Lab Experimentation Prof. Xiao-Ping Chen University of Washington Lab Experiment: the tradeoff Advantages Testing causal effects Allow for rigorous control to rule out alternative explanations Isolate the phenomenon, focus on key variables, and eliminate/ neutralize the influence of irrelevant extraneous variables Conducive to examine mediating processes Precision of measurement Study behavior itself instead of self-reports of behavior 2 Lab experiment: the tradeoff Disadvantages Reactivity of the participant Limitations posed by ethics Human subjects committee Consent from participants Artificiality of environment Seldom suited to studying long-term effects 3 Should I do a lab experiment? Depends on your research objective Parameter estimation research Hypothesis testing research Whether causal effect is of the primary interest 4 Cover story and experimental context 1. Experimental realism is the key, mundane realism is a plus 2. Provide a believable context for introducing manipulations, experimental tasks, and dependent measures and must be credible throughout the experiment 3. Hide research hypotheses from the subjects (i.e., guard demand characteristics) 4. Try to reduce the reactivity of the participant (e.g., evaluation 5 apprehension) Experimental manipulation 1. Construct validity is the key 2. Investigate previous research to see how they did it and then use your creativity 3. Use multiple operations/ methods if possible 6 Team Cooperation Prisoner’s dilemma Your Partner’s Choice C D $7 Your C Choice $7 $10 $2 $2 D $10 $4 $4 8 Social Dilemma Mean Contribution of Other Members Your Contribution 0 1 2 3 … 10 0 1 2 3 … 10 10 9.33 8.67 8.0 … 3.33 11.67 11.0 10.33 9.67 … 5.0 13.33 12.67 12.0 11.33 … 6.67 15.0 … 26.67 14.33 … 26.0 13.67 … 25.33 13.0 … 24.67 … … … 8.33 … 20.0 Your Payoff = (10 – C) + 2* [(n-1) MC + C] / n 9 Pillutla and Chen (1999): Social norms and cooperation in social dilemmas: The effects of context and feedback Experimental Implicit Condition Norm Economic Non-Econ. Perceived Norm Comp. Coop.or comp. Coop. Coop. or comp. 10 Pillutla and Chen (1999) 实验条件 1 (不合作反馈) 2 (合作反馈) 3 (无反馈) 4 (合作反馈) 5 (不合作反馈) 6 (无反馈) 第一阶段 第二阶段 经济情景 32.35% 29.68 27.54 社会情景 29.56 49.39 36.52 社会情景 43.04 39.61 32.08 经济情景 51.67 11.69 39.58 11 Chen, Bussmann, & Bachrach (2007): Reciprocity in ultimatums: the effects of fairness perceptions, negative emotions, and partner closeness Multi-operations of reciprocity: Acceptance/ rejection decision Reciprocating amount Intention for a long-term relationship 12 Chen, Bussmann, & Bachrach (2007): Reciprocity in ultimatums: the effects of fairness perceptions, negative emotions, and partner closeness Experimental design: 2 x 3 factorial Proposed amount: Partner relationship: 35%, 50%, or 65% stranger or friend Measures: Fairness perception Negative emotion Partner closeness 13 Chen, Bussmann, & Bachrach (2007): Reciprocity in ultimatums: the effects of fairness perceptions, negative emotions, and partner closeness Hypothesized structural model with parameter estimates FP1 FP2 FP3 FI1 β53=.29 (.17) Fairness Perceptions Initial Offer – DV1 γ32=-.25 (.14) Initial Offer – DV2 β54=-.35* (.14) β43=-.75* (.15) γ41=.98* (.41) Negative Emotions β64=.29 (1.31) γ42=.12 (.16) NE1 NE2 NE3 NE4 NE5 FI3 Willingness for future relationship β63=2.99 (1.59) γ31=-2.70* (.22) FI2 Counter Offer NE6 Unstandardized coefficients are provided with standard error in parentheses. * p < .05 Bold arrows indicate a significant relationship at p = .05 14 Chen, Bussmann, & Bachrach (2007): Reciprocity in ultimatums: the effects of fairness perceptions, negative emotions, and partner closeness Post-hoc structural model with parameter estimates FP1 FP2 FP3 FI1 β53=.29 (.17) Fairness Perceptions Initial Offer1 β54=-.35* (.14) β43=-.76* (.15) Negative Emotions γ41=.91* (.39) NE1 NE2 NE3 NE4 β64=.29 (1.30) NE5 FI3 Willingness for future relationship β63=3.03 (1.61) γ31=-2.55* (.22) FI2 Counter Offer NE6 Unstandardized coefficients are provided with standard error in parentheses. * p < .05 Bold arrows indicate a significant relationship at p = .05 1 Initial offer represented by first dummy variable (30% vs. 50% & 70%) 15 Conclusion Experiments can be used to rigorously test causal effects of variables; and to test mediating or moderating effects Experiments are a lot of fun! 16