laboratory experiments. benefits of the lab as treated = intent to treat (usually) confounding...
TRANSCRIPT
Laboratory Experiments
Benefits of the Lab
• As treated = intent to treat (usually)• Confounding variables hopefully controlled• DVs and IVs can be measured with greater
precision than in the field• Experimenter can create different and unique
“worlds” to examine that may not appear in nature– Advantage for testing counterfactuals– Some theories imply “dogs that don’t bark”
Today
• Classic psychological experiment in obedience• Psychological experiment on stereotyping and
minority groups• Game theory experiment using animals• Voting experiments
Milgram Experiments
Stanley Milgram. 1963. Journal of Abnormal and Social Psychology
Protocol
• 40 male subjects recruited from New Haven, CT• Told they would participate in an experiment to
determine effect of punishment on learning• Brought into waiting room, sat next to a
“confederate” of the experimenter.• Drew slips of paper from a hat labeled “Teacher”
or “Learner”– But both slips marked “Teacher” so subject was
always the Teacher
• Subjects always selected to be teacher, other person in room (confederate), always student
• Subject taken to a room with a microphone, panel of switches labeled “Slight Shock” to “Danger: Severe Shock”
• Experimenter in room with subject to give directions
• Experimenter tells subject to read list of word pairs to “Student”
• Student asked to recall which word paired with another
• If Student answers incorrectly, Teacher told by experimenter to administer increasingly severe electric shock
• Teacher hears reaction to (fake) shock through speaker
If Teacher Refuses to Administer Shock
• Prod 1: Please continue. or Please go on. • Prod 2: The experiment requires that you
continue. • Prod 3: It is absolutely essential that you
continue. • Prod 4: You have no other choice, you must go
on.
Measures
• DV is maximum shock subject administers
Maximum Shock Administered
• 5 subjects stopped at “Intense Shock”• 9 stopped at “Extremely Intense Shock” to
“Danger Severe Shock”• 26 did not stop until maximum shock• Videos of subjects show intense concern
about well-being of “Student” and doubt about the experiment
• But most subject continued
Importance
• Showed how far human obedience to authority can extend
• Also showed dangers of social science experiments and importance of human subjects protection
Categorical and Contextual Bases ofPerson Memory and Stereotyping
Shelley E. Taylor, Susan T. Fiske, Nancy L. Etcoff, and Audrey J. Ruderman
Journal of Personality and Social Psychology1978, Vol. 36, No. 7
Hypotheses
• People encode information by the race and sex of the person they observe
• Minimize within-group differences and exaggerate between-group differences
• Within-group attributes exaggerated in inverse proportion to size of the minority group
Method• 21 white Harvard students recruited to listen to an
audiotape of a group discussion• Directions: “You are going to hear a tape of six men
discussing a publicity campaign for a play. To make it easier for you to tell the six voices apart, we will also show pictures of the men. As each person speaks, his picture will be shown on the opposite wall. The conversation lasts about 15 minutes. After the conversation is over, you will be asked some questions about the discussion. We are interested in several different kinds of observations about the group. Instead of having each of you answer all the same questions, everyone will be given a separate task.”
Randomization• Don't attend to anything in particular. While you watch,
just observe the interaction of the group as a whole and the contributions of each of the participants. (Control condition.)
• During the tape, each of the six participants makes a number of specific suggestions about how to run the publicity campaign. When the presentation is over, you will be given a list of the suggestions and asked to match each suggestion with a picture of the speaker. (Memory condition.)
• The purpose of this memory versus control manipulation was to see when categorization by race occurs
Protocol
• Subjects observed 15 minute audiotape discussion with photos of 6 men, 3 black and 3 white
• Each participant made 6 suggestions during discussion
• Photos shown of man talking randomized by race
Protocol• After hearing the carefully-scripted and balanced
discussion, subjects were told:
“You will now receive a separate page with pictures of each of the six participants in the discussion you just observed. Below each person's picture is a number. You will also receive a list of all suggestions made on the tape. In front of each suggestion there is a blank. Insert the number of the picture of the person who made that suggestion. In order for us to analyze the data, all questions must be answered. Guess if you do not know an answer.”
Dependent Variables
• Total number of errors• Number of black suggestions incorrectly
attributed other blacks (BB)• Number of white suggestions incorrectly
attributed to other whites (WW)• Total number of intraracial errors (BB + WW)• Number of white suggestions incorrectly
attributed to blacks (WB)• Total interracial errors (BW + WB)
Results
Results
• People make more errors remembering statements made by people of same race
• People encode information by the race of the person they observe
Another Experiment
• Six persons, three males and three females, were recruited to improvise a public school teacher's lounge conversation.
• Six new persons, again three men and three women, were next recruited to imitate each of the original speakers; each original male speaker was imitated by a female who mimicked him until she was able to get the same intonations and speech speed. The same was done with male imitators of female original speakers.
Treatment
• Created taped conversations varying in sex composition by splicing in imitators and splicing out the appropriate original voices in whatever combination was required.
Composition of Experiment Groups
Subjects
• 140 Harvard students• Randomly assigned to hear and see slides of
one group conversation• Subjects then rated group members on
several different traits
Results
• Perception of speaker varies by sex composition of group
Experiments on Games, Markets, Voting, and Committee Decisions
Boxed Pigs Game
• Two pigs are placed in a cage.• A lever must be pressed so that food comes out
of a dispenser. There are no other source of food.• Efforts must be spent to press the lever.• Each pig has two strategies: either press the lever
or wait at the dispenser.• Each pig’s utility is the amount of food consumed
less the efforts spent to press the lever
Figure 3.14 The Boxed-Pigs GameHarrington: Games, Strategies, and Decision Making, First EditionCopyright © 2009 by Worth Publishers
The boxed pigs game
Experimental results in the boxed pigs game:
Many Experiments Test Game Theory
• Students recruited to participate• Paid small amounts of money for outcomes of
games• Researcher structures payoffs to match the
underlying game theory model• Can vary payoffs for several rounds for each
participant (within-subject) or across participants (between-subject)
Should A Person Vote?
Standard Calculus of Voting ModelVote if PB > CP=probability of decisive voteB=benefit from preferred candidate winningC=cost of voting
Since C>0 and PB is very small, implies people should not vote
Bayesian Models
• But, if no one should vote, then someone should vote since he will be decisive
• Models incorporate beliefs about how many other voters will vote
• Predictions– Positive turnout– Turnout increases with closeness of race– More of “underdog’s” voters vote
Thomas R. Palfrey, “Laboratory Experiments in Political Economy”
Jury Games
• All jurors have common preferences: – convict a guilty defendant, acquit the innocent
• Each juror has private “signal” about whether defendant is guilty
• Jurors vote simultaneously or secretly• Each juror believes each other juror’s
“informedness” (probability of receiving the correct signal about defendant’s guilt) is greater than .5
Equilibrium in Jury Game
• Solve for case in which a juror is pivotal: If he is pivotal, most other jurors must believe defendant is guilty
• One Equilibrium: No juror votes innocent when in fact every juror believes the defendant is innocent
Problems?
• The lab is an artificial environment• Man is a social animal• Context matters