Read on Twitter
In #scicomm, does communicating uncertainty have negative effects?
Our review [now out in @SciPublic] finds that in most cases, the answer is NO.
But there is an important exception where it is clearly harmful:
uncertainty in the form of DISAGREEMENT/CONFLICT.
[thread]
1/x
![In #scicomm, does communicating uncertainty have negative effects?Our review [now out in @SciPublic] finds that in most cases, the answer is NO. But there is an important exception where it is clearly harmful:uncertainty in the form of DISAGREEMENT/CONFLICT.[thread]1/x](https://pbs.twimg.com/media/EdY2fR1X0AMqXie.png "In #scicomm, does communicating uncertainty have negative effects?Our review [now out in @SciPublic] finds that in most cases, the answer is NO. But there is an important exception where it is clearly harmful:uncertainty in the form of DISAGREEMENT/CONFLICT.[thread]1/x")
Our review [now out in @SciPublic] finds that in most cases, the answer is NO.
But there is an important exception where it is clearly harmful:uncertainty in the form of DISAGREEMENT/CONFLICT.
[thread]
1/x
Experiments covered in our review (k=48) have found a variety of positive and negative effects (on audience perceptions / responses) of portraying uncertainty about science.
But most of these experiments used *very* different types of uncertainty as the manipulations.
2/x
But most of these experiments used *very* different types of uncertainty as the manipulations.
2/x
We can categorize uncertainty portrayals into four types:
A) quantified uncertainty (a MoE, a probability)
B) known unknowns (limitations, alt explanations, gaps)
C) unknown unknowns (irreducible bc knowledge is tentative forever)
D) conflict among the experts or evidence
3/x
A) quantified uncertainty (a MoE, a probability)
B) known unknowns (limitations, alt explanations, gaps)
C) unknown unknowns (irreducible bc knowledge is tentative forever)
D) conflict among the experts or evidence
3/x
We find that Type A has rarely had negative effects (only positive or null).
That is, in these exps, people usually DO NOT respond negatively to error bars, probabilistic forecasts, and other quantified uncertainties (relative to a ctrl w/o uncertainty).
4/x
That is, in these exps, people usually DO NOT respond negatively to error bars, probabilistic forecasts, and other quantified uncertainties (relative to a ctrl w/o uncertainty).
4/x
Types B and C have mixed results, but generally there is no reason to expect that people are necessarily going to respond negatively when they hear about them.
So we should feel more free to state what we don't know! And that we also don't know what we don't know!
5/x
So we should feel more free to state what we don't know! And that we also don't know what we don't know!
5/x
However... Type D.
Most of the findings of negative effects of uncertainty in #scicomm come from one type:
Portraying conflict/disagreement among the experts/evidence.
Portraying this type of uncertainty consistently harms credibility, beliefs, and behavioral intentions.
6/x
Most of the findings of negative effects of uncertainty in #scicomm come from one type:
Portraying conflict/disagreement among the experts/evidence.
Portraying this type of uncertainty consistently harms credibility, beliefs, and behavioral intentions.
6/x
Overall, this is very encouraging. We should be more confident that openly sharing the inevitable uncertainties of science is unlikely* to harm the effectiveness of the message.
*important exception noted above.
7/x
*important exception noted above.
7/x
Our review also finds that individual-level variables (ideology, prior beliefs about the topic) are often found to moderate the effects of uncertainty portrayals -- and often form a pattern of confirmation bias.
Much more work is needed in this fascinating area!
8/x
Much more work is needed in this fascinating area!
8/x
The full article is available here:
https://bit.ly/2WDSTBw
Much appreciation to my intrepid-but-Twitterless coauthor Ron Rice
https://bit.ly/2WDSTBw
Much appreciation to my intrepid-but-Twitterless coauthor Ron Rice
