepts (as can be seen by skimming through the 11 hypotheses on pgs 401-404). In short (remember that Y is adoption of a state lottery):

• MOTIVATION: Comes from poor fiscal health (thus a lottery), upcoming elections, and the interaction of these two internal characteristics.
• OBSTACLES: Internal characteristics (low personal income (no money for lotto tix) and high fundamentalism (opposition to gambling)) interact with the motivation variables.
• ABILITY to overcome obstacles: If neighboring states have already adopted a lottery, then it is easier to persuade your own people to accept a lottery. Also, single party control of govt helps. These variables interact with the motivation variables.

EVENT HISTORY ANALYSIS

• Risk set: Coded 1 if your state is "at risk" of enacting a lottery. (Coded 1 for all states after the first state adopted; coded 0 once a state adopts a lotto, since it's no longer at risk)
• Y: Whether a state-year adopts a lotto
• Unit: state-years
• Method: Probit

FINDINGS:

• Strong support for all central hypotheses
• No support for unifed party control, though; this isn't surprising since lottos aren't very conterversial. See Tables 1 and 2. Internal factors interact with diffusion.
• They claim to find support for fiscal goodness/badness, but it's not significant.
• They find support for elections being intensified by economic health (H3). They also find support for H6 and H7 by using predictions, such that obstacles increase as fiscal health deteriorates and it gets closer to an election.
• They also find support H11. Regional influence varies by motivation to innovate. This force gets stronger when it's an election year and economic health is poor.