www culturalcognition net Motivated System 2 Reasoning and

  • Slides: 79
Download presentation
www. culturalcognition. net Motivated System 2 Reasoning and Science Curiosity: Two Individual Differences in

www. culturalcognition. net Motivated System 2 Reasoning and Science Curiosity: Two Individual Differences in Politically Motivated Reasoning Dan M. Kahan Yale University & many x 103 others

Behavioural Public Policy, in press Advances in Pol. Psych. , 38 179 -199

Behavioural Public Policy, in press Advances in Pol. Psych. , 38 179 -199

“Skin cream experiment”

“Skin cream experiment”

“Skin cream experiment”

“Skin cream experiment”

Two conditions

Two conditions

Correct interpretation of data correct rash decreases rash increases incorrect Numeracy score Lowess regression

Correct interpretation of data correct rash decreases rash increases incorrect Numeracy score Lowess regression line.

Covariance & Numeracy Bars denote 0. 95 CIs.

Covariance & Numeracy Bars denote 0. 95 CIs.

“Gun ban experiment”

“Gun ban experiment”

“Gun ban experiment”

“Gun ban experiment”

Four conditions

Four conditions

Correct interpretation of data skin treatment Numeracy score Gun ban Lowess regression line.

Correct interpretation of data skin treatment Numeracy score Gun ban Lowess regression line.

Correct interpretation of data skin treatment Numeracy score Gun ban Lowess regression line. Numeracy

Correct interpretation of data skin treatment Numeracy score Gun ban Lowess regression line. Numeracy score

Lowess regression line.

Lowess regression line.

Regression model for experiment results rash_decrease rash increase crime increase z_numeracy_x_rash_decrease z_numeracy_x_rash_increase z_numeracy_x_crime_increase z_numeracy

Regression model for experiment results rash_decrease rash increase crime increase z_numeracy_x_rash_decrease z_numeracy_x_rash_increase z_numeracy_x_crime_increase z_numeracy 2_x_rash_decrease z_numeracy 2_x_rash_increase z_numeracy 2_x_crime_increase Conserv_Repub_x_rash_decrease Conserv_Repub_x_rash_increase Conserv_Repub_x_crime_increase z_numeracy_x_Conserv_repub z_nuneracy_x_Conserv_Repub_x_rash_decrease z_nuneracy_x__x_rash_increase z_nuneracy_x__x_crime_increase _constant 0. 40 0. 06 1. 07 -0. 01 0. 55 0. 23 0. 46 0. 31 0. 02 -0. 07 -0. 31 -0. 64 0. 56 1. 28 0. 63 -0. 33 0. 54 0. 26 -0. 96 (1. 57) (0. 22) (4. 02) (-0. 05) (2. 29) (1. 05) (2. 01) (2. 46) (0. 14) (-0. 39) (-1. 75) (-3. 95) (2. 64) (6. 02) (2. 82) (-1. 89) (1. 40) (2. 17) (1. 08) (-4. 70) N = 1111. Outcome variable is “Correct” (0 = incorrect interpretation of data, 1 = correct interpretation). Predictor estimates are logit coefficients with z-test statistic indicated parenthetically. Experimental assignment predictors— rash_decrease, rash_increase, and crime_increase—are dummy variables (0 = unassigned, 1 = assigned—with assignment to “crime decreases” as the comparison condition. Z_numeracy and Conserv_Repub are centered at 0 for ease of interpretation. Bolded typeface indicates predictor coefficient is significant at p < 0. 05.

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases skin treatment rash increases 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases 25%, ± 10 skin treatment rash increases 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases rash decreases skin treatment rash increases 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases rash decreases skin treatment rash increases 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy 5%, ± 6 rash increases rash decreases rash decrease rash increases skin treatment 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases rash decreases rash decreases rash increases skin treatment 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases rash decreases rash decreases rash increases skin treatment 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases rash decreases rash decreases rash increases skin treatment 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases rash decreases rash decreases rash increases skin treatment 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data crime increases 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases rash decreases rash decreases rash increases skin treatment 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% crime decreases crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime increases 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases rash decreases rash decreases rash increases skin treatment 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% crime decreases crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crimedecreases crime decreases crime increases 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases rash decreases rash decreases rash increases skin treatment 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% crime decreases crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases increases crime increases 0% 10% 20% 30% 40% 50% crime decreases 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) high numeracy = 7 correct low numeracy = 3 correct Low numeracy High numeracy rash increases rash decreases rash decreases rash increases skin treatment 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% crime decreases crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% crime increases crime decreases crime increases 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) Low numeracy High numeracy Avg. “polarization” on crime data for high numeracy partisans 46% (± 17%) Avg. “polarization” on crime data for low numeracy partisans 25% (± 9%) crime decreases crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% high numeracy = 7 correct low numeracy = 3 correct 90% 100% probabilility of correct interpretation of data crime increases crime decreases crime increases 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Bounded rationality thesis

Bounded rationality thesis

BRT prediction Reality Liberal Democrat Conservative Republican

BRT prediction Reality Liberal Democrat Conservative Republican

BRT prediction Reality Liberal Democrat Conservative Republican

BRT prediction Reality Liberal Democrat Conservative Republican

BRT prediction Reality Liberal Democrat Conservative Republican

BRT prediction Reality Liberal Democrat Conservative Republican

BRT prediction Reality Conservative Republican Adapted from Kahan, D. M. , Peters, E. ,

BRT prediction Reality Conservative Republican Adapted from Kahan, D. M. , Peters, E. , Wittlin, M. , Slovic, P. , Ouellette, L. L. , Braman, D. & Mandel, G. The polarizing impact of science literacy and numeracy on perceived climate change risks. Nature Climate Change 2, 732 -735 (2012).

There is “solid evidence” of recent global warming due “mostly” to “human activity such

There is “solid evidence” of recent global warming due “mostly” to “human activity such as burning fossil fuels. ” [agree, disagree] Probability of “agree” Liberal Democrat Conservative Republican Ordinary Science Intelligence (percentile) N = 1600. shaded areas denote 0. 95 CIs.

Probability of “agree” There is “solid evidence” of recent global warming due “mostly” to

Probability of “agree” There is “solid evidence” of recent global warming due “mostly” to “human activity such as burning fossil fuels. ” [agree, disagree] Liberal Democrat Conservative Republican Actively Open-minded Thinking (zscore) N= 750. Derived from logistic regression. Colored barsdenote 0. 95 CIs.

Probability of “agree” There is “solid evidence” of recent global warming due “mostly” to

Probability of “agree” There is “solid evidence” of recent global warming due “mostly” to “human activity such as burning fossil fuels. ” [agree, disagree] Liberal Democrat Conservative Republican Cognitive Reflection Test (no. correct) N = 1600. Brackets denote 0. 95 CIs.

N = 1759. Shaded area denote 0. 95 CIs.

N = 1759. Shaded area denote 0. 95 CIs.

N = 1759. Shaded area denote 0. 95 CIs.

N = 1759. Shaded area denote 0. 95 CIs.

N = 1759. Shaded area denote 0. 95 CIs.

N = 1759. Shaded area denote 0. 95 CIs.

N = 1759. Shaded area denote 0. 95 CIs.

N = 1759. Shaded area denote 0. 95 CIs.

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) Low numeracy High numeracy Avg. “polarization” on crime data for high numeracy partisans 46% (± 17%) Avg. “polarization” on crime data for low numeracy partisans 25% (± 9%) crime decreases crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% high numeracy = 7 correct low numeracy = 3 correct 90% 100% probabilility of correct interpretation of data crime increases crime decreases crime increases 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data

Bounded rationality thesis

Bounded rationality thesis

Advances in Pol. Psych. , 38 179 -199

Advances in Pol. Psych. , 38 179 -199

www. culturalcognition. net

www. culturalcognition. net

State of the art “Science Curiosity”/“curiosity” Test of Science Related Attitudes (TOSRA) I am

State of the art “Science Curiosity”/“curiosity” Test of Science Related Attitudes (TOSRA) I am curious about the world in which we live I find it boring to hear about new ideas I would enjoy visiting a science museum at the weekend I would like to be given a science book as a present I get bored when watching science programs on TV

Sample self-report item

Sample self-report item

Sample self-report item

Sample self-report item

Sample (self-report) behavior

Sample (self-report) behavior

Sample (self-report) behavior

Sample (self-report) behavior

Performance measure

Performance measure

“Science Curiosity Scale” (SCS_1. 0 ) (item response theory 2 PL) probability of pro-curiosity

“Science Curiosity Scale” (SCS_1. 0 ) (item response theory 2 PL) probability of pro-curiosity response 1 read. 9 BSCIENCE book on science “scientific research or discoveries” Read book in last yr . 8 how closely follow news on “scientific research or discoveries” very closely Not at all A little but not closely . 7. 6. 5. 4. 3. 2. 1 0 -3 -2 -1 0 1 science curiosity Science Curiosity Index (SCI) 2 3 Science Curiosity Index (SCI)

YIF Clip: Origins of color vision

YIF Clip: Origins of color vision

probability of requesting full show Request full show: Evolution “believers” vs. “disbelievers” Evolution believer

probability of requesting full show Request full show: Evolution “believers” vs. “disbelievers” Evolution believer Evolution disbeliever Science Curiosity (percentie) Science Curiosity Logistic regression. Colored bars reflect 0. 95 level of confidence

Was the information “believable”?

Was the information “believable”?

Study 2 goals & results: Replication & robustness of SCS Mass Extinction: Life at

Study 2 goals & results: Replication & robustness of SCS Mass Extinction: Life at the Brink Darwin’s Dangerous Idea

Study 2 goals & results: Replication & robustness of SCS

Study 2 goals & results: Replication & robustness of SCS

Study 2 goals & results: Replication & robustness of SCS Mass Extinction: Life at

Study 2 goals & results: Replication & robustness of SCS Mass Extinction: Life at the Brink Darwin’s Dangerous Idea Daily Hollywood Rundown

Study 2 goals & results: Replication & robustness of SCS

Study 2 goals & results: Replication & robustness of SCS

Study 2 goals & results: Replication & robustness of SCS

Study 2 goals & results: Replication & robustness of SCS

WTF? #2: political polarization. .

WTF? #2: political polarization. .

Q. Are “curious” partisans more likely to examine surprising contrary evidence? 1. Unsurprising Nonskeptic

Q. Are “curious” partisans more likely to examine surprising contrary evidence? 1. Unsurprising Nonskeptic vs. Surprising Skeptical 2. Surprising skeptic vs. Unsurprising skeptical

Nonskeptical

Nonskeptical

Probability of selecting surprising skeptical vs. unsurprising nonskeptical story 24% (10%, 45%) Liberal Dem.

Probability of selecting surprising skeptical vs. unsurprising nonskeptical story 24% (10%, 45%) Liberal Dem. -1 SD science curiosity N =733. Monte Carlo simulation based on logistic regression. Parenthetical values represent boundaries of 0. 95 CI.

Probability of selecting surprising skeptical vs. unsurprising nonskeptical story 24% (10%, 45%) Liberal Dem.

Probability of selecting surprising skeptical vs. unsurprising nonskeptical story 24% (10%, 45%) Liberal Dem. -1 SD science curiosity 68% (46%, 85%) Liberal Dem. +1 SD science curiosity N =733. Monte Carlo simulation based on logistic regression. Parenthetical values represent boundaries of 0. 95 CI.

Probability of selecting surprising skeptical vs. unsurprising nonskeptical story 45% (± 31%) Liberal Dem.

Probability of selecting surprising skeptical vs. unsurprising nonskeptical story 45% (± 31%) Liberal Dem. -1 SD science curiosity Liberal Dem. +1 SD science curiosity N =733. Monte Carlo simulation based on logistic regression. .

Probability of selecting surprising nonskeptical vs. surprising nonskeptical story 42% (± 31%) Conserv. Repub.

Probability of selecting surprising nonskeptical vs. surprising nonskeptical story 42% (± 31%) Conserv. Repub. -1 SD science curiosity N =733. Monte Carlo simulation based on logistic regression. .

Probability of selecting surprising nonskeptical vs. surprising nonskeptical story 42% (± 31%) Conserv. Repub.

Probability of selecting surprising nonskeptical vs. surprising nonskeptical story 42% (± 31%) Conserv. Repub. -1 SD science curiosity 62% (± 13%) Conserv. Repub. +1 SD science curiosity N =733. Monte Carlo simulation based on logistic regression. .

Probability of selecting surprising nonskeptical vs. surprising nonskeptical story 20% (± 19%) Conserv. Repub.

Probability of selecting surprising nonskeptical vs. surprising nonskeptical story 20% (± 19%) Conserv. Repub. -1 SD science curiosity Conserv. Repub. +1 SD science curiosity N =733. Monte Carlo simulation based on logistic regression. .

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on

Monte carlo simulations Liberal Democrat (-1 SD on Conservrepub) Conserv Republican (+1 SD on Conservrepub) Low numeracy High numeracy Avg. “polarization” on crime data for high numeracy partisans 46% (± 17%) Avg. “polarization” on crime data for low numeracy partisans 25% (± 9%) crime decreases crime increases crime decreases crime increases Gun ban 0% 10% 20% 30% 40% 50% 60% 70% 80% high numeracy = 7 correct low numeracy = 3 correct 90% 100% probabilility of correct interpretation of data crime increases crime decreases crime increases 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% probabilility of correct interpretation of data

There is “solid evidence” of recent global warming due “mostly” to “human activity such

There is “solid evidence” of recent global warming due “mostly” to “human activity such as burning fossil fuels. ” [agree, disagree] Probability of “agree” Liberal Democrat Conservative Republican Ordinary Science Intelligence (percentile) N = 1600. shaded areas denote 0. 95 CIs.

Probability of “agree” There is “solid evidence” of recent global warming due “mostly” to

Probability of “agree” There is “solid evidence” of recent global warming due “mostly” to “human activity such as burning fossil fuels. ” [agree, disagree] Liberal Democrat Conservative Republican Actively Open-minded Thinking (zscore) N= 750. Derived from logistic regression. Colored barsdenote 0. 95 CIs.

Probability of “agree” There is “solid evidence” of recent global warming due “mostly” to

Probability of “agree” There is “solid evidence” of recent global warming due “mostly” to “human activity such as burning fossil fuels. ” [agree, disagree] Liberal Democrat Conservative Republican Cognitive Reflection Test (no. correct) N = 1600. Brackets denote 0. 95 CIs.

2 big questions:

2 big questions:

2 big questions: 1. What next?

2 big questions: 1. What next?

2 big questions: 1. What next? 2. Implications for effective science communication?

2 big questions: 1. What next? 2. Implications for effective science communication?