2002 State Legislative Survey (ICPSR 20960)

Version Date: Mar 25, 2008 View help for published

Principal Investigator(s): View help for Principal Investigator(s)
John M. Carey, Dartmouth College; Richard G. Niemi, University of Rochester; Lynda W. Powell, University of Rochester; Gary Moncrief, Boise State University


Version V1

Slide tabs to view more

This survey of state legislators updates and expands the 1995 Carey, Niemi, and Powell survey, STATE LEGISLATIVE SURVEY AND CONTEXTUAL DATA, 1995: [UNITED STATES] (ICPSR 3021), which asked many of the same questions. Questionnaires were mailed to all 7,430 state legislators (50 states, 99 chambers) in February 2002, with follow-up letters in March and May of the same year. State legislators were surveyed on the importance of various factors in learning how to do their job, the importance of various sources of information available to them, whether they had authored any bills that became law during their most recent term, whether they specialized in single policy areas, and how much time they spent on legislative duties and tasks. Opinions were sought on the relative influence of party leaders and staff, among others, in determining legislative outcomes, and how much attention party leaders should give to various duties. Additional questions asked whether respondents followed their conscience or the wishes of their constituency when making decisions, the political views of their constituency, and which groups they considered to be their strongest supporters. Information was also collected on opposition candidates, vote percentages, campaign expenditures, previously held public and appointed offices, and future political aspirations. Demographic information includes sex, race, household income, religious preference, political party affiliation, and political philosophy.

Carey, John M., Niemi, Richard G., Powell, Lynda W., and Moncrief, Gary. 2002 State Legislative Survey. Inter-university Consortium for Political and Social Research [distributor], 2008-03-25. https://doi.org/10.3886/ICPSR20960.v1

Export Citation:

  • RIS (generic format for RefWorks, EndNote, etc.)
  • EndNote
National Science Foundation (SES-0212310)

To protect respondent privacy, the data are restricted from general dissemination. Users interested in obtaining these data must complete an Agreement for the Use of Confidential Data, specify the reasons for the request, and obtain IRB approval or notice of exemption for their research. Apply for access to these data through the ICPSR Restricted Data Contract Portal, which can be accessed via the study home page.

Inter-university Consortium for Political and Social Research

  1. The data available for download are not weighted, and users will need to weight the data prior to analysis.

  2. To protect respondent confidentiality, variables containing potentially identifying information, including state codes, were recoded in the public use data file. The state code variable and variables that include greater detail on previously held offices, vote percentages, campaign spending, and family income can only be accessed in the restricted-use file. To obtain the restricted-use file, researchers must agree to the terms and conditions of a Restricted Data Use Agreement. In both versions of the data, the principal investigators randomized a small number of survey responses so that it would be impossible to identify, with certainty, an individual legislator by his or her attributes (e.g., party or gender) and unequivocally attribute to that person a particular response.
  3. Since Nebraska has a one-chamber legislature, state legislators from Nebraska were described as serving in the upper chamber in the variable CHAMBER.

  4. With respect to a number of variables, especially those dealing with margins by which candidates won and amounts of money spent on campaigns, it should be remembered that these are survey responses. The principal investigators have not tried to edit them on the basis of plausibility, therefore some responses were very likely in error, as when a respondent claims to have won the general election with 10 percent of the vote. With respect to this variable, also remember that some districts are multi-member, so winners do not always have a majority of the vote. The largest district elects 12 members, and thus a candidate could win with a small percentage of the vote.

  5. The age variable Z26 was removed from the dataset per instructions from the principal investigator.

  6. The formats of several variables were adjusted to fit the values present in these variables.

  7. In variable names, periods were replaced with underscores in order to be compatible with current statistical software packages.

  8. The CASEID variable was created for use with online analysis.



All state legislators in all 50 states serving in February 2002.


40.1 percent



2018-02-15 The citation of this study may have changed due to the new version control system that has been implemented. The previous citation was:
  • Carey, John M., Richard G. Niemi, Lynda W. Powell, and Gary Moncrief. 2002 State Legislative Survey. ICPSR20960-v1. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2008-03-25. http://doi.org/10.3886/ICPSR20960.v1

2008-03-20 ICPSR data undergo a confidentiality review and are altered when necessary to limit the risk of disclosure. ICPSR also routinely creates ready-to-go data files along with setups in the major statistical software formats as well as standard codebooks to accompany the data. In addition to these procedures, ICPSR performed the following processing steps for this data collection:

  • Created online analysis version with question text.

The data contain a weight variable (NEWWTZ) that should be used in analyzing the data. Logistic regression analysis indicated significant differences in response probabilities across a set of individual and contextual variables. The coefficients from the regression were used to estimate the probability that individuals with given characteristics responded to the survey. Respondents were then weighted by a factor proportionate to the inverse of their response probability. The factor was chosen so that the number of respondents in the weighted dataset was the same as that in the unweighted dataset. More information on weighting can be found in the codebook documentation.