Document Type

Dissertation

Date of Award

1975

Keywords

New York State, Legislature, Legislation, New York State

Degree Name

Doctor of Philosophy (PhD)

Department

Political Science

First Advisor

Joseph M. Firestone

Second Advisor

Eduard A. Ziegenhagen

Third Advisor

Paul A. Smith

Abstract

An attempt to develop a mathematical model which predicts the extent of legislative action on any bill introduced in the New York State Senate.

The legislative process is viewed as a stochastic branching process, where the researcher is attempting to predict where in the process any introduced bill resides at the end of the legislative session. The information available about each bill consists of political information about the sponsor(s) (i.e., party, seniority, faction, etc.); the relationship between the prime sponsor and the committee to which the bill is referred; and the content of the bill. The variables were chosen on the basis of a literature review which revealed them to be among the more traditional concerns of scholars of the legislative process. Additionally, for each variable, relationships are hypothesized between the variable selected and the extent of legislative action.

The data for this research consists of all bills introduced in the 1965, 1966 and 1969 New York Senate Sessions, about 15,000 bills in all. Bill histories were read and coded to obtain the requisite information.

Each of the hypotheses relating the independent variables to the dependent variable are tested for all three sessions. While some of the hypotheses are speculative, of the fifteen hypotheses developed and tested, only six are confirmed; an additional four are confirmed in modified form, and five others are either not confirmed due to inconclusive evidence or disconfirmed.

The variables are then aggregated in a linear regression model and in a self-searching multiplicative model. The linear regression model explains between three and fifteen percent of the variance, depending upon the session and category of the dependent variable. The multiplicative model employs a relatively novel self-searching algorithm for binary splits, but the results represent only a marginal improvement over the employment of a best guess strategy.

The fact that these models are such poor predictors of legislative outcome suggests that the traditional state legislative literature, as interpreted here and applied to New York, may be based upon dubious assumptions. In any event, replication of this endeavor both longitudinally and cross-sectionally is clearly needed.

Download

Share

COinS