Rational Choice oder Framing?

Transcription

1 Hartmut Esser Rational Choice oder Framing? RCT und MFS als Ansätze zur Erklärung der Befunde aus den Experimenten zu Cooperation and Punishment in the Contribution to Public Goods von Ernst Fehr und Simon Gächter (1999) University of Mannheim Fakulty for Social Sciences Mannheim Centre for European Social Konferenz Analytische Soziologie: Theorie und Empirische Anwendungen Venice International University, San Servolo

2 ... a dèjá vu!

3 Fehr and Gintis 2007 (in: Annual Review of SOCIOLOGY) RCT? social motives! with punishment no punishment MFS? wide RCT! GTA!

4 Fehr and Gintis 2007 (in: Annual Review of SOCIOLOGY) RCT? social motives! with punishment no punishment MFS? Ministerium für Staatssicherheit Modell der Frame-Selektion wide RCT! GTA!

5 Fehr and Gintis 2007 (in: Annual Review of SOCIOLOGY) RCT? social motives! with punishment MFS no punishment!!!!!! Ministerium für Staatssicherheit Modell der Frame-Selektion wide RCT! GTA!

6 Framing and Priming

7 Framing

8 The Wallstreet-Game (Liberman, Samuels & Ross 2004) Activation of an accessible mental model by understanding a (verbal) cue against preferences! % Cooperation First Round Community Game W all Street Game "Most Likely Cooperators" "Most Likely Defectors" cheap talk? Choice?

9 Priming

10 Keizer, Lindenberg and Linda Steg person sweeping the street prosocial cue helping to pick up dropped oranges Learning? Activation of an accessible behavioral program by observation of a meaningful symbol! Choice?

11 The full (and true) story by Fehr&Gächter 1999

12 1. Version: No Punishment first

13 Contributions steady rise up to ca 100% WiP NoP 1. ca. 50% at start 3. strong jump up steady decay Rounds

14 2. Version: With Punishment first

15 Contributions same pattern 15.0 WiP 6. but small(er) decrease NoP Rounds

16 Contributions same pattern 4. steady rise up to ca 100% WiP 15.0 WiP 6. but small(er) decrease 10.0 NoP 1. ca. 50% at start NoP 3. strong jump up steady decay Rounds

17 Contributions 20.0 WiP 15.0 WiP 6. but small(er) decrease NoP 10.0 NoP Rounds

18 Summary

19 Findings 1. ca. 50% contribution at start 2. steady decay of conribution 3. strong jump up after announcement P 4. steady rise up to ca. 100% 5. general pattern for sequences similar 6. smaller decrease than increase Explanation RCT Explanation MFS

20 RCT and MFS

21 Aspects RCT MFS Mechanism formal axioms empirical interpretation Opportunities Beliefs (p) Preferences (U) Emotions Symbols/cues Effect on: Utility-function Type of Rationality Unconditionality social situation Processes Changes: short-term long-term

22 Aspects RCT MFS as-modus (PBB) rc-modus (RCT*) Mechanism formal axioms empirical interpretation Choice p*u future necessary by axioms Categorization M = f(o*a*l*c) past desirable by measurement Opportunities yes yes Beliefs (p) yes yes Preferences (U) yes yes Emotions (no) yes Symbols/cues Effect on: cheap talk beliefs Definition beliefs&preferences Utility-function one for all situation-dependent Type of Rationality fixed variable Unconditionality no yes social situation strategic interaction symbolic interaction Processes Sequences Sequences Changes: short-term long-term (Bayesian) Learning Internalisation Priming/Re-Framing Internalisation

23 Aspects RCT MFS as-modus (PBB) rc-modus (RCT*) Mechanism formal axioms empirical interpretation Choice p*u future necessary by axioms Categorization M = f(o*a*l*c) past desirable by measurement Opportunities yes yes Beliefs (p) yes yes Preferences (U) yes yes Emotions (no) yes Symbols/cues Effect on: cheap talk beliefs Definition beliefs&preferences Utility-function one for all situation-dependent Type of Rationality fixed variable Unconditionality no yes social situation strategic interaction symbolic interaction Processes Sequences Sequences Changes: short-term long-term (Bayesian) Learning Internalisation Priming/Re-Framing Internalisation

24 Aspects RCT MFS as-modus (PBB) Mechanism formal axioms empirical interpretation CHOICE p*u future necessary by axioms MATCH M = f(o*a*l*c) past desirable by measurement Opportunities yes yes Beliefs (p) yes yes Preferences (U) yes yes Emotions (no) yes Symbols/cues Effect on: cheap talk beliefs Definition beliefs&preferences Utility-function one for all situation-dependent Type of Rationality fixed variable Unconditionality no yes social situation strategic interaction symbolic interaction Processes Sequences Sequences Changes: short-term long-term (Bayesian) Learning Internalisation Priming/Re-Framing Internalisation rc-modus (RCT*) Object Accessibility Link Clarity

25 Aspects RCT MFS as-modus (PBB) Mechanism formal axioms empirical interpretation CHOICE p*u future necessary by axioms MATCH M = f(o*a*l*c) past desirable by measurement Opportunities yes yes Beliefs (p) yes yes Preferences (U) yes yes Emotions (no) yes Symbols/cues Effect on: cheap talk beliefs Definition beliefs&preferences Utility-function one for all situation-dependent Type of Rationality fixed variable Unconditionality no yes social situation strategic interaction symbolic interaction Processes Sequences Sequences Changes: short-term long-term (Bayesian) Learning Internalisation Priming/Re-Framing Internalisation rc-modus (RCT*) Object Accessibility Link Clarity deliberated Definition

26 Aspects RCT MFS as-modus (PBB) Mechanism formal axioms empirical interpretation CHOICE p*u future necessary by axioms MATCH M = f(o*a*l*c) past desirable by measurement Opportunities yes yes Beliefs (p) yes yes Preferences (U) yes yes Emotions (no) yes Symbols/cues Effect on: cheap talk beliefs Definition beliefs&preferences Utility-function one for all situation-dependent Type of Rationality fixed variable Unconditionality no yes social situation strategic interaction symbolic interaction Processes Sequences Sequences Changes: short-term long-term (Bayesian) Learning Internalisation Priming/Re-Framing Internalisation rc-modus (RCT*) Object Accessibility Link Clarity deliberated Definition

27 Aspects RCT MFS as-modus (PBB) Mechanism formal axioms empirical interpretation CHOICE p*u future necessary by axioms MATCH M = f(o*a*l*c) past desirable by measurement Opportunities yes yes Beliefs (p) yes yes Preferences (U) yes yes Emotions (no) yes Symbols/cues Effect on: cheap talk beliefs Definition beliefs&preferences Utility-function one for all situation-dependent Type of Rationality fixed variable Unconditionality no yes social situation strategic interaction symbolic interaction Processes Sequences Sequences Changes: short-term long-term (Bayesian) Learning Internalisation Priming/Re-Framing Internalisation rc-modus (RCT*) Object Accessibility Link Clarity deliberated Definition

28 Aspects RCT MFS as-modus (PBB) Mechanism formal axioms empirical interpretation CHOICE p*u future necessary by axioms MATCH M = f(o*a*l*c) past desirable by measurement Opportunities yes yes Beliefs (p) yes yes Preferences (U) yes yes Emotions (no) yes Symbols/cues Effect on: cheap talk beliefs Definition beliefs&preferences Utility-function one for all situation-dependent Type of Rationality fixed variable Unconditionality no yes social situation strategic interaction symbolic interaction Processes Sequences Sequences Changes: short-term long-term (Bayesian) Learning Internalisation Priming/Re-Framing Internalisation rc-modus (RCT*) Object Accessibility Link Clarity deliberated Definition

29 Aspects RCT MFS as-modus (PBB) Mechanism formal axioms empirical interpretation CHOICE p*u future necessary by axioms MATCH M = f(o*a*l*c) past desirable by measurement Opportunities yes yes Beliefs (p) yes yes Preferences (U) yes yes Emotions (no) yes Symbols/cues Effect on: cheap talk beliefs Definition beliefs&preferences Utility-function one for all situation-dependent Type of Rationality fixed variable Unconditionality no yes social situation strategic interaction symbolic interaction Processes Sequences Sequences Changes: short-term long-term (Bayesian) Learning Internalisation Priming/Re-Framing Internalisation rc-modus (RCT*) Object Accessibility Link Clarity deliberated Definition

30 Aspects RCT MFS as-modus (PBB) Mechanism formal axioms empirical interpretation CHOICE p*u future necessary by axioms MATCH M = f(o*a*l*c) past desirable by measurement Opportunities yes yes Beliefs (p) yes yes Preferences (U) yes yes Emotions (no) yes Symbols/cues Effect on: cheap talk beliefs Definition beliefs&preferences Utility-function one for all situation-dependent Type of Rationality fixed variable Unconditionality no yes social situation strategic interaction symbolic interaction Processes Sequences Sequences Changes: short-term long-term Updating beliefs Internalisation Activation frames Internalisation rc-modus (RCT*) Object Accessibility Link Clarity deliberated Definition

31 Theoretical reconstruction

32 ... remember

33 Contributions same pattern 4. steady rise up to ca 100% WiP 15.0 WiP 6. but small(er) decrease 10.0 NoP 1. ca. 50% at start NoP 3. strong jump up steady decay Rounds

34 Findings 1. ca. 50% contribution at start 2. steady decay of conribution 3. strong jump up after announcement P 4. steady rise up to ca. 100% 5. general pattern for sequences similar 6. smaller decrease than increase Explanation RCT Explanation MFS

35 Explanations

36 Finding 1. ca. 50% contribution at start 2. steady decay of conribution 3. strong jump up after announcement PUN 4. steady rise up to ca. 100% 5. general pattern for sequences similar 6. smaller decrease than increase Explanation RCT EP- and AP-Types conditional cooperation: updating beliefs (only) common knowledge of P with WiP-Option common knowledge of P with WiP-Option same (RCT-)effects in both versions Explanation MFS situation not defined and EF-/AF- Types Mis-Match of A-Frame for AF- Types Activation of R-Frame with announcement oft WiP-Option Increase of match for R-Frame strong match of R-Frame in both cases

37 Contributions WiP MFS Priming C-Frame increased match C-Frame! WiP 6. but small(er) decrease 10.0 NoP NoP... Explanation for down after WiP first and with annoncement NoP? RCT common knowledge NoP no priming down to round 1 with NoP(first) Rounds?

38 Contributions 20.0 RCT MFS WiP 15.0 WiP 6. but small(er) decrease 10.0 NoP? NoP!... Explanation for down after WiP first and with annoncement NoP? Rounds

39 Finding 1. ca. 50% contribution at start 2. steady decay of conribution 3. strong jump up after announcement PUN 4. steady rise up to ca. 100% 5. general pattern for sequences similar 6. smaller decrease than increase Explanation RCT EP- and AP-Types conditional cooperation: updating beliefs (only) common knowledge of P with WiP-Option common knowledge of P with WiP-Option same (RCT-)effects in both versions Explanation MFS situation not defined and EF-/AF- Types Mis-Match of A-Frame for AF- Types Activation of R-Frame with announcement oft WiP-Option Increase of match for R-Frame strong match of R-Frame in both cases

40 Finding 1. ca. 50% contribution at start 2. steady decay of conribution 3. strong jump up after announcement PUN 4. steady rise up to ca. 100% 5. general pattern for sequences similar 6. smaller decrease than increase Explanation RCT EP- and AP-Types conditional cooperation: updating beliefs (only) common knowledge of P with WiP-Option common knowledge of P with WiP-Option same (RCT-)effects in both versions no explanation Explanation MFS situation not defined and EF-/AF- Types Mis-Match of A-Frame for AF- Types Activation of R-Frame with announcement oft WiP-Option Increase of match for R-Frame strong match of R-Frame in both cases Priming of R-Frame with sequence of cooperation before

41 Finding 1. ca. 50% contribution at start 2. steady decay of conribution 3. strong jump up after announcement PUN 4. steady rise up to ca. 100% 5. general pattern for sequences similar 6. smaller decrease than increase Explanation RCT EP- and AP-Types conditional cooperation: updating beliefs (only) common knowledge of P with WiP-Option common knowledge of P with WiP-Option same (RCT-)effects in both versions Explanation MFS situation not defined and EF-/AF- Types Mis-Match of A-Frame for AF- Types Activation of R-Frame with announcement oft WiP-Option Increase of match for R-Frame strong match of R-Frame in both cases?!

42 A (separate) Test

43 ... remember

44 Contributions 20.0 WiP 15.0 WiP 6. but small(er) decrease NoP 10.0 NoP Rounds

45 Hypotheses RCT and MFS for changes WiP to NoP for E- and R-Types

46 Contributions RCT MFS 20.0 R-Type no priming Priming E-Type no priming no priming RCT Priming not in tool box Priming, but dependent of accessibility MFS Rounds

47 Measurement of Types: Contribution in first round!

48 Findings

49 Contributions 20.0 R-Type RCT MFS E-Type Rounds

50 Contributions 20.0 RCT MFS 15.0 R-Type! 10.0 E-Type MFS explains a (crucial) anomaly for any version of RCT... Activation of an internalized program by mere observation situational cues Rounds no change of preferences (by learning or choice)

51 the complete picture

52 Contributions R-Type R-Type E-Type E-Type WiP NoP Community... jump : not neccessarily by framing alone! Mechanical Solidarity WiP NoP Rounds

53 ... and also:

54 Zürich&St.Gallen pooled (Fehr&Gächter 1999 plus Hermann, Thöny and Gächter 2008)

55 EF- und RF-Typen Zürich&St. Gallen... and for WiP to NoP even stronger WiP1 RF WiP2 RF WiP2 EF... now: morethan-ever for R-Types! NoP1 RF NoP2 RF 100 WiP1 EF NoP1 EF NoP2 EF interaction-effect with accessibility also for NoP to WiP!

56 Why important? A really General Theory of (Inter-)Action!... last point: Integrated AS!

57

58

59 Appendix 1: F&G and Variants of RCT

60 theoretical predictions and empirical fit RCT MFS Finding RCT 1.0 (E only) 1 ca. 50% contribution at start 2 steady decay of contribution 3 strong jump up for WiP (2) 4 steady rise up to ca. 100% 5 similar for WiP(1) and WiP(2) 6 small(er) decrease for NoP(2)

61 theoretical predictions and empirical fit RCT MFS Finding RCT 1.0 (E only) 1 ca. 50% contribution at start refuted 2 steady decay of contribution refuted 3 strong jump up for WiP (2) refuted 4 steady rise up to ca. 100% refuted 5 similar for WiP(1) and WiP(2) refuted 6 small(er) decrease for NoP(2) refuted

62 theoretical predictions and empirical fit RCT MFS Finding RCT 1.0 (E only) RCT 2.0 (E and A) 1 ca. 50% contribution at start refuted 1 2 steady decay of contribution refuted 2 3 strong jump up for WiP (2) refuted refuted 4 steady rise up to ca. 100% refuted refuted 5 similar for WiP(1) and WiP(2) refuted 5 6 small(er) decrease for NoP(2) refuted refuted

63 theoretical predictions and empirical fit RCT MFS Finding RCT 1.0 (E only) RCT 2.0 (E and A) RCT 3.0 (E and R) 1 ca. 50% contribution at start refuted steady decay of contribution refuted strong jump up for WiP (2) refuted refuted 3 4 steady rise up to ca. 100% refuted refuted 4 5 similar for WiP(1) and WiP(2) refuted small(er) decrease for NoP(2) refuted refuted refuted

64 theoretical predictions and empirical fit RCT MFS Finding RCT 1.0 (E only) RCT 2.0 (E and A) RCT 3.0 (E and R) as-modus (PPB) 1 ca. 50% contribution at start refuted steady decay of contribution refuted strong jump up for WiP (2) refuted refuted steady rise up to ca. 100% refuted refuted similar for WiP(1) and WiP(2) refuted small(er) decrease for NoP(2) refuted refuted refuted 6

65 theoretical predictions and empirical fit RCT MFS Finding RCT 1.0 (E only) RCT 2.0 (E and A) RCT 3.0 (E and R) as-modus (PPB) rc-modus (RCT*) 1 ca. 50% contribution at start refuted (1) 2 steady decay of contribution refuted (2) 3 strong jump up for WiP (2) refuted refuted 3 3 (3) 4 steady rise up to ca. 100% refuted refuted 4 4 (4) 5 similar for WiP(1) and WiP(2) refuted (5) 6 small(er) decrease for NoP(2) refuted refuted refuted 6 (6)

66 Appendix 2: St. Gallen and Zürich

67 NoP STG 100 NoP ZUR

68 WiP STG 160 WiP ZUR

69 WiP STG 160 WiP ZUR NoP STG 100 NoP ZUR

70 Appendix 3: EF- vs. AF-Types Zürich&St.Gallen pooled

71 EF- und RF-Typen Zürich&St. Gallen WiP1 RF WiP2 RF NoP1 RF WiP1 EF NoP1 EF NoP2 RF NoP2 EF WiP2 EF