U.S. Department of Justice Office of Justice Programs Bureau of Justice Statistics Technical Report Assessing the Accuracy of State Prisoner Statistics February 1999, NCJ 173413 By Patrick A. Langan, Ph.D. David Levin BJS Statisticians Included in the wide array of statistics routinely compiled and disseminated by the Bureau of Justice Statistics (BJS) are those describing persons sentenced to State prisons during the year. Prisoner characteristics are available in two different sources: BJS reports that are based on State prison records, and BJS reports that use data from State court records. This report investigates the extent to which prisoner characteristics from one source match those from the other. Results of the investigation indicate a close correspondence between court and prison records. The need for the investigation arises because the court records and the prison records are not from geographically identical places. For example, in 1992 the prison records are from 37 States, and the court records are from a sample of 300 counties scientifically drawn to be nationally representative. State prison records Since 1926 the Federal Government has been regularly compiling national prisoner statistics from records that State prisons voluntarily submit. These statistics give information on persons sentenced to State prisons during the year: their age, race, sex, the length of their prison sentence, and the offense they committed. These statistics serve as one of the Nation's most valuable sources of information on its prison population. They answer such questions as whether blacks are a growing fraction of the persons being sent to prison (they are); whether prison sentences are getting longer (generally they are not); and whether imprisonment of violent offenders accounts for much of the recent increase in the prison population (it does). State court records In 1986 the Federal Government expanded its data gathering to include State courts. The data, voluntarily sent by State courts, serve as the sole source of national information on two major roles of courts: convicting and sentencing criminals. The data are from annual records that State courts keep on each person convicted of a felony offense. The data show how many people are convicted in the year; the crimes they are convicted of; their age, race, and sex; and what sentence the court imposed. Among other things, the data reveal that a substantial number of the violent offenders convicted in State courts each year get probation rather than a prison sentence; that about 50% of convicted felons (a category that includes both violent and nonviolent criminals) are sentenced to State prison; and that, compared to women, men generally receive longer prison sentences, longer jail sentences, and longer probation sentences. State court records compared to State prison records In the State court records, convicted felons who received a prison sentence are distinguishable from those who received other sentences. Consequently, characteristics of persons sent to prison in a particular year according to State court records can be compared to characteristics of persons who entered prison that same year according to State prison records. The comparison is instructive. To the extent that characteristics of one match characteristics of the other, confidence is enhanced in each as a reliable source of national records on convicted offenders. The comparison covers offenders sentenced to State prison in 1992. In general, results indicate a close correspondence between the two sources. What State prison records indicate about persons sent to prison generally mirrors what State court records show. Results Conviction offense State prison records and State court records both show that, among persons sentenced to State prisons in 1992, robbery was the conviction offense for 10% and larceny was for 11% (table 1). Percentages differed for most other offenses, but usually by small amounts. Table 1. Court records compared to prison records: Percent of State prison sentences, by offender characteristics, 1992 Percent of persons sentenced to prison in 1992 according to -- ------------------------------------- Characteristic Court records Prison records -------------------------------------------------------- Sex 100 % 100 % Male 91 91 Female 9 9 Number of cases 349,722 288,211 Race 100 % 100 % White 45 45 Black 54 54 Other 1 1 Number of cases 264,742 288,211 Age 100 % 100 % 13-17 1 1 18-24 31 34 25-29 22 22 30-34 20 18 35-44 20 18 45-54 5 5 55 or older 1 1 Mean 30 yrs 30 yrs Number of cases 322,164 288,211 Most serious conviction offense All offenses 100 % 100 % Violent offenses 26 % 29 % Murder a/ 3 2 Rape 4 2 Robbery 10 10 Assault 7 8 Other violent b/ 2 6 Property offenses 31 % 27 % Burglary 15 13 Larceny c/ 11 11 Fraud d/ 5 3 Drug offenses 29 % 26 % Possession 9 7 Trafficking 20 19 Weapons offenses 3 % 2 % Other offenses e/ 11 % 16 % Number of cases 390,917 288,211 Note: The number of court records is estimated from the sample, and the number of prisoners is that submitted by 37 States. Detail may not sum to total because of rounding. a/Includes nonnegligent manslaughter. b/Includes offenses such as negligent manslaughter, sexual assault, and kidnaping. c/Includes motor vehicle theft. d/Includes forgery and embezzlement. e/Composed of nonviolent offenses such as receiving stolen property. Sex of prisoners State prison records and State court records both show that males were 91% of the persons sentenced to prison in 1992. Court records show that burglars were 16% of all male offenders sentenced to prison, and prison records show 14%, a difference of 2 percentage points (table 2). Differences of this size or less characterized most other offenses, for males and females alike (table 2). Table 2. Court records compared to prison records: Percent of State prison sentences, by sex and most serious conviction offense, 1992 Male Female Most serious conviction offense Court records Prison records Court records Prison records All offenses 100 % 100 % 100 % 100 % Violent offenses 26 % 30 % 15 % 16 % Murdera 3 3 3 2 Rape 4 3 1 0 Robbery 10 10 5 5 Assault 7 8 5 5 Other violentb 2 6 1 3 Property offenses 31 % 27 % 39 % 34 % Burglary 16 14 6 5 Larcenyc 11 10 17 17 Fraudd 4 3 16 12 Drug offenses 29 % 25 % 36 % 32 % Possession 9 7 14 8 Trafficking 20 18 22 24 Weapons offenses 3 % 3 % 1 % 1 % Other offensese 11 % 15 % 9 % Number of cases 319,063 261,188 30,660 24,640 Note: Detail may not sum to total because of rounding. a/Includes nonnegligent manslaughter. b/Includes offenses such as negligent manslaughter, sexual assault, and kidnaping. c/Includes motor vehicle theft. d/Includes forgery and embezzlement. e/Composed of nonviolent offenses such as receiving stolen property. Race of prisoners Both sources show that, among all persons sentenced to State prisons in 1992, 45% were white, 54% were black, and 1% were other races (table 1). Among whites, 7% were persons sentenced for robbery according to both prison and court records (table 3). Among blacks, 12% were robbers according to court records, nearly identical to what prison records show (13%) (table 3). Such small differences between court and prison records characterized most offenses. Table 3. Court records compared to prison records: Percent of State prison sentences, by race and most serious conviction offense, 1992 White Black Other Most serious -------------------------------- ------------------------------- ------------------------------- conviction offense Court records Prison records Court records Prison records Court records Prison records All offenses 100 % 100 % 100 % 100 % 100 % 100 % Violent offenses 23 % 28 % 25 % 29 % 46 % 41 % Murder a/ 3 3 3 3 4 5 Rape 5 3 2 2 18 4 Robbery 7 7 12 13 12 11 Assault 6 7 7 8 11 12 Other violent b/ 3 9 1 4 1 9 Property offenses 39 % 34 % 27 % 24 % 32 % 28 % Burglary 19 17 13 11 19 15 Larceny c/ 13 12 11 10 11 10 Fraud d/ 7 5 4 3 2 3 Drug offenses 23 % 18 % 36 % 30 % 8 % 12 % Possession 8 4 12 9 4 3 Trafficking 15 14 25 22 4 10 Weapons offenses 2 % 2 % 3 % 3 % 2 % 2 % Other offenses e/ 14 % 19 % 8 % 14 % 13 % 19 % Number of cases 118,326 117,936 143,844 143,168 2,572 2,542 a/Includes nonnegligent manslaughter. b/includes offenses such as negligent manslaughter, sexual assault, and kidnaping. c/Includes motor vehicle theft. d/Includes forgery and embezzlement. e/Composed of nonviolent offenses such as receiving stolen property. Age of prisoners State prison records and State court records both show that the average age of prisoners sentenced in 1992 was 30 years (table 1). Nearly a quarter of all persons sentenced in 1992 were between the ages of 25 and 29 years. The percentages that were in other age categories generally differed between the two sources, but usually by no more than 2 percentage points. According to both sets of records the average age of persons sentenced for murder in 1992 was 28 (table 4). For most offense categories, State prison records and State court records agreed on the average age of sentenced persons. Where differences existed, they were usually no more than 1 year. Table 4. Court records compared to prison records: Mean ages of State prisoners, by most serious conviction offense, 1992 Average age of persons sentenced to prison in 1992 according to -- ---------------------------------- Most serious Most serious conviction offense Court records Prison records ----------------------------------------------------------- All offenses 30 yrs 30 yrs 30 yrs Violent offenses 29 yrs 29 yrs Murder a/ 28 28 Rape 33 33 Robbery 26 26 Assault 29 29 Other violent b/ 34 33 Property offenses 30 yrs 29 yrs Burglary 28 28 Larceny c/ 30 30 Fraud d/ 32 32 Drug offenses 30 yrs 30 yrs Possession 31 30 Trafficking 30 30 Weapons offenses 28 yrs 29 yrs Other offenses e/ 32 yrs 29 yrs Number of cases 322,165 285,921 Note: Means exclude sentences to death or to life. a/Includes nonnegligent manslaughter. b/Includes offenses such as negligent manslaughter, sexual assault, and kidnaping. c/Includes motor vehicle theft. d/Includes forgery and embezzlement. e/Composed of nonviolent offenses such as receiving stolen property. Prison sentence length The biggest difference between the two sets of records is prisoner sentence length. Court-recorded sentence lengths averaged 6 months longer than prison-recorded sentence lengths. For example, the average prison sentence length for rape was nearly 14 years 8 months according to court records and 14 years 2 months according to prison records (table 5). Court-recorded sentence lengths were longer than prison recorded sentence lengths for males, females, whites, and blacks (table 6). Table 5. Court records compared to prison records: Mean State prison sentence lengths, by most serious conviction offense, 1992 Average prison sentence length (in months) for persons sentenced to prison in 1992 according to -- Most serious ---------------------------------- conviction offense Court records Prison records --------------------------------------------------------- All offenses 85 mo 79 mo Violent offenses 130 mo 117 mo Murder a/ 253 265 Rape 170 176 Robbery 119 107 Assault 93 82 Other violent b/ 96 112 Property offenses 71 mo 66 mo Burglary 80 77 Larceny c/ 58 53 Fraudd 69 62 Drug offenses 72 mo 71 mo Possession 62 71 Trafficking 75 72 Weapons offenses 54 mo 51 mo Other offensese 62 mo 52 mo Number of cases 349,434 293,162 Note: Means exclude sentences to death or to life in prison. Ohio data from the NCRP are excluded due to a coding error. a/Includes nonnegligent manslaughter. b/Includes offenses such as negligent manslaughter, sexual assault, and kidnaping. c/Includes motor vehicle theft. Table 6. Court records compared to prison records: Mean sentence lengths for persons sentenced to prison in 1992, by sex, race, and most serious conviction offense Average sentence length (in months) for persons sentenced to prison Sex Race ----------------------------------- --------------------------------- Male Female White Black Black ---------------- ---------------- ---------------- -------------- Most serious Court Prison Court Prison Court Prison Court Prison conviction offense records records records records records records records records ----------------------------------------------------------------------------------------------------- All offenses 87 mo 80 mo 59 mo 59 mo 79 mo 72 mo 89 mo 83 mo Violent offenses 129 mo 118 mo 96 mo 92 mo 125 mo 111 mo 128 mo 122 mo Murder a/ 257 272 191 189 231 242 248 277 Rape 167 176 110 202 167 168 166 191 Robbery 117 108 77 81 111 98 117 112 Assault 95 83 73 66 87 74 93 87 Other violent b/ 99 113 73 92 102 111 87 119 Property offenses 74 mo 67 mo 47 mo 51 mo 72 mo 64 mo 72 mo 66 mo Burglary 81 77 50 58 85 74 75 79 Larceny c/ 59 54 40 74 45 57 51 60 53 Fraud d/ 83 64 53 57 65 63 94 62 Drug offenses 74 mo 73 mo 59 mo 60 mo 63 mo 60 mo 76 mo 77 mo Possession 64 72 50 58 51 58 69 73 Trafficking 78 73 64 61 68 61 79 78 Weapons offenses 55 mo 52 mo 44 mo 41 mo 54 mo 44 mo 51 mo 54 mo Other offenses e/ 63 mo 53 mo 53 mo 45 mo 46 mo 48 mo 97 mo 58 mo Number of cases 285,740 269,106 26,343 23,687 106,064 121,069 126,721 146,716 Note: Means exclude sentences to death or life in prison. Ohio data from the NCRP are excluded due to a coding error. a/Includes nonnegligent manslaughter. b/Includes offenses such as negligent manslaughter, sexual assault, and kidnaping. c/Includes motor vehicle theft. d/Includes forgery and embezzlement. e/Composed of nonviolent offenses such as receiving stolen property. The 6-month difference between court records and prison records found for 1992 is not found for other years. In 1990 the court records were 3 months longer than the prison records: 6 years 9 months in court records versus 6 years 6 months in prison records. In 1994, the prison records were longer than the court records by 2 months: 6 years 7 months in prison records versus 6 years 5 months in court records. Number of persons sentenced to prison in 1992 Using the most comparable figures, 1992 court records indicate 349,400 persons were sent to prison with sentences greater than a year; prison records indicate 293,200 persons (table 5), difference of 16%. Nearly all of the difference is explained by the fact that 13 States are not included in the data compiled from prison records. Had all 50 States been included in the prison records, it is estimated that the prison records would have indicated 338,600 prison sentences, leaving a difference of 4%. Methodology Significance tests All comparisons discussed in this report were statistically significant at the 95% confidence level. 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. Data sources Data for this report were drawn from two BJS statistical programs: the National Corrections Reporting Program (NCRP) based on prison records, and the National Judicial Reporting Program (NJRP) based on court records. Data in this report shown under the heading prison records in tables 1 through 4 are from the Bureau of Justice Statistics (BJS) statistical report National Corrections Reporting Program, 1992 (NCJ 145862, October 1994, calculated from tables 1-4, 1-5, 1-6, 1-7, 1- 9). Sentence lengths in tables 5 and 6 under the heading prison records were produced specifically for this report based on analysis of the 1992 NCRP database. Data shown under the heading court records in tables 1 through 6 are tabulated from the 1992 NJRP data-base. For more information on the 1992 NJRP, see State Court Sentencing of Convicted Felons, 1992, BJS Report, NCJ 152696, May 1996. Average sentence length for 1990 and 1994 were not based on previously published figures, but on new analyses of the court and prison databases for these years. For published figures and more information on 1990 and 1994 data, see National Judicial Reporting Program, 1990, BJS Report, NCJ 145323; State Court Sentencing of Convicted Felons, 1994, BJS Report, NCJ 164614, March 1998; National Corrections Reporting Program, 1990, BJS Report, NCJ 141879; National Corrections Reporting Program, 1994, BJS Report, forthcoming. Comparability Because of numerous differences between procedures used to collect NJRP and NCRP data, the two will not give identical results: * The NJRP data are based on a sample survey of persons sentenced to prison for a felony offense in State courts in 1992. The sample consisted of 300 counties selected to be nationally representative. * The NCRP data pertain to persons entering State prison in 1992 directly from a sentence by a court. The data were volunteered by 38 states. * NJRP data are limited to persons convicted of a felony, whereas NCRP data pertain to all persons entering State prison, whether convicted of a felony or a misdemeanor. The two data sources use similar, but not identical, definitions of the term "most serious conviction offense." For the NCRP, if the entering prisoner was convicted of more than one offense, the prisoner's "most serious conviction offense" was the offense with the longest sentence. For the NJRP, if the felon was convicted of more than one felony offense, the offense designation for the felon 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. Sentence length Information on average sentence lengths given in tables 5 and 6 of this report were not obtained from existing BJS publications of 1992 data from the NCRP and the NJRP. The reason is that these publications define aver-age sentence length differently: * Persons receiving consecutive sentences are treated one way in the 1992 NCRP publication's measure of average sentence length and a different way in the 1992 NJRP's measure. In the NJRP publication, the sentences for the different offenses are summed. For example, the NJRP measure treats a person receiving a 10-year sentence for rape and a consecutive 5-year sentence for robbery as a 15- year sentence. By contrast, the person is treated as a 10-year sentence in the NCRP measure because the NCRP measure counts just the longest sentence. * Unlike the 1992 NJRP publication, the 1992 NCRP publication does not include probationers and parolees returned to prison with new sentences. * Unlike the 1992 NJRP publication, the 1992 NCRP publication excludes prisoners with sentences of a year or less. To make average sentence length (shown in tables 5 and 6) more comparable between the NCRP and the NJRP, 1992 prison sentence length data were subjected to an analysis in which average sentence length was defined the same way for this report: * Maximum sentence length for consecutive sentences was defined as the sum of the consecutive sentences a person receives. * Probationers and parolees returned to prison with new sentences were included. * Only sentences over a year were included. Using a common definition, the NCRP gave an estimated average sentence length of 6 years 7 months (table 5) and the NJRP gave 7 years 1 month (table 5), a difference of 6 months. When the common definition was not used, the two differed by 10 months (5 years 7 months according to the 1992 NCRP publication versus 6 years 7 months according to the 1992 NJRP publication). Therefore, using a common definition (table 5) narrowed the difference. Even with the use of a common definition, differences remain between the NCRP and the NJRP. For example, California (which has relatively short sentences) accounts for a greater fraction of the NCRP cases than the NJRP cases. The 6-month difference (table 5) between the NCRP and the NJRP would be reduced to 4 months if California accounted for the same fraction of the NCRP as the NJRP. No doubt other differences exist between the two that affect comparisons of average sentence length. Estimate of the number of persons sentenced to prison in 1992 according to prisoner data The 293,162 figure for the 37 States plus an estimate for the 13 missing States forms a national estimate of persons admitted. To obtain the national estimate, a prisoner database was located that gave an aggregate count of the number of persons sentenced to prison in 1992 for each of the 50 States (Correctional Populations in the United States, 1992 , NCJ 146413, January 1995, tables 5.10a, 5.16 ). According to the database, the 13 States accounted for 12% of prison admissions nationwide in 1992. Using the 12% figure as a basis for estimation, it is estimated that 338,000 persons were sentenced to prison nationwide in 1992. ****************************************************************************** The Bureau of Justice Statistics is the statistical agency of the U.S. Department of Justice. Jan M. Chaiken, Ph.D., is director. This report belongs to the BJS Technical Report series. It was written by Patrick A. Langan, Ph.D., and David Levin, BJS statisticians. The report was produced by Yvonne Boston and Tom Hester. Marilyn Marbrook supervised final production and printing. February 1999, NCJ 173413 ****************************************************************************** Readers who want further information about the National Judicial Reporting Program and the National Corrections Reporting Program should access the BJS Internet Web site: http://www.ojp.usdoj.gov/bjs/ Data presented in this report may be obtained from the National Archive of Criminal Justice Data at the University of Michigan, 1-800-999-0960. The archive may be accessed through the BJS Web site. When at the archive site, search for data sets ICPSR 6038, 6509, 6855, 6141, 6400, and 6681. ****************************************************************************** (End of file) 3/10/99 yb