JMIR Publications

ABSTRACT

Background:

Traditionally, surveys are conducted to answer questions related to public health but can be costly to execute. However, the information that researchers aim to extract from surveys could be potentially retrieved from social media – data that is highly accessible and lower in cost to collect.

Objective:

In this study, we evaluate whether attitudes towards COVID-19 vaccines collected from the Household Pulse Survey can be predicted using attitudes extracted from Twitter. Ultimately, we would like to determine whether Twitter can provide us with similar information to what is observed in traditional surveys, or, if saving money comes at the cost of losing rich data.

Methods:

COVID-19 vaccine attitudes were extracted from the Household Pulse Survey collected between January 6th and May 25th, 2021. Twitter’s streaming API was used to collect COVID-19 vaccine tweets during this same period. A sentiment and emotion analysis of tweets was conducted to examine the attitudes towards the COVID-19 vaccine on Twitter. Regression analyses were conducted to evaluate the ability for COVID-19 vaccine attitudes on Twitter to predict vaccine attitudes in the HPS.

Results:

The results revealed that attitudes of COVID-19 vaccines found in tweets explained 61-72% of the variability in the percentage of HPS respondents that were vaccine hesitant or compliant. We also found significant statistical relationships between perceptions expressed on Twitter and in the survey.

Conclusions:

These results suggest the information researchers aim to extract from surveys could also potentially be retrieved from a more accessible data source, such as Twitter data.