10
courts, institutional characteristics, state context, and case related characteristics.
With the findings
divided by elected and appointed courts, the author hopes to better understand the separate and distinct
conditions that contribute to consensus in state supreme courts.
VI.
Research Design
Both the consensus and ideological diversity hypotheses are tested with data collected from the
Brace-Hall State Supreme Court Data Project. In addition, diversity data also draws from data collected
by Laura Langer’s judge level data (Langer 2001; Langer and Wilhelm, forthcoming). The model of
ideological diversity contains information related to all state supreme courts from 1995 through 1998,
including the Texas Court of Criminal Appeals and the Oklahoma Court of Criminal Appeals.
Consensus data includes all murder decisions from 1995 to 1997.
Diversity Data
For the model of ideological diversity, the unit of analysis is the individual court by year from
1995 to 1998. The continuous dependent variable, ideological diversity is the range of justice ideology
found in each court and year with larger values representing greater court polarization.
Justice
preferences are obtained using the PAJID measure of judge ideology (Brace, Langer, and Hall, 2000;
Langer, 2001).
A party-adjusted, contextual measure of judge ideology, PAJID provides a proxy
measures for understanding judge personal preferences.
With a continuous dependent variable, an
ordinary least squares regression is used to determine the effect of each parameter. Robust standard
errors are used to correct for violations of any assumptions within the model.
As noted from the
hypotheses, measures of court structure, elected methods of selection, and the contextual political
environment are included within the model to assess the impact of candidate perceptions of office value,
risks, and costs, respectively.
4
The two independent variables of primary interest are the formal state method of judicial
selection and vacancy appointment authority by the governor, both representing risks associated with
office.
First, elected method of selection (American Judicature Society, 2005) is coded using a dummy
variable, coded 1 for states with elected selections and 0 in states with appointed forms of selection,
including merit selection systems. Second, vacancy appointments (American Judicature Society, 2005) is
coded as a dichotomous independent variable, coded 1 in states where the governor has the authority to
make appointments upon the event of a vacancy and 0, otherwise, where no early appointments are made.
Related to the value of office, I test whether court size, term length, or court professionalization
affects recruitment and ideological diversity. A continuous variable, term length (The Book of the States,
4
The ideological diversity model also controls for region, using dummy variables for the Northeast, South, and
West regions. Prior examinations of court diversity and composition (Cannon, 1972; Glick and Emmert, 1987) note
the explanatory value of such measures. To control for the effects of regions, they are included within the model.
Convention