WikiSummary, the Social Science Summary Database

 

Gerber and Green. 1998. Rational learning and partisan attitudes. AJPS 92: 794-818.

Objective

The authors attempt to develop a learning model based on Achen's model (1992, G&G called it "a static model of partisanship") and the Kalman Filter to investigate the dynamics of partisan attitudes. They model party identification as a result of prospective evaluation (but compare this with retrospective models in Fiorina and Downs).

Model

Achen's Model

Achen's crucial assumption: The average "benefits" a party provides remain constant over time; Achen equates party identification with prospective evaluations for the dynamics of partisan attitudes. This assumption cannot explain how individuals learn from the history of party performance, though, the way retrospective models do. Achen's is a Bayesian model (voters are rational, forward-looking and they update their beliefs with additional information). Party identification is the voter's estimate of the benefit differential between parties.

Gerber and Green's Model

• Each voter begins with prior beliefs about party differential.
• People use past and current information to make inferences about future performance.
• This model allows party benefits to vary over time, which cause (in this context) that partisan attitude can also vary; and that frequent and abrupt changes in the public's evaluations of parties suggests that citizens are attentive to current information.

Data

Panel data during 1990-1992, with questions about economic performance (800).

Conclusions

1. Disregard Achen's assumption of fixed party benefit level, because it is implausible.
2. Party benefits can vary over the time; when this happens, partisan attitudes can change in the wake of new information about the parties.

This model is more flexible, and more accurate, because it

• Incorporates periods of realignment and periods of stable two-party politics.
• Permits voters to beware for signs of change in party competence; and "even when voters are aware that such changes seldom occur, the possibility of change alerts the way in which they would optimally update their beliefs about party capabilities. In other words, new information is important even for experienced individuals (old, for example) and even if the rate of party change is gradual (the static model individuals are supposed to reach one point where current observations are no longer informative... in other words, individuals know enough).

The Kalman Filter model captures basic features of "how the public updates its prospective performance evaluations: at the aggregate, these evaluations change rapidly with changes in party leadership or economic conditions; at the individual level, the interaction between age and current information is confined to those under 30" (813)

• Perceptual bias seems not to prevent individuals from updating their evaluations.
• Nonetheless, there is one problem: It is difficult for these models to explain "period effects" or the differences between generations about partisan bias.

Party identification concerns the way in which people think of themselves, nonetheless, these perceptions are not fixed. "People maintain their partisan identities as long as their image of the partisan groups remains intact. But when secular realignment is afoot, the public image of the partisan groups shifts, which in turn produces a shift in party identifications and perhaps further alters perceptions of partisan groups" (816)