Screening Households to Select a Sample: Evaluating a Multi-Mailing Contact Strategy
Submission ID: 5332

Date: Thursday, 10:15 AM to 11:45 AM 
Session: Session D: T10:15 - 11:45 AM

Primary Presenter

Jessica Graber, National Center for Health Statistics

Additional Authors or Round Table Presenters

Jay Clark, Westat ,

Kevin Wilson, Westat ,

Allan Uribe, National Center for Health Statistics ,

Juliana McAllister, National Center for Health Statistics ,

, ,

, ,

, ,

, ,

Abstract

The National Health and Nutrition Examination Survey (NHANES) is a nationally representative survey that, since 1999, has used in-person household screening to identify sampled persons.  In-person surveys require trained interviewers to conduct them, which can be an expensive and inefficient use of resources.  These considerations – coupled with the safety concerns raised by the COVID-19 pandemic, and the general survey environment of declining response rates – led NHANES to implement a multi-mode household screener in the 2021-2022 survey cycle.

Sampled dwelling units received four mailings inviting them to respond via a secure website with a unique pass code.  All mailings also included a toll-free number that the respondent could call to complete the screener by telephone.  The mailings spanned six weeks and included: (1) an initial invitation letter with a study-branded magnet; (2) a reminder postcard; (3) a FedEx package that included a letter and a paper version of the questionnaire with a postage-paid return envelope; and (4) a final letter that, in addition to the website and phone response options, informed the recipient of an upcoming visit by a field interviewer.  Responding addresses were removed from subsequent mailings.  Nonresponding addresses were visited by a field interviewer who attempted to administer the screener in person. 

In this presentation, we examine completed screener cases to identify which contact prompted the successful response and identify any variation in mode of response, respondent or household characteristics, including household size, demographic characteristics of the respondent, and number of people sampled. We will also report on available instrument paradata by mode, such as time to complete, device type and time of day of completion. Finally, we compare responses by mode to identify any systematic mode effect.