B. Statistical Methods
Respondent Universe and Sampling Methods
The respondent universe is members on GfK’s KnowledgePanel® and Knowledge Panel Latino®.
KnowledgePanel® is a probability-based online non-volunteer access panel. Panel members are recruited using a statistically valid sampling method with a published sample frame of residential addresses that covers approximately 97% of U.S. households. Sampled non-Internet households, when recruited, are provided a netbook computer and free Internet service so they may also participate as online panel members. KnowledgePanel® consists of about 50,000 adult members (ages 18 and older) and includes persons living in cell phone only households. The multi-dimensional Hispanic population is also represented in KnowledgePanel with members recruited in both English and Spanish and thereby representing different levels of language proficiency and acculturation levels. Due to voluntary withdrawal, involuntary retirement of high-tenured members, and the addition of new panelists from the on-going recruitment, actual panel size fluctuates across the year.
Currently, 40% of KnowledgePanel®’s members were recruited through the random-digit-dialing technique and 60% using the address based sampling (ABS) technique (based on the U.S. Postal Service’s Delivery Sequence File). KnowledgePanel® is based on a household sampling frame which recruits households (1) with unlisted telephone numbers, (2) without landline telephones, (3) that are cell phone only, (4) without current Internet access, and (5) without devices to access the Internet. After initially accepting the invitation to join the panel, participants are asked to complete a short demographic survey (the initial profile survey); answers to these questions allow efficient panel sampling and weighting for surveys. Completion of the profile survey allows participants to become panel members.
The sample for KnowledgePanel Latino® uses national RDD samples targeting telephone exchanges associated with census blocks that have a 45% or greater Latino population density (this density level covers just over 50% of the U.S. Hispanic population). Households are screened in the Spanish language to recruit only those homes where Spanish is spoken at least half the time. This all probability-based RDD Spanish-language sample supplements the Latino households (English and Spanish) that are now recruited through the KnowledgePanel’s general ABS recruitment sample.
More details about KnowledgePanel®’s design and general sampling approaches can be found at http://www.knowledgenetworks.com/ganp/irbsupport/.
This study will aim to obtain 3,000 completes by adult members of its KnowledgePanel® and KnowledgePanel Latino®, evenly distributed among three subsamples: 1,000 English-dominant Hispanics, 1,000 Spanish-dominant Hispanics, and 1,000 English-speaking non-Hispanics. GfK will draw each of the three subsamples using probability-based sampling and ensure the ending composition of each subsample matches that of the corresponding population group’s Census benchmarks. More specifically, the overall sample is drawn using a weighted proportional sampling approach so that all major population categories are correctly proportionally represented in the sample and all members of the panel have a known probability for inclusion.
English- or Spanish-dominance for KnowledgePanel Latino® participants is established using their answers to the following questions after they agreed to join the panel (response options are: very well, pretty well, just a little, not at all, and don’t know):
Would you say you can carry on a conversation in Spanish, both understanding and speaking?
Would you say you can read a newspaper or book in Spanish?
Would you say you can carry on a conversation in English, both understanding and speaking?
Would you say you can read a newspaper or book in English?
From these questions, four variables are created that capture proficiency in speaking and reading English and speaking and reading Spanish. Based on the four variables, Spanish (English) dominance means a participant’s self-reported conversational and reading proficiencies are stronger in Spanish (English) than in English (Spanish).
Procedures for the Collection of Information
Members on KnowledgePanel® and KnowledgePanel Latino® panels will be invited by email to complete the study online (see Appendices A and C for the pretest and full-study invitation, respectively). Participants can choose to answer either an English questionnaire or a Spanish questionnaire; the choice will be recorded. The study is expected to take an average of 15 minutes to complete (see Appendices B and D for the full-study questionnaire, respectively). Reminder emails will be sent to those who have not completed their questionnaires every three days during the field period (see Appendix E for the reminder).
The study includes the following topics: (1) food consumption habits such as frequency of eating food items that particularly might pose risk of foodborne illness (e.g., hamburgers, uncooked vegetables, raw fish, salsa); (2) food handling practices (e.g., washing hands, handling cutting boards, storing leftovers); (3) food safety information (e.g., sources of information, trust in sources of information); (4) risk perception and foodborne illness experience; and (5) personal background information such as a measure of acculturation and health condition. Many of the substantive questions are borrowed and/or adapted from the FDA Food Safety Survey that we have regularly commissioned over the years (OMB No. 0910-0345). Spanish translation of many questions is already available from existing surveys and will be used in this study. The translation of remaining questions will be developed.
Methods to Maximize Response Rates and Deal with Non-response
We will keep the study questionnaire at a reasonable length to encourage participation and minimize break-offs. We will offer Spanish-dominant Hispanic participants a $5 incentive to encourage their participation (see Part A, Item 9). We also will send reminders (Appendix E) three days after initial invitations if no responses to the initial invitations are received. We will regularly monitor the progress of the fieldwork to identify and resolve any administration issues that might prevent the study from achieving a successful implementation.
Test of Procedures or Methods to be Undertaken
We will conduct nine in-person cognitive interviews to help improve understandability of the questionnaire, to reduce participant burden, and to enhance interview administration. We will use a screener (Appendix F) to select eligible participants. Eligible participants will first be asked to complete a test questionnaire (Appendix G) by themselves on a computer. Then, they will be debriefed by an interviewer about any difficulties they had in understanding and responding to questions. We plan to use the results of the debriefing to refine and improve the quality of the questionnaire.
Individuals Consulted on Statistical Aspects and Individuals Collecting and/or Analyzing Data
Chung-Tung Jordan Lin, Ph.D., at the FDA will supervise the data collection and lead data analyses.
List of Appendices
A Pretest invitation
B Pretest questionnaire
C Study invitation
D Study questionnaire
E Reminder
F Cognitive interview screener
G Cognitive interview questionnaire
| File Type | application/msword |
| File Modified | 2016-06-23 |
| File Created | 2016-06-23 |