Supp Stmt Part B 103107

Supporting Statement PART B

for Paperwork Reduction Act ICR

Evaluation of the Spanish-Language Campaign “Good Morning Arthritis, Today You Will Not Defeat Us.”

Project Officer:

Lee Ann Ramsey

Public Health Advisor

Arthritis Program

Centers for Disease Control and Prevention

Atllanta GA 30341

Phone: 770-488-6036

Fax: 770-488-5486

e-mail: [email protected]

October 31, 2007

Table of Contents

B.	Statistical Methods (used for collection of information employing statistical methods)
1.	Respondent Universe and Sampling Methods
2.	Procedures for the Collection of Information
3.	Methods to Maximize Response Rates and Deal with Nonresponse
4.	Test of Procedures or Methods to be Undertaken
5.	Individuals Consulted on Statistical Aspects and Individuals Collecting and/or Analyzing Data

2. COLLECTION OF INFORMATION EMPLOYING STATISTICAL METHODS

Overview of “Good Morning Arthritis, Today you will not defeat us.” Evaluation

The evaluation of the “Good Morning, Arthritis, Today you will not defeat us” campaign will use a quasi-experimental design, and will collect data across three points in time. The primary method for data collection will be a 13 minute campaign tracking survey administered via telephone in six markets, including four test markets and two comparison markets. The markets are listed in the table below.

Experimental Market 1 Market 2 Market 3 Market 4	South Valley, NM Orlando, FL Oxnard, CA Aurora, IL and Waukegan, IL
Control Market	Brownsville City, TX
Control Market	Las Vegas, NV

The test markets will each be exposed to the campaign materials, while the two comparison markets will not. Pre and post-exposure telephone surveys will be collected in each of the six markets, allowing us to compare before and after effects of the campaign. (See screening tool in Appendix D1, and pre- post- questionnaire in Appendix D2). Both the pre and post-exposure market readings will be conducted with different samples, not with the same individuals across both waves. The six month follow up measure, however, will be with respondents originally contacted in the pre and post-campaign data collection.

Aeffect, Inc., the vendor selected to conduct the evaluation, will utilize the following quasi-experimental controlled market study design incorporating pre and post-campaign exposure data in four experimental markets and two control markets (also a non-equivalent pretest/posttest design). We have also included a six month follow-up measure with up to 100 individuals in each experimental market, as well as up to 100 randomly selected individuals in each control market. The purpose of the follow-up will be to determine the extent to which campaign messages may have been retained and whether progress toward intermediate outcomes can be observed.

Study Design

-------------------- Time -----------------------

	1 (pre)		2 (post)	3 (follow-up)
Experimental Market 1	O	X	O	O
Experimental Market 2	O	X	O	O
Experimental Market 3	O	X	O	O
Experimental Market 4	O	X	O	O
Control Market 1	O		O	O
Control Market 2	O		O	O

Note: In the grid shown above, “O” indicates an observation and “X” represents the campaign treatment. The dashed line indicates non-random assignment of respondents to the experimental and control groups given that separate markets/cities will be used.

1. Respondent Universe and Sampling Methods

It should be noted the respondent universe is not intended to be representative of the U.S. That is, results are not intended to be projected to the larger U.S. population. Rather, the respondent universe represents those people within each market who could have been exposed to the campaign being evaluated. The sample sizes below were selected based on power analysis and the known limitations of response rates among Hispanic audiences. Aeffect will interview 200 Hispanic, lower SES PWA per market in each pre- and post-campaign wave of data collection. The following table details the sampling plan for each location.

	Before Campaign Launches	6 Weeks After Campaign	6 Months After Campaign
Experimental Market 1	200	200	100
Experimental Market 2	200	200	100
Experimental Market 3	200	200	100
Experimental Market 4	200	200	100
Control Market 1	200	200	100
Control Market 2	200	200	100
Total Sample	1,200	1,200	600

The potential respondent universe for the survey is estimated at 75,000. This is based on the total size of the sample list that will be purchased and used for contacting potential respondents. Respondents will be required to be people with arthritis, Hispanic, between the ages of 45-64, and with household incomes below $35,000. Furthermore, since the survey will be conducted in Spanish, all respondents must be fluent in Spanish. Taking into account these demographic screening criteria, refusals, and the potential for disconnected numbers (which are noted to be higher among Hispanics) the rate used for determining list sample size was a ratio of one to twenty-five. Thus, with a target sample size of 3,000, a list of 75,000 respondents will be purchased. Although it is highly unlikely that all respondents will be called, the complete sample list represents the potential respondent universe.

The respondents from the purchased sample lists will be chosen at random through Random Digit Dialing (RDD). This will insure no bias in the selection of which respondents are called at what times throughout the day. Furthermore, gender distribution will be monitored to obtain an approximate 2:1 ratio of women to men based on the overall prevalence of people with arthritis.

2. Procedures for Collection of Information

All data collection will be completed via computer aided telephone interviewing (CATI). All interviewers will be professionally trained and will also be required to attend an additional training session specific to the questionnaire used in the study. During the training session, interviewers will have the opportunity to ask questions and practice with the questionnaire on a CATI program. All interviewers will be proficient in Spanish with prior experience conducting interviews by telephone in Spanish. Furthermore, all interviewers will also be randomly monitored on a weekly basis by telephone for any additional issues that may require clarification or instruction.

As the demographic screening criteria for participation in the survey are quite extensive and the study is not meant to be representative of the U.S., CDC has decided to purchase targeted lists of randomly generated telephone numbers in geographic areas (neighborhoods) known to have higher proportions of lower SES individuals and Hispanic populations. This should lead to a higher incidence of qualification and survey completion.

After completion of the pre-campaign data collection, a different sample will be utilized for the post-campaign data collection. That is, prior to the second wave of data collection (post campaign), the sample will be purged to ensure that no duplicate phone numbers exist from the pre-campaign data collection. In both the pre- and post campaign data collection, respondents who complete a survey will also be asked if they would be willing to participate in a six month follow up survey. First names (not last names) and telephone numbers of respondents who complete a survey will also be recorded for use in the six month follow up survey. All records of names and phone numbers will be destroyed immediately after completion of the six month follow up survey and prior to any data analysis.

Six months after the post campaign data collection, CDC will re-contact respondents who participated in the pre and post campaign questionnaire. The sample list will be created by retrieving phone numbers of respondents from the pre- and post-campaign data collection. Random digit dialing from this list of telephone numbers will be used to select follow-up respondents.

3. Methods to Maximize Response Rates and Deal with Non-response

As previously stated, every effort will be made to ensure a high response rate. In addition to the use of targeted lists, CDC has planned several methods to ensure high response rates. First, CDC has kept the CATI questionnaire to an average length of 15 minutes or less (including the screening questionnaire) because longer surveys tend to have higher drop-out rates toward the end of the survey. Furthermore, to keep the questionnaire simple and easy to understand, nearly all questions are closed-ended. All respondents will also have the option to skip questions at any time. The survey will also be conducted entirely in Spanish.

To further improve response rates, interviews will occur throughout the day and evening on weekdays and weekends, ensuring that everyone has an equal opportunity to participate in the survey. Particularly given the lower economic status and older age of the target audience, this will be important as respondents may have unique work schedules (late evening shifts) or physical limitations that cause them to be at home during the day. All household phone numbers will be called with predictive dialers (RDD), allowing for faster dialing of numbers and reducing the potential for human error. For the six month follow up data collection, respondents from the pre and post-campaign data collection will be asked if they would be willing to participate in the follow up survey.

To address the issue of potential non-response bias, at least 10% of households with initial refusals will be called back a second time in an attempt at re-conversion. Those households that then subsequently choose to complete the survey will be marked for comparison (assuming sufficient sample size) with the larger pool of surveys to see if there are significant differences in response. If significant differences are found, these findings will be noted in the report and the data set can, if necessary, be weighted to account for any differences.

4. Tests of Procedures or Methods to be Undertaken

The survey instrument has been reviewed by several experts and revisions made as recommended. This instrument is a slight modification of the instrument used in the evaluation of the English language version of the same campaign (OMB clearance #0920-0627) in 2004. A Spanish language expert will be utilized to ensure accurate translation (including back translation) and description of key measures.

5. Individuals Consulted on Statistical Aspects and Individuals Collecting and/or Analyzing Data

The CDC has contracted Aeffect, Inc. to collect and analyze the data. The account team listed below completed the evaluation design. Aeffect will manage and conduct all survey programming, data processing, and report preparation. Specific contributors from Aeffect are as follows:

Ms. Michelle Kuhn, President

Mr. Jed Lam, Vice President Account Services, [email protected]

Ms. Wendy Joice, Vice President Information Technology

Aeffect, Inc.

520 Lake Cook Road, Suite 100

Deerfield, IL 60015

Tel: 847-267-0169

Appendices

1. Regulatory Authority for Data Collection

2. Federal Register Notices

B1. 60-day FRN, March 22, 2006 (Vol. 71, No. 55, pp 14531-14532)

B2. 60-day FRN, April 4, 2007 (Vol. 72, No. 64, pp 16369-16370)

3. Public Comment and Response to Federal Register Notice

4. Baseline and Immediate Post Intervention Data Collection Tools

D1. Baseline and Immediate Post intervention Screener

D2. Baseline and Immediate Post intervention Telephone Interview Questionnaire

5. Six Month Follow Up Data Collection Instruments

E1. Six Month Follow up Screener

E2. Six Month Follow up Telephone Interview Questionnaire

6. Sample Campaign Materials

7. Aeffect Non-Disclosure Agreement

2