CPES logoCollaborative Psychiatric
Epidemiology Surveys

Data Collection

The organizational structure of the field and central data collection staff for the NCS-R, NSAL, and NLAAS was divided into teams of 6 to 12 interviewers. Each team was supervised by a team leader. Approximately three to four teams formed a workgroup, which was supported by a team leader coordinator. Each workgroup was assigned to a regional field manager, who was responsible for the workgroup's interview production efforts, quality control, and personnel management. Whenever possible, teams were comprised of groups of interviewers from the same region. Every effort was made to assign interviewers to teams prior to training so that interviewers on the same team would be able to work together during training.

An overview of the field organization and structure of the data collection for all three studies is described in Figure 1.

Figure 1. Organizational Structure for CPES data collection
 

For NCS-R, a sample of 13,054 addresses was fielded. In 98.1% (11,222) of occupied housing units (11,443), interviewers were able to determine whether or not the household was eligible for inclusion in the study. A total of 10,622 addresses yielded an eligible household. As summarized in Table 8 below, 9,282 adult interviews were completed: 7,693 interviews with the main respondent and 1,589 interviews with a second adult in the household. A total of 554 additional interviews (not included in Table 8) were collected using a shortened form of the instrument with a subsample of non-respondents in an effort to assess non-response bias.

Table 8. Interviews, response rate, interview length, and number of contacts for NCS-R, NSAL, NLAAS
Component Interviews Response rate (%) Average
interview
length (mins)
Average
Contacts
per interview
NCS-R
      Main respondent 7,693 70.9 126 7.1
      Second respondent 1,589 80.4 124 4.7
NSAL
      Adult respondent 4,842 (core)
1,357 (supplement)
71.5 (core)
76.4 (supplement)
145 7.4
NLAAS
      Main respondent 3,620 75.7 161 9.2
      Second respondent 1,029 80.3 152 11.6

The final weighted response rate for NCS-R, excluding the Part 1 interviews, was 70.9% for primary respondents (American Association for Public Opinion Research (AAPOR) Response Rate 3; AAPOR, 2004). The response rate for the second respondent was 80.4%. See Kessler et al. (2004) and Heeringa et al. (2004) for a complete discussion of final sample outcomes and response rates.

In contrast to the relatively high sample yield for NCS-R, in NSAL, 11,634 eligible households were identified from 26,495 randomly sampled addresses. A total of 6,199 adult respondents were interviewed as part of NSAL (1,006 White respondents, 1,623 respondents of Caribbean descent, and 3,570 African American respondents). The overall response rate for the core NSAL national sample was 71.5% (AAPOR Response Rate 3; AAPOR, 2004). The Caribbean Supplement sample, which was designed to target areas with high concentrations of persons of Caribbean origin, yielded a weighted response rate of 76.4%. See Jackson et al. (2004) and Heeringa et al. (2004) for additional description of the final distribution of sample dispositions and response rates.

The NLAAS project also involved a significant amount of screening. From a total sample of 27,026 addresses, 4,345 eligible main respondents and 1,234 eligible second adult respondents were identified. For NLAAS, 3,620 main respondent and 1,029 second adult interviews were completed. The weighted response rate for NLAAS was 75.7% among main respondents (77.6% for Latinos, 69.3% for Asians) (AAPOR Response Rate 3; AAPOR, 2004). For second respondents, the final response rate was 80.3% (82.4% for Latinos, 73.7% for Asians). As noted earlier, 215 non-Asian, non-Hispanic white controls were interviewed prior to the elimination of the interviewing of White controls due to budget constraints. More detailed information regarding sample outcomes and calculation of response rates can be found in Heeringa et al. (2004). See Table 8 for a general summary of data collection results across the three studies.