Thread by @DG_Rand

🚨New working paper!🚨

"Fighting COVID-19 misinformation on social media:
Experimental evidence for a scalable accuracy nudge intervention"

We test if an intervention we developed for political fake news works for #COVID19- seems like YES!

PDF: psyarxiv.com/uhbk9/

1/

Previously we found people share political misinfo b/c social media distracts them from accuracy- NOT b/c they cant tell true v false, NOT b/c they dont care about accuracy

So nudging them to think about accuracy improved quality of news they shared!

https://twitter.com/DG_Rand/status/1196171145227251712

2/

Like everyone else we're losing sleep over #COVID19

To try to feel (slightly) useful, we decided to see how similar COVID-19 misinfo was to political misinfo from a cog psych perspective- and if the accuracy nudge we'd come up with might help fight COVID-19 misinfo online

3/

We ran survey exps on @lucid_hq recruiting Americans, quota-sampled to match US distribution on age gender ethnicity and region

We showed them true & false stories about COVID-19 in FB format- photo headline source (false sample below)

Headlines here
osf.io/7d3xh/

4/

Study 1

Asked N=853 participants for accuracy judgments OR sharing intentions

⟶ Accuracy: people rate false COVID-19 headlines as much less accurate than true

⟶ Sharing: ~equally likely to share false v true

50% more likely to share false misinfo than to believe it!

5/

Also:

Higher Cognitive Reflection & science knowledge🡆Greater discernment for belief+sharing
CRT🡆less concern about COVID-19

Predisposed to seek medical care🡆More belief in false; more sharing of both; more concern

Republicans 🡆 Less sharing of true info; less concern

6/

Study 2=INTERVENTION!

Can nudging people to think about accuracy make them more discerning in their sharing?

Treatment: People rate accuracy of 1 non-COVID-19 headline at outset of study (ostensibly as pretest for other study)

Primes concept of accuracy/fights distraction

7/

RESULTS

It worked! Sharing discernment in treatment 2.5x higher than in control

Item-level: treatment effect on sharing proportional to headline's perceived accuracy in S1
🡆Treatment makes people more likely to think about accuracy when deciding what COVID-19 info to share

8/

CONCLUSION

Sharing of COVID-19 info is similar to political news

Inattention is big part of the problem- as opposed to lack of digital literacy or "post-truth"ness

Nudging people to consider accuracy= scalable intervention platforms could easily stack w other approaches

9/

Huge thanks to tireless coauthors @GordPennycook @JonMcphetres Zhang- was a sprint, from conception to working paper in 10 days

Tried to be extremely careful despite speed-but pls let us know if you find any mistakes!!/Comments much appreciated

Data
osf.io/7d3xh/
/end

Perhaps of interest to @NAChristakis @CassSunstein @R_Thaler @davidlazer @j_a_tucker @sinanaral @deaneckles @katy_milkman @Mantzarlis @JuddLegum @natematias @persily @alexleavitt @STWorg @Sander_vdLinden @jayvanbavel @RobbWiller @william__brady @mollycrockett

OH also, we are planning on replicating this study in China in the immediate future - so any feedback or suggestions prior to our next run would be particularly useful!

Whoops forgot @andyguess @BrendanNyhan @jplotkin @JasonReifler

Another addendum: you might be worried that the effect of our intervention won't generalize b/c we use a self-report sharing measure.

But there are two reasons to be hopeful that the result from the survey experiment WILL generalize to actual sharing:

1) Self-report sharing correlates with actual sharing (headlines Turkers say they will share more get more actual shares on Twitter)
journals.plos.org/plosone/articl…

2) For political misinfo, we show the accuracy nudge improves actual sharing in Twitter field exp

https://twitter.com/DG_Rand/status/1196171188613132290?s=20