Schaffer Online Library of Drug Policy Sign the Resolution
Contents | Feedback | Search
DRCNet Home
| Join DRCNet
DRCNet Library | Schaffer Library | Government Publications


Sourcebook of Criminal Justice Statistics

Appendix 13

National Judicial Reporting Program

Survey sampling procedures and definitions of terms

Note: The following information has been excerpted from the U.S. Department of Justice, Bureau of Justice Statistics, Felony Sentences in State Courts, 1994, Bulletin NCJ-163391 (Washington, DC: U.S. Department of Justice, January 1997), pp. 10-12. Non-substantive editorial adaptations have been made.

Survey sampling procedures

A sample of 300 counties was drawn for the 1994 National Judicial Reporting Program (NJRP) survey. Every county in the Nation had a nonzero probability of being in the sample. In general, the more felony cases a county had, the more likely it was to be in the sample.

The survey used a two-stage, stratified cluster sampling design. In the first stage the Nation's 3,109 counties or county equivalents were divided into 8 strata. Strata 1 and 2 consisted solely of the 75 largest counties in the United States as defined by the 1985 resident population. Strata 3 through 8 consisted of the remaining 3,034 counties.

Because the 75 largest counties account for a disproportionately large amount of serious crime in the Nation, they were given a greater chance of being selected than the remaining counties.

Stratum 1 consisted of both the 19 counties with the largest number of felony convictions in 1985, and 12 counties whose participation in the survey had been prearranged. Every county in stratum 1 was selected for the sample.

Stratum 2 consisted of the 44 most populous counties that were not in stratum 1. The 44 were ordered by their number of felony convictions in 1985, and then approximately every other county was selected. Stratum 2 thus contributed 23 counties to the sample. Altogether, 54 of the 75 largest counties were sampled. Data on 1985 felony convictions were obtained from a mail survey described in State Felony Courts and Felony Laws (NCJ-106273) and Census of State Felony Courts, 1985 (ICPSR 8667). The 54 sampled counties in the 1994 NJRP survey were the same 54 as in the 1986, 1988, 1990, and 1992 surveys.

The 3,034 counties not among the 75 largest were placed into 6 strata defined by the total number of felony convictions in 1985 and then arrayed within stratum by region, and within region from largest to smallest on felony conviction totals.

The final sample included 246 counties from among the 3,034 counties outside the 75 largest. One county did not participate in the 1994 survey and was subsequently replaced by another county in the same stratum. Case-level data were successfully obtained on convicted felons sentenced in 1994 from these 300 sampled counties. One of the 300 counties in the 1994 survey had no felony convictions during the survey period.

The 60 sampled counties in strata 1 and 3 were self-representing only, and their sampled cases therefore had a first-stage sampling weight of 1. The remaining 240 counties sampled from strata 2 and 4 through 8 were selected to represent their respective strata so that the felony conviction cases sampled had first-stage weights greater than 1.

At the second stage of sampling, a systematic sample of felons sentenced for murder/nonnegligent manslaughter, rape, robbery, aggravated assault, burglary, felony larceny/motor vehicle theft, fraud/forgery/embezzlement, drug trafficking, drug possession, weapons offenses, and other offenses was selected from each county's official records. The sample for the 1994 survey numbered 85,191 cases. Of these, 59,152 cases were in the 75 largest counties.

Rates at which cases were sampled varied by stratum and crime type. In smaller counties every felony case was selected. In larger counties all murder cases and rape cases were typically included, but other offense categories were sampled.

The survey targeted and recorded initial sentences imposed in 1994. If a sentence was imposed on one date and then modified at a later date, the revision was ignored. The survey recorded sentences that were actually executed and excluded suspended sentences.

Because the year of conviction was not a defining characteristic, some cases in the sample were of persons convicted before 1994, but not sentenced until 1994.

In a few counties, where it was impractical to target sentences in 1994, the target was felons convicted in 1994. Therefore, in some of the cases the data relate to sentences imposed after 1994.

Sources of data

For 85% of the 300 counties sampled for the 1994 survey, NJRP data were obtained directly from the State courts. Other sources included prosecutor offices, sentencing commissions, and statistical agencies. Individual-level NJRP records were obtained through a variety of collection methods, including magnetic tape (64% of the counties) and field collection (9% of the counties). Data on other cases were obtained from photocopies of official documents and survey questionnaires completed by court officials (27% of the counties).

Data collection was performed by the U.S. Bureau of the Census.

Sampling error

NJRP data were obtained from a sample and not from a complete enumeration. Consequently, they are subject to sampling error. A standard error, which is a measure of sampling error, is associated with each number reported. In general, if the difference between two numbers is at least twice the standard error of that difference, there is at least 95% confidence that the two numbers do in fact differ; that is, the apparent difference is not simply the result of surveying a sample rather than the entire population.

National estimates of the number of convictions for individual crime categories and for the aggregate total had a coefficient of variation of 3%.

Crime definitions

Before the sample was drawn, each felon sentenced in the sampled counties in 1994 was placed into 1 of the 11 offense categories identified above. If the felon was convicted of more than one felony offense, the offense category was the most serious offense. The hierarchy from most to least serious offense was murder, rape, robbery, aggravated assault, burglary, drug trafficking, weapons, forgery/fraud/embezzlement, larceny/motor vehicle theft, drug possession, and all other felonies. The hierarchy was determined from an analysis of two factors that reflect how seriously the justice system treats different offenses: the sentence length imposed and the time actually served in prison before release. In general, the higher the offense is in the hierarchy, the more serious it is in terms of the two factors.

Sample selection procedures gave each sentenced felon a single chance to be in the sample. However, felons who appeared in court on more than 1 day for different offenses and received a sentence at each reappearance had more than a single chance.

At the data analysis stage, cases were aggregated according to their offense designation at time of sampling, with the single exception of "other violent." "Other violent" is a category shown in the tables, but it was not a category at sampling. The "other violent" category was formed from the sampling category "other felonies." That is, after sampling, sampled cases designated "other felonies" were coded either "violent," "nonviolent," or "not ascertained," based on data available. Cases coded "not ascertained" were rare. For data analysis purposes, cases coded "other violent" were removed from the "other felonies" category and shown separately in the tables. The 11 original offense categories, and "other violent" are defined as follows:

Sourcebook of Criminal Justice Statistics