You are here

Job Approval Ratings Database

U.S. Officials' Job Approval Ratings (JARs) Database

The U.S. Officials' Job Approval Ratings (JARs) Database is a depository for job approval ratings obtained at the state level for state governors, U.S. senators and U.S. presidents from the mid-1900s to 2009. I host this page on behalf of the three scholars who led the JARs project for many years – Richard Niemi, the principal investigator on the National Science Foundation grant that funded the project, and his co-PIs, Thad Beyle and the late Lee Sigelman. According to Richard Niemi, Thad did the heavy lifting on the project and the text below is taken, with little editing, from Thad’s old web site for the database.

If you use these data, please cite the database as the U.S. Officials' Job Approval Ratings (JARs) Database, compiled by Richard Niemi, Thad Beyle and Lee Sigelman—the shorthand is the JARs Database—and include in your citations a note that the database is currently hosted by Jennifer Jensen and can be found at this web site.  Someday, I (Jennifer) may expand the dataset, but not now.

Richard Niemi is the Don Alonzo Watson Professor Emeritus of Political Science at the University of Rochester. Thad Beyle is professor emeritus at the University of North Carolina at Chapel Hill. Lee Sigelman was Columbian College Distinguished Professor of Political Science at George Washington University.

Funded in part by the National Science Foundation, the JARs Database is drawn from public opinion surveys—typically state-level surveys—that include items pertaining to satisfaction with political officials.

Prior to downloading data, please review our information on question type and rating scales, downloading job approval ratings, and treatment of missing or incomplete data in order to ensure that you are able to take full advantage of the information provided.

Question Type and Rating Scales
As you will find in the codebook, the survey results represent a variety of questions asked of respondents regarding their assessments of officials. With the exception of the senator data, for which question type was limited to one, questions varied from general job performance assessments to assessments of the official's attention to one or more key policy areas. For this reason, two or more records may be identical except for the question type codes. In addition, rating scales vary. Possible responses will range from "good, very good, fair, poor" to "approve, disapprove," and so forth. To make aggregation of the results possible, all responses have been collapsed into "percent positive" and "percent negative."  A "/" indicates the division for positive and negative totals. Listings for these and remaining fields are provided in the codebook.

Downloading Job Approval Ratings
Before downloading JARs on this site, please note the following. Databases have been composed in Microsoft Excel, versions 97 forward, and are transferable into statistical applications such as SPSS, STATA, and SAS. For ease of data manipulation, codes for all three primary databases are consistent with each other--hence, a single codebook for gubernatorial, senatorial and presidential data. Just a reminder if you are unfamiliar with Excel -- Upon opening the Excel files, you may encounter ##### in one or more of the spreadsheet cells. If this occurs, it normally means that the column width is not great enough to show the full cell entry. To widen the column, place your cursor at the top of the column, left click, and drag the edge of the column to the appropriate width.

The "date into field" (DATEIN) column in all datasets is accompanied by 3 separate columns for day (DAYIN), month (MONTHIN), and year (YEARIN); this is true likewise for the "date out of field" (DATEOUT) columns. Some statistical applications "prefer" the 3-column split over the full date style (i.e. 03/34/79). Please be sure to check for proper translation of dates when you open the datasets in the statistical application you choose.

Treatment of Missing or Incomplete Data

As we continue to improve this resource, our goal is to keep missing data to a minimum. However, in some cases full information pertaining to a given survey is not available. Where information is lacking in a record, the cell will be empty. This should translate appropriately into the statistical application. Please also note that for some records the exact date of the survey in question is incomplete. Where this occurs, the expansion of the date code into three separate variables for day, month, and year operates as follows:                                         

Information available (example) Recorded as
Year (1983) and Month (April) 4    1983
Year (1983) 1983





So, when there is just a month and year, it records the date as the first of the relevant month and year. Note that the original variable--DATEIN--does not change; it remains a month and year or just a year. When using the data, please note that column-scale changes in the format of the dates will compromise the integrity of the data due to automatic adjustments made by the Excel software to compensate for "missing" information. For example, where the date is given as Mar-79, a format change for the entire column could result in the date "permanently" reading as 03/01/79, thereby losing the original information noting that the exact date of the survey is not known. It is recommended that any changes to the format of dates be made per cell(s) in order to avoid making unwanted changes.

Data and Codebook: