Iowa Survey of Public Attitudes: An Optimal Approach

University Of Iowa, Iowa City IA

Investigators

Abstract

SES-0825588 Kevin T. Leicht, Naresh Kumar The University of Iowa The Social Science Research Center (SSRC) at the University of Iowa will complete a pilot study in 2010 to test the efficacy of using a spatially dispersed sampling design to collect survey data. The design will capture the maximum variance in socio-economic conditions and contextualize the neighborhood and physical environment surrounding each survey respondent (using their precise location). The PIs will draw a full household sample for the 2010 and systematically compare the sampled regions and respondents with those selected in the conventional General Social Survey (GSS). The PIs will field individual-level surveys in one EPA region in 2010 for the purposes of comparing their entire data collection method and results with those from the conventional GSS for these same regions. While the GSS has provided unprecedented access to the attitudes and circumstances of surveyed individuals, the links between these and the larger cultural, economic, political, and environmental contexts where people live and work have been indirect at best. This approach takes advantage of the rapid quantification of multi-level geographic data and its extensive integration with information on built environments, toxic exposures, and other features of human and natural geography that are increasingly used in theories in social and behavioral science and public health. The PIs will construct a spatial sampling frame of inhabited areas and then draw a sample that can capture the maximum variation in population distribution and SES. They will utilize geographic information systems, remote sensing and locational analysis for constructing a sampling frame and contextualizing individual GSS respondents. This approach offers several advantages over the conventional sampling designs by (a) capturing the maximum spatial dispersion in population size and SES; (b) avoiding spatial autocorrelation; (c) linking sampled respondents (via geocodes) to many additional pieces of contextual information including SES characteristics of a place, land use and land-cover type, and potential sources of toxic emission; (d) ensuring spatial coverage and adequate representation across SES and ethnic groups; and (e) providing interviewers with information that allows for a better assessment of respondents' answers in relation to their lived context. This study presents a mechanism for conducting and integrating the General Social Survey's traditional strengths in collecting nationally-representative survey data through face-to-face interviews with spatial samples of adults in the United States with a geographic information data management system that will locate GSS respondents in multiple layers of geographic space. The approach builds on the rapid quantification of geographic data and its extensive integration with information on built environments, toxic exposures, and other features of human and natural geography that are increasingly being used to theorize social science research. The pilot survey in 2010 allows for a systematic comparison of the results from an optimal spatial sampling approach with more conventional sampling methods for a geographic region and allow for a comparison of both full-sampling frames at the national level. Broader Impacts: The world is becoming more interconnected, and the ability to link personal circumstances to broader social structure is central to the social scientific enterprise. The ability to develop these linkages through the GSS will provide an unprecedented opportunity to build social science infrastructure in the United States and (eventually) around the world by allowing researchers and policy makers to have access to spatially linked, demographically and statistically sound data sets. The ability to train future generations of social science students and practitioners in GIS and hierarchical analysis methods will be greatly enhanced, and high quality data will at a previously unheard-of level of detail will be available to the social science and social policy communities. The pilot test of this optimal spatial sampling approach is an important first step toward revolutionizing survey data collection for national surveys combining the latest technologies with conventionally sound survey research techniques.