May 5, 2008

IX. Methodology

Prison Admissions for Drug Offenses

State-by-state data on prison admissions from the National Corrections Reporting Program (NCRP) for 2003[130] (the latest year with available data) were used to compute all the statistics in this report related to admissions to prison. In 2003, 35 states participated in the reporting of admission data to NCRP. The data were cleaned as follows before any analysis:

  • The admission database had a total of 559,526 cases. Cases were attributed to states first using the state of jurisdiction. Where the state of jurisdiction was unknown (136,573 cases; 24.4 percent) the county where sentence was imposed was used to assign the case to a state. A total of 1,546 cases (0.28 percent) were excluded when the case could not be attributed to a participating state. These include cases from: "Shared Jurisdiction" (2), "Federal Prison System" (195), "State Other than Reporting State" (62), "Unknown State" (777), and cases from non-participating states: Arizona (7), Arkansas (3), Connecticut (134), Delaware (1), Idaho (5), Kansas (5), Massachusetts (1), Montana (3), New Mexico (7), Ohio (1), and Vermont (343). The remaining 557,980 cases were from participating states.
  • Only new court commitments were considered for analysis, representing 357,114 cases (64.0 percent), including: "New Court Commitment," "Parole Revocation with a New Sentence," "Mandatory Parole Release with a New Sentence," "Suspended Sentence Imposed," and "Probation Revocation with a New Sentence." A total of 200,866 cases (36.0 percent) from the participating states that were not new court commitments (for example, "Parole Revocation with No New Sentence") were excluded.
  • Alaska did not report any new commitments in the data it submitted to the NCRP. We have therefore not included Alaska in this report and in the data analyses that we have done. We refer in the report to the 34 NCRP reporting states for simplicity's sake, even though there were technically 35.
  • Out of the 357,114 new commitments, 3,397 concerned children. These cases were excluded. All the figures used in this report are provided for adults only. Finally, 15 cases were excluded because they had incomplete data on demographics (gender or age). The final number of new prison commitments considered for analysis therefore was 353,702 cases.
  • From this admissions database, drug offenders were selected for analysis. Drug offenders identified for this report were defined as new prison admissions in the NCRP database for which the most serious offense (that is, the offense that carried the longest sentence-variable V26 in the database) was a drug-related offense. These included offense codes 340 to 450.[131] Only those cases were selected for analysis. Offenders who were admitted to prison with more serious offenses-such as murder-in addition to drug offenses are not included among drug offenders in our analyses. The final database for drug offender admissions included 111,247 cases.

Race and Prison Admissions

The NCRP database treats race and ethnicity separately, with one variable for race and one variable for ethnicity. We only used race in our analyses of prison admissions because of the large amount of missing data on ethnicity (23.8 percent cases with missing ethnicity). We recoded race into three categories: whites, blacks, and "other."  The "other" category includes cases for which the race was Indian American, Asian, Native Hawaiian, other, unknown, or blank cases. Each racial category can include Hispanics and non-Hispanics.

Rates of Admission

We used US Census Bureau projected population data for 2003[132] to compute rates of admissions in this report. Rates were calculated per 100,000 adult (age 18 years or older) residents of the designated race and gender groups in each state for which we had drug admissions data. Races were recoded into three categories: white, black, and other, each of which may include Hispanics and non-Hispanics.

Total Figures

In the charts and figures in this report, unless otherwise specified, "total" frequencies and rates of admission were calculated on the basis of the total number of new drug admissions and the total populations for the 34 reporting states combined. The "totals" do not reflect averages.

Limitations

External Validity

The National Corrections Reporting Program does not provide data for all 50 states.  In 2003, 35 states participated in the program (the number of participating states varies each year). The state of Alaska did not report any new prison admissions and we therefore excluded it from our analysis. The analysis presented in this report therefore is only valid for the 34 reporting states with new prison admissions. How the non-participating states differ in terms of drug admissions and racial disparity is unknown.

 

Reporting of the Cases

The reliability of the data contained in the NCRP database cannot be assessed.The NCRP database is based on a structured questionnaire completed annually on the basis of official prison records of prisoner population movement. After the questionnaires are processed by the Census Bureau, state tallies are sent to state officials for verification and comment. Limitations and information on data processing are provided in the codebook associated with the data.[133]

Duplication

The NCRP database considers every prison admission as a new case. It is therefore possible that the same individual is represented more than once in the 2003 database if he or she was admitted more than once over the course of that year. However, since offenders are rarely sent to prison unless they have a sentence of one year or more, the possible number of duplicates among new prison admissions is unlikely to affect the analyses presented in this report.

 

Missing Data on Race

Among the new drug offender admissions in the NCRP data base, 12.3 percent listed race as unknown, "other" or left the variable blank. There is no way of knowing whether or to what extent the results of our analyses would change had there been more complete reporting on race in the NCRP database.    

[130] Bureau of Justice Statistics (BJS), "National Corrections Reporting Program Series," 2003, http://www.icpsr.umich.edu/cocoon/ICPSR/SERIES/00038.xml (accessed December 1, 2007).

[131] BJS, "Offense Code for the National Correction Program," ICPSR 20741 Codebook.

[132]US Census Bureau, "Population Estimates," March 2008, http://www.census.gov/popest/estimates.php (accessed April 16, 2008).

[133] BJS, ICPSR 20741 Codebook.