Poster Presentation at the APHA 2019 Annual Meeting and Expo
background: Trust is considered an important factor in vaccine-related decisions. Traditional surveys measure this with thoughtful and rigorous research, but that can be slow and expensive compared to social media analyses. Social media research currently lags behind surveys in terms of rigor because, among several reasons, it requires measuring via inference instead of self-report. There is an opportunity to innovate to leverage strengths of both data sources. objective: This session will exemplify how to combine social media and survey data when characterizing online discussion on trust in vaccines to leverage strengths of both data sources while mitigating weaknesses.
methods: We build a factor analysis linking social media messages with traditional survey questions conveying constructs related to trust in vaccines. Specifically, we recruit respondents to both answer survey questions and indicate agreement with (and whether they would share) tweets related to trust in vaccines. We assemble a set of scaled tweets that convey constructs in relevant survey research, such as trust in government or vaccine uptake. We perform a factor analysis to identify significant relationships between survey answers and agreement with tweets. These significant factors provide a means to translating from Twitter to surveys. We randomly sample Twitter data, apply this translation, and statistically characterize Twitter discussion as it relates to traditional survey results.
results: We will show significant relationships between constructs on surveys and constructs on Twitter, and we will show how these relationships can be used to characterize social media discussions on trust in vaccines.
discussion: This type of analysis can be used to quantify associations and relationships across different data sources like social media and surveys as constructs are conveyed. Relating more free-form social media to traditional surveys whose set of responses is more bounded would facilitate interpretation along important constructs related to trust in vaccines.