6. Game Theory Flashcards
Describe set up of Guth Et Al 1982 experiment
-ultimatum game
-subjects were in one room but no subject knew the person they were paired with
-real monetary stakes (pie of roughly £4-11)
-2 treatments
Naive (inexperienced subjects)
Experienced subjects- same experiment one week later with same subjects
Results of Guth Et Al 1982
-naive subjects
Modal offer 50%
Mean offer 37%
-experienced subjects
Mean offer 32%
-systematic deviation from game theoretic prediction
“Game theory is of little help in exploring ultimatum game behaviour”
Describe the set up of Binmore Et Al 1985
Stage 1
The cake is of size 100 pence. P1 makes a proposal, P2 accepts or rejects (game continues)
Stage 2
The cake is of size 25 pence. P2 makes proposal, P1 accepts or rejects and both receive 0
Game A: as described
Game B: P2 gets to return and play as P1
Results of Binmore Et Al 1985
-a tendency to play fair in Game A becomes a strong tendency to play like a game theorist in game B
-generally poor documentation of experimental procedures and results including leading instructions
Key norms in experimental econ
-transparency and clear documentation of experimental procedures
-original instructions should always be made available
-original date should be made available
-important for interpretation of the data and replication
What did Forsythe Et Al 1994 find out about fairness concerns in UG behaviour?
People offer significantly less in dictator games so UG behaviour can’t be fully attributed to fairness concerns
McCabe & Smith 1986 set up and results
Several treatments of dictator game with varying degree of social isolation.
Single blind: anonymity among subjects
Double blind: anonymity wrt other subjects and experimenter
Results: with double (single) blind procedures approx 65% (40%) give nothing. Exp results can be sensitive to procedural details
What is replicability key for?
Trust in scientific knowledge
What is RPP and what did they do?
Reproducibility project psychology replicated 100 original experimental studies in top 3 journals in psychology. Found a significant effect in only 36% of them
What happens when Camerer (2016) tried replicating papers?
In 2018 he replicated 18 papers 2011-2014 and found 61% of papers were replicated well
Why does Camerer say economics paper are more replicable than psychology papers?
Experimental economists have strong norms about incentives and not using deception. These norms make subjects more responsive and may reduce variability in how experiments are done across different research teams. Editorial practices of printing instructions and data led to norms of transparency and made replication and reanalysis relatively easy
Neutral vs loaded framing
Neutral framing (standard practice)
-advantage: focuses on strategic situation, comparability with other studies
Loaded framing
-can be easier to understand
-can be morally loaded
-might induce associations, attitudes, memories from real life
-can produce framing effects
Possible interpretations of Guth Et Al 1982 results
-subjects fail to apply solution concept (backward induction)
-other regarding fairness predictions
How did Binmore Et Al 1985 interpret their results?
Subjects faced with a new problem simply chosen equal division as an obvious and acceptable compromise such considerations are easily displaced by considerations of strategic advantage once players fully appreciate the structure of the game
What is HARKing?
Hypothesising after results are known. Presenting a post hoc hypothesis as if it was an a priori hypothesis
