A proposed hybrid sampling frame for the National Survey on Drug Use and Health final report
CONCLUSION: With proper training and monitoring, the hybrid sampling frame can be implemented in a way that reduces survey costs while maintaining the high scientific standards of NSDUH.
| Main Authors: | , , , |
|---|---|
| Corporate Authors: | , |
| Format: | eBook |
| Language: | English |
| Published: |
Rockville (MD)
Substance Abuse and Mental Health Services Administration (US)
March 2012, 2012
|
| Series: | CBHSQ methodology report
|
| Subjects: | |
| Online Access: | |
| Collection: | National Center for Biotechnology Information - Collection details see MPG.ReNa |
| Summary: | CONCLUSION: With proper training and monitoring, the hybrid sampling frame can be implemented in a way that reduces survey costs while maintaining the high scientific standards of NSDUH. Section 4.2 provides further details about the experimental design and the results of the MLFS II.3. To ensure correct implementation of the CHUM procedure, field staff must receive adequate training. The at-home training combined with in-person training that was used on the MLFS II was generally effective. (see Section 4.2 for further details); however, during fieldwork, FIs reported difficulty with several concepts such as performing the CHUM at apartments and trailer parks and knowing when to contact field support for assistance. As a result, improved training procedures and materials are needed.4. After training, FIs must be monitored in the field through the use of seeding and other techniques to ensure they are correctly implementing the CHUM procedure. They must also be provided with field support to answer questions that arise while implementing the CHUM procedure in the field (see Section 4.2 for further discussion).With proper training and monitoring, the hybrid sampling frame can be implemented in a way that reduces survey costs while maintaining the high scientific standards of the NSDUH. Further efficiencies can be gained by developing techniques that accurately allocate segments with low ABS coverage (e.g., segments with high concentrations of GQs) to the FE frame, and by continuing to explore sources of supplemental addresses BACKGROUND: Researchers view address-based sampling (ABS) as a potential remedy for the rising costs associated with in-person surveys of the U.S. general population. The Substance Abuse and Mental Health Services Administration considered ABS methodology for the National Survey on Drug Use and Health (NSDUH) by using a hybrid sampling frame. The NSDUH sampling frame has traditionally relied on field enumeration (FE) to identify dwelling units (DUs) that are eligible for selection. The proposed hybrid sampling frame would use ABS as a frame for the vast majority of NSDUH segments but would still rely on FE and the half-open interval procedure for the remaining (primarily rural) segments where ABS coverage is low. The Check for Housing Units Missed (CHUM) procedure would be used to cover missed DUs in ABS segments. METHOD: This report summarizes research to evaluate the coverage, cost, and implementation of the proposed hybrid sampling frame for NSDUH. RESULTS: Theoretically, the hybrid frame provides 100 percent coverage of the target population; however, ABS coverage of group quarters (GQs) is problematic. Without a frame supplementation procedure like the CHUM, geocoding error can lead to over-coverage error and under-coverage error of an ABS frame. Geocoding is problematic in rural areas and in area segments comprised of smaller levels of census geography. The cost savings afforded by the hybrid frame depend on how many segments are assigned to ABS, and as stated, there are problems associated with ABS for rural areas and GQs. ABS frames supplemented with the CHUM procedure provide 100 percent coverage of the target population when the CHUM is implemented correctly; however, the challenge remains with ensuring that field staff are adequately trained to identify when CHUM procedures are necessary. It is more problematic in rural areas than urban areas, and for area segments comprised of smaller levels of census geography (see Section 5.3 for further details). COST SAVINGS OF THE HYBRID FRAME: 1. The cost savings afforded by the hybrid frame depend on how many segments are assigned to ABS. In general, the lower the ABS coverage threshold, the more segments will be allocated to ABS and the higher the cost savings. However, because the CHUM procedure is designed to supplement areas with adequate ABS coverage, it is more efficient to allocate segments with very low ABS coverage to the FE frame (see Section 4.1.2 for further details).2. Cost savings are also a function of how well ABS coverage is predicted at the segment level. Inefficiencies arise when segments are allocated to FE that should be allocated to ABS and when segments are allocated to ABS that should be allocated to FE (because the high reliance on the CHUM procedure mitigates the cost savings of ABS). The target population for the National Survey on Drug Use and Health (NSDUH) is the civilian, noninstitutional population of the United States aged 12 or older. Currently, the NSDUH sampling frame relies on field enumeration (FE) to identify dwelling units (DUs) that are eligible for selection. The half-open interval (HOI) frame-linking procedure is designed to cover DUs that are missed during FE. The proposed hybrid sampling frame would use address-based sampling (ABS) as a frame for the vast majority of NSDUH segments, but would still rely on FE and the HOI procedure for the remaining (primarily rural) segments where ABS coverage is low. The Check for Housing Units Missed (CHUM) procedure would be used to cover missed DUs in ABS segments (see Section 3 for details). This report summarizes research conducted by RTI to evaluate the coverage, cost, and implementation of the proposed hybrid sampling frame for NSDUH. For a further discussion of how coverage prediction affects the costs of the hybrid frame, see Section 4.1.2. IMPLEMENTATION OF THE CHUM PROCEDURE: 1. The CHUM procedure is an ABS frame-supplementation procedure. The CHUM is implemented by field staff from selected DUs to pick up any DUs that are not included on the ABS frame. When implemented correctly, it gives every DU in a sampled segment a chance of selection with known probability. The implementation of the CHUM procedure is discussed further in Section 3.2. ABS frames supplemented with the CHUM procedure provide 100 percent coverage of the target population when the CHUM is implemented correctly. We conducted a field study (MLFS II) to measure how well field staff implement the CHUM procedure in various situations they are likely to encounter in the field. The success rate for typical CHUM intervals2 was 90.7 percent compared with 60.0 percent for high-difficulty intervals. Segments that meet or exceed a specified coverage threshold will be assigned to ABS; otherwise, they will be assigned to FE. ABS coverage thresholds of 50, 65, or 80 percent respectively would require that approximately 8, 14, or 27 percent of NSDUH segments be assigned to FE (see Section 5.1 for details).3. ABS coverage of group quarters (GQs) is problematic. Therefore, segments with high concentrations of GQs should be allocated to FE whenever possible. The 2010 Decennial Census is the only feasible source for predefining segments requiring FE based on having a large noninstitutional GQs population. However, as the data age, the quality of the predictor will deteriorate (see Section 5.2 for further details).4. Geocoding error occurs when the geographic coordinates assigned to a DU do not correspond to its actual location. Without a frame supplementation procedure like the CHUM, geocoding error can lead to both over-coverage error and under-coverage error of an ABS frame. Topics include results from the Mailing List Field Studies (MLFS) I and II, subsequent work developing and testing the CHUM procedure, and exploratory analyses on coverage prediction, group quarters coverage, geocoding error, and potential supplemental sources of addresses. Major research findings include the following:COVERAGE OF THE HYBRID FRAME: 1. Theoretically, the hybrid frame provides 100 percent coverage of the target population. In FE segments, the coverage is equivalent to the current NSDUH coverage rate. In ABS segments, DUs that are not included on the ABS frame are covered by the CHUM procedure. The only known sources of under-coverage occur when field staff incorrectly implement the CHUM and/or HOI procedures.2. The ABS coverage threshold is defined for each segment as the ratio of the number of DUs with locatable mailing addresses1 to the total number of DUs in the segment. |
|---|---|
| Physical Description: | 1 PDF file (ix, 41 pages) illustrations |