Comparing Public Opinion Polling with Social Media Listening for Purposes of Communications Research

• Activity Number: 105
• Type: Invited
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Social Statistics Section
• Author(s): Jennifer Hunter Childs*+ and Monica Wroblewski
• Companies: U.S. Census Bureau and U.S. Census Bureau
• Keywords: social media listening ; text analytics ; privacy ; trust ; confidentiality ; statistical system

Abstract

The Census Bureau collaborated with other federal statistical agencies to understand public trust in official statistics in the United States. For a period of almost 2 years, a survey was fielded daily to a sample of 200 respondents using the Gallup Daily Poll. This data collection, start-to-finish, cost the Census Bureau approximately $5 million dollars. We hypothesize that we could elicit the same type of information through Social Media Listening, in a method similar to that used to measuring brand sentiment (see Schweidel and Moe, 2011) or to track flu activity (see Signorini, Segre, and Polgreen, 2011).

Using a Social Media Listening text mining approach, we sought to identify public discussion about the government and topics related to privacy, confidentiality, and administrative records usage. Passive data collection allows us to gather data in a way that will not influence responses. Real-time systems, such as these, could possibly allow the Federal Statistical System to react to an emerging public relations problem prior to it becoming a big problem. This paper will compare results between the Gallup Poll and Social Media Listening.