The protests look dramatic on TV: crowds of people clamouring for their rights and freedoms amidst business closures, curtailed socializing and mask wearing, among other restrictions.
Does this discontent represent the wider views of Ontario residents regarding measures to stop the spread of COVID-19?
Not necessarily, says new Brock University-led research.
Assistant Professor of Health Sciences Antony Chum and graduate students Andrew Nielsen and Zachary Bellows, have found that conversation in the general public is much more nuanced and balanced when it comes to views on pandemic measures.
The research team used the social media platform Twitter to monitor and analyze Ontarians’ reactions to a range of pandemic restrictions including business and school closures, regional lockdown differences and other public health restrictions such as social distancing and masking.
The researchers examined 1.5 million Tweets about COVID posted in Ontario between March 12 and Oct. 31, 2020.
The team was looking for three trends: the volume of COVID-related Tweets, especially following major government announcements; the emotional tone of the posting; and the level of disagreement or polarization during Twitter discussions.
An artificial intelligence-powered natural language processing program classified Tweets into positive, negative or neutral categories.
The researchers discovered some interesting associations.
“Our findings actually show that a partial lockdown is associated with more positive tweets compared to a provincewide lockdown,” says Nielsen, an Applied Health Sciences master’s student.
“We believe this is the case because people were aware of their local infection rates, and would want their local policies to reflect their perceived level of danger,” he says, adding that many people agreed that areas with high infection rates should be under lockdown while other areas with low rates should be more open.
Fellow Applied Health Sciences master’s student Zachery Bellows notes that, as expected, the volume of Tweets rose with new COVID-related announcements.
“While there was a lot more discussion in general, we found that, overall, there weren’t too many people up in arms on either side,” he says.
Chum, Bellows and Nielsen collaboratively conducted the study, which has been accepted for publication in the journal JMIR Public Health and Surveillance.
Other team members include Brock research associate Eddie Farrell, software engineer Pierre-Nicolas Durette, Professor Gerald Cupchik from the University of Toronto and Assistant Professor Juan M. Banda from Georgia State University.
Highlights of their study, “Changes in public response associated with various COVID-19 restrictions in Ontario, Canada: an observational study using social media time series data,” include:
- Higher new COVID-19 case counts in Ontario increased negative opinions; each additional 100 cases per day increased negative tweets by three per cent.
- The effect of business closures on public opinion depended on the number of new COVID-19 cases; closing businesses when there were 50 new cases was associated with three times more negative tweets compared to closing with 200 new cases.
- The announcement of additional public health restrictions was associated with 544 additional tweets but did not affect public opinion.
- During a provincewide lockdown, there were five times more negative tweets compared to a partial lockdown.
Chum says monitoring and analyzing Twitter posts is a good way for policy-makers to get a handle on what the general public is saying about COVID restrictions. He estimates that 40 per cent of Canadians are regular Twitter users.
“It’s very easy for politicians to look to the protests and large gatherings and say, ‘OK, that’s what represents Canadians,’” says Chum. “If you look at larger data sets like all the discussion that goes on in the Twitter community, we can actually get a better understanding of what the public’s overall emotional tone is, and it’s not that negative.”
Public health practitioners can look to reactions and conversations on Twitter to help them shape and improve their public health messaging, since “when people agree with the restrictions, they’re more likely to follow them, so understanding the public’s reaction to these restrictions is very important,” says Chum.