Walker: The diffusion of innovations among the American states

Walker. 1969. The diffusion of innovations among the American states. American Political Science Review 63:880-899.

MAIN POINT

States vary in how rapidly they tend to adopt new programs. This variation can be explained with the "tree" model. TREE MODEL: There are regional leaders of innovation who emulate and compete with one another (this is the center of the tree, the main few branches). The rest of the states are smaller branches, sorted out according to the regional pioneer from which they take their cues.

DEFINITION

Innovation: When Walker speaks of innovation, he doesn't refer to anything more than adopting a new program. Even if a state adopts a new program begrudgingly and appropriates only $1000 to it, the state has adopted the new program. Furthermore, Walker refers only to programs adopted by state legislatures (not by bureaucrats).

MEASUREMENT: Y

The appendix lists 88 programs that were adopted by at least 20 state legislatures prior to 1965. For each program, Walker gives each state a percentile score (btw 0 and 1) based on the order in which it adopted the program (0 if first, 1 if last or still unadopted). He then averages each state's 88 program scores and subtracts the result from 1. Thus, the most innovative state (NY) scores 0.656; the least innovative state (MI) scores 0.298. PROBLEM: If the policies have an ideological tilt, then you would be giving a higher "innovativeness" score to more liberal states (which appears to be the case--see Table on pg 883). Many of the policies in the Appendix appear slanted toward urban and liberal innovativeness.

TWO COMPETING Xs FROM THE LIT:

Walker isn't so interested in determining which variable matters more as in his "tree" model. But still, both of these Xs seem to matter. X1 matters more.

• X1: Demographic factor, such as income, urbanity, manufacturing/farm output, literacy, and years of education.
• X2: Political factors: Looks mainly at party competitiveness, turnover in office, and malapportionment.

THE TREE MODEL: PREMISES

• Premise 1: Legislators satisfice (Simon). Rationality and information are bounded. Thus, we use heuristics (rules of thumb).
• Premise 2: A common heuristic among state legislators is, "Look for an analogy between the situation you are dealing with and some other situation, perhaps in some other state, where the problem has been successfully resolved" (889). There is inertia against change; people worry about potentially dire consequences. Legislators have a better probability of passing proposed solutions to the problem if they can say, "Look, this program worked in Nevada!"
• Premise 3: We perceive the quality of our policy by comparing to reference groups. Thus, poor states spend more (relative to income) on education than rich states because poor states are below average (but rich states are above average)--and all states are comparing themselves to the average.
• Premise 4: States compare themselves more to regional reference groups than national reference groups.

THE TEST:

Walker runs some factor analyses (of Y) and finds that the states really do sort out primarily by region (see Tables 5 and 6).