Challenges of Big Data: Abt SRBI Experts Present Insights

Abt SRBI's Timothy Michalowski presents at the TSE15 conference. Abt SRBI experts presented two papers and chaired several sessions at the international Total Error Conference (TSE15), held September 20-22 in Baltimore. The conference focuses on survey quality and the challenges of big data.

Total Survey Error summarizes the ways a survey estimate may deviate from the corresponding value in the population. Presentations explore ways of reducing survey error to produce more accurate estimates.

Researchers from Abt SRBI, which is the survey subsidiary of Abt Global, presented the following:

The Value of Self-reported Frequently Visited Addresses in GPS Assisted Travel Surveys Timothy Michalowski, Dara Seidl, Rena Peña
 Abt SRBI has been in the forefront of utilizing rapidly improving small personal GPS loggers. These digital devices track, record and, in some cases, transmit in near-real time, a respondent’s travel behavior. Abt SRBI has deployed over 10,000+ GPS devices for various projects throughout the USA. This study examines the value of asking respondents to report frequent location addresses in recruitment and then compare them with their final GPS travel data. The frequent locations from travel survey recruitment are geocoded and compared to GPS travel data in Geographic Information Systems (GIS) using spatial statistics. The study concludes that reporting frequent locations reported in recruitment enhances GPS travel data analysis.
 
Mode Effects in American Trends Panel: A Closer Look at the Person-Level and Item-Level Characteristics Stanislav Kolenikov, Abt SRBI, Kyley McGeeney and Scott Keeter, Pew Research Center, and Courtney Kennedy, Abt SRBI
The American Trends Panel by Pew Research Center and Abt SRBI is an RDD-based probability panel. The panel currently has 4,165 recruited active members, of whom approximately 3,200 complete a typical wave. Panel surveys have been conducted on different modes in different waves, including web for most panel members, and mail or phone for those who do not have access to the Internet. This paper analyzed the results of the July 2015 wave (Wave 5) that included a comprehensive, large-scale mode-of-interview experiment that randomly assigned respondents to telephone and web modes, with approximately 1,500 respondents in each mode. For the purposes of mode effect analysis, each experimental group was weighted separately to national parameters for the general public. To measure mode effect, the researchers built a regression model to identify the properties of survey questions that make them susceptible to mode effects, as well as the demographic groups that tend to exhibit mode effects.
 
Abt SRBI Research Chief and Cofounder, Mark Schulman, and Advanced Methods member Rafael Nishimura also chaired panels at TSE15.