ESRA 2017 Programme

Tuesday 18th July      Wednesday 19th July      Thursday 20th July      Friday 21th July     

     ESRA Conference App


Thursday 20th July, 14:00 - 15:30 Room: F2 103

Uses of Geographic Information Systems Tools in Survey Data Collection & Analysis

Chair Dr Stephanie Eckman (RTI )
Coordinator 1Mr Ned English (NORC)

Session Details

The techniques of remote sensing and geographic information systems (GIS) can provide new methods for frame creation and sample selection and help us study and reduce coverage, nonresponse and measurement error. GIS-based tools are also used in the analysis of survey data and presentation of results. As these technologies become less expensive and easier to use, and geographic data availability grows, we expect survey researchers to make more use of GIS. While we embrace geospatial methods, however, we should also maintain a healthy skepticism about their capabilities and limitations.

This series of sessions at the ESRA 2017 conference in Lisbon will bring together survey researchers from different countries to discuss novel applications of GIS technology to data collection and analysis and to share ideas. We encourage papers that discuss the use GIS or GPS technologies in any stage of the survey process, and how these tools can help us understand, reduce or adjust for different error sources. We are also interested in papers that review errors in GIS technology and how they can impact survey quality.

Paper Details

1. Exit Polling and Geolocation Technology
Dr Rene Bautista (NORC at the University of Chicago)
Dr David Sterrett (NORC at the University of Chicago)
Dr Jennifer Benz (NORC at the University of Chicago)
Mr David Pace (The Associated Press)
Mrs Emily Swanson (The Associated Press)
Mr Trevor Tompson (NORC at the University of Chicago)

In order to explore potential improvements in exit po lling, this study explores the feasibility of using geolocation technology to send people a survey on their smartphones after they leave their voting places. The Associated Press and NORC at the University of Chicago, with funding from the Knight Foundation, conducted the feasibility study with participants in Florida during the 2016 US Presidential Election. NORC leveraged its nationally representative AmeriSpeak panel to test people’s willingness to complete an exit poll on their smartphone. About 1,000 participants were invited to download the SurveySwipe app, which uses geolocation technology to send people surveys when they enter certain locations. The latitudes and longitudes of participants’ early voting and Election Day polling places were programmed into the app and people received a short survey via the app after entering their polling place. The results highlight people’s general willingness to participate, the download rate of the app, and the c ompletio n rates for the survey. The analysis illustrates the demographic profiles of those who cooperate and download the app. In addition, the study findings highlight various challenges and opportunities when using geolocation technology to conduct rapid response surveys.


2. Geo-Sampling for Establishment Survey Enumeration and Sampling: Nigeria Non-facility Based Health Service Providers Case Study
Miss Justine Allpress (RTI International)

Establishment survey frames usually suffer from coverage bias and outdated records. A full listing or even a re-listing exercise to update the frame is an intense, time consuming, and costly process. In many studies, researchers seek a dual frame approach in order to reduce the coverage bias but that still raises challenges and weighting issues. In this paper, we explore the use of Geo-sampling, a GIS-driven sampling methodology, to determine areas to be enumerated for sampling non-facility based health service providers in two states in Nigeria (Lagos and Kaduna) as a case study.
A survey of households based on a Geo-sampling design is being conducted concurrently to a list-based survey of health providers to assess both demand and supply of health services within the same geographic areas. The health provider survey includes both facility-based and non-facility-based providers. Comprehensive list frames exist for the facility-based providers such as hospitals and clinics. However, for the non-facility based health providers, which includes patent medicine vendors, pharmacists, and traditional birthing assistants, lists are only partially available in the case of the pharmacists and suffer from incomplete information or coverage issues when they do exist. No lists exist for patent vendors and birthing assistants. The study aims at selecting a representative sample of non-facility based health service providers in Lagos and Kaduna. A full enumeration effort of these providers is not possible given the study’s limited resources and tight project timeline.
The sample of the geographic units selected for the household portion of the study was available, so Geo-sampling emerged as a solution for selecting a sample of units for enumeration and selecting a representative sample of non-facility based service providers. The strategy involves selecting a complex sample of geographic areas from the set of areas to be worked for the household study and conducting enumeration of the non-facility based providers within these units. This approach facilitates the creation of a spatial linkage between the two targeted populations and limits the number of geographic areas required for fieldwork across both surveys. Rules are devised to maximize the benefit of having a partial existing list of pharmacists .