Twitter Study of Vaccine Messages: Opinions Are Contagious, But In Unexpected Ways

Remember 2009? The H1N1 pandemic we were all waiting for? I do. I was pregnant; H1N1 was particularly risky for pregnant women. The vaccine wasn’t available until after I had my baby, but when they held a clinic an hour north of where I live, I brought my husband there so we could both get our shots. My infant son was too young to be vaccinated, so I wanted to protect him through herd immunity.

study came out recently on twitter messages from that time. How did pro-vaccine sentiments spread, versus anti-vaccine ones? Which messages were more contagious?

I talked to one of the authors, Marcel Salathe, today. He’s an infectious disease researcher studying the spread and transmission, not (just) of disease, but of information. “We assume people infect each other with opinions about vaccinations,” he said, and the H1N1 scare was a good opportunity to put some of his group’s theories to the test.

They collected nearly half a million tweets about the H1N1 flu vaccine. In 2009, H1N1 wasn’t included in the regular flu shot, and became available partway through flu season as a separate dose. With a possible pandemic looming, people had plenty of motivation to get the vaccine and encourage others to get it—butanti-vaccine sentiments were in circulation too.

The result, striking but perhaps not surprising: negative opinions were more contagious than positive ones. (Specifically, someone who read a lot of anti-vaccine messages was more likely to follow up by tweeting or retweeting negative messages of their own.)

What’s more, being exposed to a lot of positive messages also predicted negativity. Salathe interprets this as a sign of public health messages backfiring: maybe they’re being seen as pushy.

I asked him, is it just that people can be passionate about being anti-vaccine, but nobody is passionate about “yeah, I got my flu shot”? He says maybe not: people who are pro-vaccine are often concerned that those who don’t vaccinate are putting others at risk. On the other hand, the anti-vax folks included people who sent out hundreds of negative tweets; nobody had that kind of enthusiasm on the positive side.

The research really didn’t get into the content of the tweets. In fact, they were categorized as positive or negative by students who read just 10% of the tweets. Computers, following the humans’ lead, automatically tagged the rest. (If you’re interested in the methods, the paper is open access, and the Salathe group’s website is also full of fun reads. If you find statistics and computational analysis fun, that is.)

Does sharing ideas mean sharing germs?

Seth Mnookin has said that people are drawn to groups critical of vaccines because those groups provide a supportive community, especially for parents of kids with autism. I thought about that when reading this paper, about the “echo chamber” style clusters of opinion the researchers found. 

In real life, a cluster of people who don’t vaccinate is a potential hotspot for an outbreak. I asked if people who get that supportive community online might be less of a threat to herd immunity: their buddies could be anywhere in the world, well out of germ-sharing range.

Salathe pointed out that past research has shown that people’s online contacts tend not to stray too far geographically: if you live in San Francisco, you’ll have more contacts in the bay area than you would if you lived in another part of the country.

What would this study look like on facebook, I wondered? (Salathe doesn’t have that data.) In my experience, people who are friends on facebook often know each other in real life: On twitter, I follow scientists, science writers, and interesting people the world over. Meanwhile, most of my facebook friends are my actual friends. The local La Leche League has a lively facebook page, as does my son’s preschool.

This landscape is changing all the time. In 2009, Twitter had about 18 million users; that’s since grown more than tenfold. Facebook, meanwhile, has over a billion.

“Ten years ago we talked about Dr. Google; now we can start to talk about Dr. Facebook and Dr. Twitter,” Salathe said. “People get their information [about health and disease] from social networks.” On the bright side, this means well-crafted public health messages could go viral too. If what we read online influences what we do in meatspace, like getting vaccinations, that could have a big impact.

Beth Skwarecki is a science writer based in Pittsburgh, PA. You can follow here here at her personal website or on Twitter at @BethSkw. This post originally appeared in PLOS: Public Health Perspectives.

1 reply »

  1. Beth, you wrote:

    >>>In my experience, people who are friends on facebook often know each other in real life: On twitter, I follow scientists, science writers, and interesting people the world over. Meanwhile, most of my facebook friends are my actual friends. <<<<

    This is not true of my FB experience. I have never read any study supporting or refuting either of our FB experiences.

    You have just begun a viral message that may, or not be correct. Just sayin'