Overview
During the end of the National Survey on Drug Use and Health (NSDUH) questionnaire, the interviewer asked each respondent (R) a series of questions to obtain details about the R's employment, including the type of business or industry and the main duties done in the job. In prior years, specially trained coders at RTI used this information to assign an industry and an occupation code for each respondent. Beginning in 2003, the work of assigning codes was completed by the National Processing Center (NPC) of the Census Bureau through an InterAgency Agreement between the Substance Abuse and Mental Health Services Administsration (SAMHSA) and the Bureau of the Census.
Process
RTI sent compiled industry and occupation questionnaire data to the NPC in eight separate deliveries, or waves. Each quarter, RTI sent two electronic files, the first with data from the first 4 or 5 weeks of the quarter and the second with all remaining data for the quarter. NPC coders determined both an industry and an occupation code for each record; each code was determined at the 4-digit level of detail. Coders used the Census Bureau's 2003 standard industry and occupation classification coding system to assign the codes, meaning they used the 2002 North American Industry Classification System (NAICS) for industry coding and the 2000 Standard Occupational Classification (SOC) system for occupation coding.
Two different coders assigned the codes for each record. During the second verification coding, if the first and second codes did not agree, the second coder reconciled the discrepancy and assigned the final code. In some instances, cases were referred to a third party for assignment of a final code. The NPC then returned the codes to RTI for inclusion in the final NSDUH results.
As a quality control measure, the NPC provided feedback regularly on production and error rates to coding staff. To improve the quality of the data collected, RTI used NPC data to learn of situations where coders had trouble coding three or more cases completed by a particular interviewer. RTI supervisors used this information to retrain those specific interviewers.
Results
The NPC sent SAMHSA progress reports including production rates per hour and numbers and percentage of codes requiring reconciliation separately for industry and occupation codes. Based on those reports, Tables F.1 through F.3 display the production information for the NPC coding process. Table F.2 contains the coding production result by wave, while Table F.3 shows the production rates for each wave.
Completed Interviews by Wave | |||||||||
Wave 1 | Wave 2 | Wave 3 | Wave 4 | Wave 5 | Wave 6 | Wave 7 | Wave 8 | Total | |
---|---|---|---|---|---|---|---|---|---|
Completed Interviews | 13,571 | 3,099 | 14,751 | 2,565 | 14,503 | 3,381 | 14,503 | 2,271 | 68,644 * |
Interviews with I&O data | 8,739 | 2,172 | 9,418 | 1,823 | 9,414 | 2,287 | 8,949 | 1,618 | 44,420 |
* Completed interviews delivered to Census throughout the year have not gone through the data cleaning and editing process, thus the total is higher than the final number of completed interviews for the year. |
Production Results by Wave | ||||||||||||||||||
Wave 1 | Wave 2 | Wave 3 | Wave 4 | Wave 5 | Wave 6 | Wave 7 | Wave 8 | Total | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Count | % | Count | % | Count | % | Count | % | Count | % | Count | % | Count | % | Count | % | Count | % | |
Total Coded | 8,739 | 100.0 | 2,172 | 100.0 | 9,418 | 100.0 | 1,823 | 100.0 | 9,414 | 100.0 | 2,287 | 100.0 | 8,949 | 100.0 | 1,618 | 100.0 | 44,420 | 100.0 |
Total Verified | 8,739 | 100.0 | 2,172 | 100.0 | 9,418 | 100.0 | 1,823 | 100.0 | 9,414 | 100.0 | 2,287 | 100.0 | 8,949 | 100.0 | 1,618 | 100.0 | 44,420 | 100.0 |
Industry Codes Requiring Reconciliation | 525 | 6.0 | 155 | 7.1 | 610 | 6.5 | 107 | 5.9 | 627 | 6.7 | 184 | 8.0 | 592 | 6.6 | 85 | 5.3 | 2,885 | 6.5 |
Occupation Codes Requiring Reconciliation | 866 | 9.9 | 231 | 10.6 | 938 | 10.0 | 176 | 9.7 | 875 | 9.3 | 241 | 10.5 | 812 | 9.1 | 131 | 8.1 | 4,270 | 9.6 |
Total Referred Cases | 1,077 | 12.3 | 288 | 13.3 | 1,178 | 12.5 | 218 | 12.0 | 1,100 | 11.7 | 308 | 13.5 | 1,048 | 11.7 | 157 | 9.7 | 5,374 | 12.1 |
Total Coded: Codes assigned by first coder. Total Verified: Codes assigned and confirmed by second coder. Reconciled Codes: First and second codes did not match. Second coder reconciled and assigned final code. Total Referred Cases: Second coder could not reconcile case. Final code assigned by third-party coder using additional resources (Internet, Dun and Bradstreet) to resolve discrepancy. |
Production Rates by Wave | ||||||||||||||||||
Number per Hour | Average Number per Hour | |||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Wave 1 | Wave 2 | Wave 3 | Wave 4 | Wave 5 | Wave 6 | Wave 7 | Wave 8 | Total | ||||||||||
Coding Production Rates | 76.65 | 60.75 | 84.08 | 87.86 | 89.87 | 94.30 | 99.70 | 107.9 | 87.6 | |||||||||
Coding Verification Rates | 82.83 | 54.64 | 84.46 | 91.15 | 93.21 | 79.54 | 123.00 | 92.5 | 87.7 | |||||||||
Problem Referral Rates | 24.47 | 19.86 | 22.23 | 21.80 | 24.18 | 19.87 | 33.80 | 22.4 | 23.6 | |||||||||
Coding: Codes assigned by first coder. Verification: Codes assigned and confirmed by second coder. Referred Cases: Second coder could not reconcile case. Final code assigned by third-party coder using additional resources (Internet, Dun and Bradstreet) to resolve discrepancy. |