Release Date: 11/23/2013
Use of Blended Sample Frames in Surveying Rare Population
November 23, 2013 -- The challenge facing Abt SRBI researchers in a recent major sensitive survey was how to reach 18-26 year olds in selected, often high-crime New York City neighborhoods. The two-wave survey was probing experiences with and views about the police in their neighborhoods, particularly any experiences with the city’s controversial “stop and frisk” enforcement program.
The Abt SRBI research team, together with researchers from Yale and Columbia universities, presented a paper at the late November conference of the Midwest Association for Public Opinion Research (MAPOR) in Chicago, detailing the elaborate sample design to identify and interview this very difficult-to-reach population in very small geographic areas.
To meet these challenges efficiently, the team developed and carefully evaluated a representative multi-frame sample -- list frame, landline and cell RDD samples. Abt SRBI completed a total of 1,000 interviews with 18 to 26 year old respondents in 37 New York City low SES neighborhoods.
The paper presented and then evaluated the blended sample approach. Consumer database lists can serve as efficient sampling frames for rare and hard-to-reach populations because the incidence of the rare population on the list is relatively high. However, the drawback of using list frame alone is that it often provides incomplete coverage of the population. Therefore, the blended strategy was utilized, integrating the various sample frames.
To locate respondents within the small neighborhood areas, the Abt SRBI team implemented a sequential geographic screening with real-time customized GIS “geoprocessing.” Abt SRBI has been a leader in applying GIS technology to surveys. For cell frame sample, the team matched the cell’s billing code to the target area and excluded cell numbers outside the billing area.
The sample evaluation found the following:
Respondents from each sample were equally likely to report being stopped by a police officer in the past 12 months.
List sample respondents more likely to be:
lived in their neighborhood for more than 10 years.
College educated, and
RDD samples were better at reaching minorities, those who never attended college, and who lived in public housing projects
The paper’s authors were Chintan Turakhia, Dean Williams, and Marci Schalk of Abt SRBI, Tom Tyler, Yale University and Jeff Fagen, Columbia University.
Copies of the paper may be obtained from Chintan Turakhia, email: firstname.lastname@example.org.