The Houston Area Survey on Health, Education, and the Arts (SHEA)

Development of the Survey Instruments

The studies focus on the residents of Harris County, since almost 70 percent of everyone in the vast, ten-county, multi-centered Houston metropolitan region resides in this county.

Using "back translation" and the reconciliation of discrepancies, the questionnaires were translated into Spanish, and bilingual supervisors and interviewers were assigned to the project at all times.

The research team organized and convened a series of planning sessions, focus group discussions, and one-on-one interviews with local experts and stakeholders in a concerted effort to ensure that the research is of real and lasting value to the wider Houston community and across the nation.

We have also identified a group of important national figures in each of the three substantive areas, and invited the most knowledgeable and experienced experts available to serve on the advisory panels guide the design and structure of the new surveys.

Click here for a list of the researchers and experts who participated  in this process.

Survey Methodologies

In this new era of "caller-id," voice mail screenings, declining landline penetration, ubiquitous cell phones, and constant telemarketing, the use of cold telephone calling based on conventional random-digit-dialing is becoming less effective as a technique for reaching random samples of a community's residents. It is important for us to revise and expand the sampling procedures for this new study.

The research team has reviewed the wide range of alternative survey methodologies being used in comparable survey projects across the nation. We also discussed the issue with several survey methodologists at the American Association for Public Opinion Research (AAPOR) Annual Conference.

Given that a growing number of Americans are eliminating the traditional landlines in favor of cellphones, a trend led by the young and the poor, it is critical to secure interviews from a random-digit-dialing (RDD) sample of cellphone numbers. Broadly speaking, the sampling design for this study will include 70% of landline and 30% of cellular telephone numbers selected randomly from the pool of numbers expected to serve the Harris County area. We hope that the use of a dual-frame RDD sampling design may compensate for undercoverage bias.

Upon final approval of the questionnaires, in both English and Spanish, the data collection vendor formatted and programmed the survey into the computer-assisted telephone interviewing system, also known as CATI, for use in interviewing.

- Last updated on August 6, 2013.