Study Finds Twitter Sentiment Positively Correlates with ACA Health Insurance Enrollment

A while ago, several investigators from the University of Pennsylvania—Maarten Sap, Andrew Schwartz, Robert Town, Tom Baker, JD, and Lyle Ungar—conducted a study to examine whether Twitter activity is associated with health insurance enrollment. Is there indeed a correlation? VCBeat has compiled the survey findings as follows:

It is well known that Twitter has played a positive role in tracking the risk of heart disease and conducting medical research. However, a new study from the Perelman School of Medicine at the University of Pennsylvania has found that Twitter is also associated with an increase in health insurance enrollment under the Affordable Care Act (ACA).

Researchers tracked positive or negative sentiment on Twitter, using it as a “real-time measure of public opinion” to enable the rapid detection of changes in health insurance market enrollment and emerging issues.

In March 2014, the study analyzed nearly one million tweets related to the Affordable Care Act (ACA) and Obamacare, including those linking to the Obamacare website HealthCare.gov and 17 official Twitter accounts.

We then tested the correlation between Twitter sentiment and health insurance market enrollment. Twitter sentiment was determined using the National Research Council’s emotion lexicon, which comprises more than 54,000 words ranging from positive to negative.

For instance, “excellent” conveys a stronger and more positive meaning than “good.” The same applies to “bad” and “awful.”

After identifying the vocabulary in the tweets, researchers found that a one-percentage-point increase in tweet sentiment corresponded to a 19% increase in registrations. Although these figures may appear modest, the researchers noted that “when examining sentiment scores across more than one million tweets, these numbers indicate a significant relationship between Twitter sentiment and registrations.”

In addition, the researchers pointed out that these findings have broader implications beyond merely representing a trend.

“Twitter is a quite powerful tool when it comes to researching trends in healthcare policy,” said Raina Merchant, a senior writer, director of the Penn Social Media and Health Innovation Lab, and assistant professor in the Department of Emergency Medicine. “We can see that this approach can also be used to promptly improve healthcare, identify issues, and make timely adjustments accordingly.”

To further validate the National Research Council lexicon, they randomly sampled 300 tweets for manual scoring, and found a significant correlation between the human and computer-generated scores.