Vaccine misinformation is a worldwide concern, but little is known about its spread and adoption. How do people seek vaccine-related information and decide whether to trust it? This is very much an open question.
Using questions and answers retrieved from “Yahoo! Answers,” we explored how individuals seek vaccine-related information online and evaluate its trustworthiness.
Research Question
What factors are associated with the perceived trustworthiness of online content about vaccines?
Specifically, we were interested to know if trustworthiness was explained by the following two factors:
- Stance towards vaccines (”what is said”): Does it matter if the content encourages or discourages vaccination?
- Answerers’ identity (”who said it”): Does it matter if the content is written by someone who identifies as a health professional, as someone who identifies as a parent, or as someone who does not identify as either?
Methods
We conducted an online experiment to find out what factors are associated with the perceived trustworthiness of answers in online Q&A plaforms.
Main Part of the Study
Participants were shown authentic answers from a Q&A platform that:
(1) were either…
✅ pro-vaccination or
❌ anti-vaccination;
(2) and were written by…
🧑⚕️ “Health Professionals” — people who self-identified as health professionals in the answer,
👪 “Parents” — people who self-identified as parents, or
👤 “Neither” — by answerers who did not self-identify as either.
Participants were then asked to rate the trustworthiness of the answerers.
Additional Parts of the Study
On top of the above, participants were asked to:
- answer a question that made sure they were paying attention, and
- indicate how much they agreed with four statements about vaccines (e.g., "Overall I think vaccines are safe").
Findings and Discussion
However, the strength of the association had a lot to do with answer's stance towards vaccination, namely:
- The “who-said-it” factor was more closely associated with the scores of pro-vaccine answers (Figure 1a, orange lines with squares); in other words, in pro-vaccine answers, the author’s identity makes a relatively big difference in trustworthiness scores.
- by contrast, when it came to anti-vaccine answers (Figure 1a, blue lines with diamonds), the “who-said-it” factor had a smaller effect.
All in all, it seems that when it comes to vaccines, on average, health professionals are considered more trustworthy than parents.
Self-identifying as an authority has a relatively small effect but it still measurable, even among vaccine-hesitant raters.
Conclusion
By and large, internet users consider primarily "what is said" and (to a lesser extent) "who said it" when assessing online vaccine information.
Vaccine-hesitant internet users’ considerations seem to be more complex than those of other users, in this context. More research is needed to figure out how to better cater to this audience over the internet.
Figure 1. Epistemic trustworthiness ratings by answer's stance towards vaccination and answerer self-identification.
(a) Descriptive statistics of trustworthiness ratings for all raters, for vaccine-hesitant raters, and for vaccine-confident raters. Ratings are based on the mean of 14 questionnaire items on a 7-point Likert scale. Error bars denote the standard error of the mean.
(b) Moderation analyses predicting epistemic trustworthiness ratings. These analyses examine whether different variables predict trustworthiness ratings, and if so, to what extent, and whether the relationship between answerer self-identification and trustworthiness depends on the answer's stance towards vaccines. One moderation analysis was performed for all raters, and one analysis each for vaccine-hesitant raters and vaccine-confident raters. Each independent variable was attributed a standardized regression coefficient (beta coefficient), which describes the extent to which changes in that variable are related to changes in trustworthiness ratings, when holding all other variables constant, where changes in variables are measured in standard deviations.
One, two and three bullets (•, ••, •••) denote significance at the 0.05, 0.01 and 0.001 levels, respectively.
Acknowledgements
Some of the statistical analyses for this project were conducted by a freelance statistical consultant, Dr. Amihai Rigbi. I designed the experiment, wrote up the findings and created the visualizations.
Press Releases and Media Coverage
New Technion study identifies factors used when trusting vaccine info - Jerusalem Post, 16.3.2020
איך מעבירים מדע ברשת? - מכון דוידסון, 22.3.2020
חיסונים: האם אתם מאמינים למה שכתוב ברשת? - ynet, 15.3.2020
Citation
Sharon, A. J., Yom-Tov, E., & Baram-Tsabari, A. (2020). Vaccine information seeking on social Q&A services. Vaccine, 38(12), 2691–2699. https://doi.org/10.1016/j.vaccine.2020.02.010
