Sup Stmt Part B

Generic ICR for

BRFSS Preliminary Experiment for Sample and Mode

for the National Center for Chronic Disease Prevention

and Health Promotion

Submitted under

CDC/ATSDR Formative Research and Tool Development

(OMB #0920-1154)

Supporting Statement

Part B

September 8, 2017

List of Attachments:

Attachment 1: BRFSS Data Collection Protocols

Attachment 2: Informed consent script for telephone interviews

Attachment 3: Informed consent script for web-based and mailed questionnaires

Attachment 4: Formatted version of the BRFSS Core Questionnaire for mailing

Attachment 5: Example of letter of invitation to participate in web-based BRFSS

B. Collections of Information Employing Statistical Methods

1. Respondent universe and sampling methods

The purpose of the BRFSS Preliminary Experiment for Sample and Mode is to explore new modes of data collection for the BRFSS core questionnaire. Comparisons of the sequence of data collection modes across two types of samples will allow for understanding of the cost effectiveness and data quality of each method.

Two types of samples will be used: 1) a telephone sample with appended addresses (where available). Approximately 30% of all telephone numbers are likely to have correct address matches; 2) an Address-Based Sample (ABS) from the US Postal Service (USPS) with appended phone numbers (where available). Approximately 40% of all addresses are likely to have correct phone numbers. Both samples (telephone numbers and addresses) that would be used in this experiment are traditional probability samples. This is not an experiment using any nonprobability sampling approach. Both samples will be purchased from an outside vendor.

The samples will be divided into groups, based on their matched address or phone number. Matched sample numbers and addresses will be contacted in sequential modes to complete the BRFSS core questionnaire (see Attachment 4). No BRFSS optional modules will be included in the preliminary experiment. The respondents will continue to be recruited to participate. A target of 300 respondents is set for each group. Five states will be recruited from those state health departments that volunteer to participate. States will be recruited by geographic region (northeast, south, mid-west, west and southwest) and population characteristics. If more than 5 states volunteer, the selection of states will be determined by a range of states based on population density, geographic location and/or population size. At this time it is not possible to name specific states as state recruitment will be determined in part by the ability of the state health department to take on the project in the future. The selected states will be diverse in terms of their populations and locations, but there is no intention of using the participating states to calculate state-wide or national prevalence estimates. Sampling strategies within each state will be uniform. A simple random sample of working phone numbers will be used without regard to substate strata for this experiment. This is different from the state-level sampling designs used by the BRFSS which include sub-state strata. The use of a single sampling design across the five states is intended to standardize methods across the participating states.

The table below illustrates the estimated respondent universe and burden hours for the single administration of the BRFSS Preliminary Experiment for Sample and Mode .

Table 1. Respondent Burden by Type of Information Collected
Type of Respondents	Type of Information Collection	Number of respondents per IC	Annualized number of respondents	Average burden per response (in hours)	Total annualized burden hours
General U.S. Adult Population	Telephone Sample Address Matched Web-based questionnaire	1,500	1,500	15/60	375
	Telephone sample Address-Matched Mailed Questionnaire	1,500	1,500	15/60	375
	Telephone Sample Address Matched Telephone Interviewer	1,500	1,500	15/60	375
	Telephone Sample Not Address- matched telephone interview	1,500	1,500	15/60	375
	Address Based Sample (ABS)
	ABS Phone Matched Web-based questionnaire	1,500	1,500	15/60	375
	ABS Phone Matched Mailed Questionnaire	1,500	1,500	15/60	375
	ABS Phone Matched Telephone Interviewer	1,500	1,500	15/60	375
	ABS Not Phone matched Web-based interview	1,500	1,500	15/60	375
	ABS Not Phone Matched Mailed Survey	1,500	1,500	15/60	375
Total		13,500	13,500		3,375

2. Information Collection Procedures

Information collection will proceed in the following way:

Telephone Sample

• Sample telephone numbers matched to addresses (approximately 30%)

  • Matched numbers will be sent letters with unique IDs and invitation to web-based questionnaire (see Attachment 5 for an example of a letter of invitation)

  • After two months, matched numbers which have not responded to web invitation will be send full printed questionnaires of the BRFSS core survey (see Attachment 4 for a formatted version of the core BRFSS)

  • After two month, matched numbers with have not responded to web invitation nor returned mailed surveys will be called, using the BRFSS Data Collection Protocol

• Sample telephone numbers not matched to addresses (approximately 70%)

  • Data will be collected using the BRFSS Data Collection Protocols

Address-Based Sample

• Sample addresses matched to phone numbers (approximately 40%)

  • Matched numbers will be sent letters with unique IDs and invitation to web-based questionnaire (see Attachment 5 for an example of a letter of invitation)

• After two months, matched numbers which have not responded to web invitation will be send full printed questionnaires of the BRFSS core survey (see Attachment 4 for a formatted version of the core BRFSS)

• After two month, matched numbers with have not responded to web invitation nor returned mailed surveys will be called, using the BRFSS Data Collection Protocol

• Sample addresses not matched to phone numbers (approximately 60%)

  • Matched numbers will be sent letters with unique IDs and invitation to web-based questionnaire

  • After two months, matched numbers which have not responded to web invitation will be send full printed questionnaires of the BRFSS core survey

It should be noted that phone numbers in the telephone sample which are matched to addresses will not overlap locations in the address-based sample. The two samples are independent of each other and are not linked by the matching process. It is highly unlikely that any single residence and/or person would appear in both samples.

3. Methods to Maximize Participation Rates

Since the purpose of this Preliminary Experiment for Sample and Mode is to assess whether different modes are superior in terms of data quality (including response rates) and cost effectiveness, standard methods to achieve the highest response rate will be adhered to. The BRFSS Data Collection Protocol (Attachment 1) provides information on calling times, number of attempts and introductory scripts designed to recruit participants. Since the purpose of the experiment is to determine which mode results in the optimal balance of response rate, cost and burden no changes to the protocol will be permitted, even if they would result in increased participation.

4. Tests of Procedures

The web-based questionnaire will be tested prior to implementation by staff members at the CDC and at participating states. Formatting of the Mailed Survey Questionnaire (see Attachment 4) will also be conducted using health department staff. The BRFSS core survey has been used for many years as a telephone interview and will not require further testing.

5. Statistical Consultants

Other than statisticians within the state health departments and/or employed by the CDC, no outside consultants will be used in the preliminary experiment.

5