Do contingency estimates inform our causal judgments? A survey of controversial health-related beliefs
Julie Y.L. Chow1, Ben Colagiuri1, Benjamin M. Rottman2, Micah B. Goldwater1 & Evan J. Livesey1 1School
of Psychology, University of Sydney 2Department of Psychology, University of Pittsburgh
Address correspondence to:
[email protected]
Background
•
Laboratory research has found that under zero contingency, i.e. p(recovery|treatment) = p(recovery|no treatment), manipulations that inflate causal judgments do not influence contingency estimates to the same extent (e.g. Chow et al., 2018). To date, there is no direct comparison between people’s contingency estimates and their causal judgments about realworld health beliefs, particularly when the relationship between the putative cause (e.g. treatment) and its effect (e.g. health recovery) is potentially non-contingent.
Causal belief ratings Causal belief ratings
60 80 40 60 20 40 -1.0
200.0
-0.5
0.5
1.0
2
r change = .314, -1.0 -0.5 p < .001 BFM = 8.75 x
109
0.0
0.5
2
r change = .146, -1.0 -0.5p < .001
1.0
BFM = 44781
1.0
0.0
0.5
1.0
r2 change = .146, p < .001 BFM = 44781
80 100
60 80 40 60 20 40
-1.0
200.0
-0.5
0.5
60 80
BFM = 514
0.0
200.0
-0.5
r2 change = .121, -1.0 -0.5 p < .001
1.0
BFM = 9797
Contingency Estimates
r2 change = .140, p < .001 BFM = 514
Regular exercise increases lifespan 100
20 40
0.5
1.0
40 60 20 40
-1.0
20 0.0
-0.5
-1.0 -0.5 p = .033 r2 change = .036,
1.0
BFM = 2.62
r2 change = .238, p < .001 BFM = 1.797 x 107
r2 change = .036, p = .033 BFM = 2.62
Childhood vaccination causes autism Childhood vaccination causes autism
BFM = 1.48
N.S.
0.0
Contingency Estimates
1.0
Causal belief ratings Causal belief ratings 0.5
1.0
60 40 40 20
-1.0
-0.5
20 0.0
0.5
1.0
0.5
1.0
Contingency Estimates 0.5
1.0
2) Of those people you know who experience back pain and visit a chiropractor, what percentage of them have experienced an improvement in their condition?
-1.0 N.S.
BFM = 2.60
-0.5
0.0
50%
100%
Estimating p(back pain relief | no chiropractic) = 3) Of those people you know who experience back pain but do not visit a chiropractor, what percentage of them have experienced an improvement in their condition? 50%
100%
To what extent do you agree that chiropractic therapy is an effective treatment for back pain? 0 Strongly disagree
50 Neither agree nor disagree
100 Strongly agree
Conclusion • Causal judgments reflect contingency estimates, particularly for complementary health treatments.
80 60
Contingency Estimates -0.5
0.5
100 80
40 20
-1.0 N.S.
0.0
100 Wi-Fi causes cancer
60 40
-0.5
1.0
Contingency Estimates
100 80
-1.0
0.5
Radiation from mobile phones and Wi-Fi causes cancer Radiation from mobile phones and
100
80 60
Estimating p(back pain relief | chiropractic) =
Causal judgment ratings:
Contingency Estimates 0.5
100%
qN/A I don’t know anyone with back pain
0%
60 80
BFM = 1.797 x 107Contingency Estimates
20 0.0
1.0
Exposure to wind turbines can cause adverse health effects Exposure to wind turbines can cause 100 adverse health effects
Contingency Estimates 0.0
0.5
r2 change = .121, p < .001 BFM = 9797
Causal belief ratings Causal belief ratings
40 60
-1.0 -0.5 p < .001 r2 change = .238,
0.0
80 100
60 80
20 0.0
1.0
Contingency Estimates
80 100
-0.5
0.5
Contingency Estimates 0.5
50%
0%
20 40
Contingency Estimates r2 change = .140, -1.0 -0.5 p < .001
0%
40 60
-1.0
1.0
p(back pain relief | chiropractic) - p(back pain relief | no chiropractic)
Contingency Estimates
80 100
Causal belief ratings Causal belief ratings
Chiropractic therapy is an effective treatment for back pain Herbal remedies are effective in treating the common cold Acupuncture is an effective treatment for prolonged pain Prolonged smoking causes heart, lung & breathing problems Regular exercise increases lifespan Exposure to wind turbines can cause adverse health effects Childhood vaccination causes autism Radiation from mobile phones and Wi-Fi causes cancer
0.5
Acupuncture is an effective treatment Prolonged smoking causes heart, lung for persistent pain and breathing problems Acupuncture is an effective treatment Prolonged smoking causes heart, lung 100 100 for persistent pain and breathing problems
Contemporary Opinions on Factors that Impact Human Health
Ø Ø Ø Ø Ø Ø Ø Ø
200.0
Contingency Estimates
r2 change = .314, p < .001 BFM = 8.75 x 109
-1.0
Eight target health-related beliefs that differed in degree of consensus within the community:
20 40 -0.5
Contingency estimate, Δp (Allan, 1980) =
1) Of all the people you know, what percentage of people visit a chiropractor for back pain?
40 60
-1.0
Contingency Estimates
Causal belief ratings Causal belief ratings
Do people’s reported beliefs about controversial treatment-outcome relationships reflect their contingency estimates?
60 80
Contingency Estimates
Regular exercise increases lifespan
Research Question
80 100
Causal belief ratings Causal belief ratings
•
The perceived contingency between events, i.e. the difference in the probability of the outcome occurring with and without the putative cause present, is thought to be important in determining cause-effect relationships, e.g., the extent to which a treatment is effective in improving health recovery.
80 100
Causal belief ratings Causal belief ratings
•
Health treatment choices, like many decisions we make in everyday life, are motivated by beliefs about cause and effect. The efficacy of available treatments varies widely, and some very popular choices may be completely ineffective.
Exemplar of critical measures
Herbal remedies are Estimates effective in Chiropractic therapy is an effective Regression of Causal Ratings on Contingency treating the common cold treatment for back pain Herbal remedies are effective in Chiropractic therapy is an effective 100 100 treating the common cold treatment for back pain
Causal belief ratings Causal belief ratings
•
Results
Contingency Estimates
N.S.
*Beliefs with very high frequency of cue exposure (>80% of everyone BFM = 1.48 BFMthey = 2.60know) and very low frequency of perceived exposure to the outcome (