Presentation is loading. Please wait.

Presentation is loading. Please wait.

Repetition I  Causes of biases in contingency and causal judgments:  Pattern recognition capabilities.  Belief in the law of small numbers.  Hot hand.

Similar presentations


Presentation on theme: "Repetition I  Causes of biases in contingency and causal judgments:  Pattern recognition capabilities.  Belief in the law of small numbers.  Hot hand."— Presentation transcript:

1 Repetition I  Causes of biases in contingency and causal judgments:  Pattern recognition capabilities.  Belief in the law of small numbers.  Hot hand in basketball.  Regression to the mean.  Incorrect weighting of contingency information.  Attentional and saliency effects.

2 Repetition II  ANOVA model (Kelley, 1967)  3 possible causes influencing behavior: Person, Object, and Situation  3 sources of variance are analyzed:  Consensus: Variation due to the person.  Distinctness: Variation due to the object.  Consistency: Variation due to the situation.

3 Repetition III  ANOVA Model: Shortcomings:  Data driven model: Ignoring the impact of personal theories.  Ignoring motivational aspects.  Ignoring saliency or attentional effects.  Various sources of variance are weighted differently or are ignored all together.

4 Ignoring relevant influences  Example 1: Halo effect:  Empirical result.  Explanation of halo effects: Examples.  Example 2: Anchor effects.  Example 3: ignoring consensus information (base rate information).

5 Overestimating irrelevant influences  Example: Assessment of the effect or reassurance.  Fundamental attribution error: Erroneous attribution to personal factors.  Influence of Saliency: Attributional asymmetry.

6 Problem of explanationby means of saliency  Saliency need itself explanation.  Cultural aspects of attentional and saliency effects.  Evolutionary roots of attentional and salience effects.

7 2.6 Methodological aspects: Overview  Causal reasoning: methodology  Simpson’s paradox.  Ecological fallacy.  Regression to the mean.

8 2.6.1 Causal reasoning  Three important activities (in science and everyday life): 1.Diagnosis. 2.Prediction. 3.Causal explanation.  Problem of inferring causal relations:  Correlations and spurious effects  Confounding

9 2.6.1 Causal reasoning  Common causes /Confounding:

10 2.6.1 Causal reasoning  Example of confounding: Full moon and car accidents:

11 2.6.1 Causal reasoning  Common causes: Factor analytic model:

12 Causal reasoning: Method  Experimental method and causal reasoning: Claim:  Causal relations can be only by established by means of experiments (and not via observational studies)  This is due to the following characteristics of experiments: 1.Random assignment of units to experimental conditions. 2.Balancing / Parallelization.

13 Causal reasoning: Method  Example: placebo-controlled, double blind studies with random assign- ment of units to treatments.

14 Causal reasoning: Method Question: Is it possible to assess the causal influence of a target cause on an outcome variable without doubt (e.g. by using sophisticated experimental methods)? Answer: No! Argument: It is not possible to exclude counfounders without doubt.

15 Causal reasoning: Method  Conclusion: It is impossible to establish causal relations without doubt (David Hume).  Alternative Approach (Karl Popper):  Inferring causal relation is but a specific case of inductive inference.  Inductive inference cannot be justified.  Causal theories and assumptions can be tested.

16 Causal reasoning: Method  Assessing causal theories / assump- tions /models:  Causal models implicate pattern of co- variances.  Specifically: Causal models implicate missing (partial) covariances.

17 Causal reasoning: Method  Assessing causal theories / assump- tions /models:  Important Problem: Equivalent causal models: Models the make the same predictions (or no predictions). X1X1 X3X3 X2X2 X4X4 X1X1 X3X3 X2X2 X4X4 X1X1 X3X3 X2X2 X4X4

18 Causal reasoning: Method  Bad practice in causal modeling (as well as other branches): Explorative models are termed as confirmatory in publications.

19 Simpson’s Paradox: Example 1 Death sentence GroupYesNo  %YesYule’s Q Black59244825072.4-0.16 White72218522573.2  13146334764

20 Simpson’s Paradox: Example 1 Color Death sentence VictimDelinquent YesNo  %YesYule’s Q Black 11220922200.051.00 White 0111 0.00 WhiteBlack 4823928716.700.71 White 72207421463.40  13146334764

21 Simpson’s Paradox: Example 2 Successful  LocalityTreatmentYesNo% SuccessYule’s Q Goat-townNew2018020010%0.36 Old5951005% Cow-cityNew901010090%0.50 Old1505020075%  265335600

22 Simpson’s Paradox: Example 2 Successful  TreatmentYesNo% SuccessYule’s Q New11019030037%-0.30 Old15514530052%  265335600

23 Simpson’s Paradox: Example 3 Success FieldSexYesNo  % SuccessYules Q Social workMan1273516278%-0.20 Woman2753284% PsychologyMan17425929%-0.14 Woman9217026235%  263252515

24 Simpson’s Paradox: Example 3 Success SexYesNo  % SuccessYules Q Man1447722165%0.47 Woman11917529440%  263252515


Download ppt "Repetition I  Causes of biases in contingency and causal judgments:  Pattern recognition capabilities.  Belief in the law of small numbers.  Hot hand."

Similar presentations


Ads by Google