Midterm

Question 1

Happenstance data

Answer 1

product of ongoing uncontrollable process (GDP, unemployment, inflation, gender wage gap)

Question 2

Pros of happenstance data (4)

Answer 2

• ecological validity
• cheaper to obtain
• does not involve interference
• large dataset

Question 3

cons of happenstance data

Answer 3

• internal validity is not clear (not sure causal relationship and interpretation; opp of controlled variable)
• external validity is not clear ((how can findings apply to other populations, environments, time ie. gender pay gap in one country isn’t the same in another)

Question 4

what is an experiment?

Answer 4

Method to collect data in controlled (most factors held constant, only one factor of interest is varying at a time) environment

Question 5

How to collect data in controlled environment?

Answer 5

Through randomization= indiv. Randomly assigned to treatment and control to eliminate prior systematic differences bw two groups

Question 6

Pros of experimental data? (2)

Answer 6

• Internal validity

- suitable for testing theory

Question 7

What is experimental data?

Answer 7

active intervention by the researcher to produce and measure change or to create difference when a variable is altered.

Question 8

Cons of experimental data?

Answer 8

Cons:
Need to think about external validity in other contexts, w other populations
More costly
**differences in lab and field experiments
*Most surveys not incentivized

Question 9

Ways to develop theories?

Answer 9

• Analytical models

- Computer simulations

Question 10

Purpose of experiments (6)

Answer 10

Documenting empirical regularities (reaction to framing effects)
Studying alternative institutions
Providing evidence to influence policy makers
Marketing applications
Measuring individual characteristics (risk preferences, time preferences)
Pedagogical purposes

Question 11

Scientific purposes of experiments (theory based)

Answer 11

• If no theory exists, discover empirical regularities where existing theory has little to say
• If many theories exist, test range of applicability of several competing theories
• If one theory exists, under which conditions can theory account for data

Question 12

Conjunction Fallacy

Answer 12

occurs when it is assumed that specific conditions are more probable than a single general one.
(Tversky and Kahneman)

Question 13

representativeness heuristics

Answer 13

estimating the likelihood of an event by comparing it to an existing prototype that already exist… It involves making judgments by comparing things to concepts we already have in mind.
*The problem with this is that people often overestimate the similarity between the two things they are comparing.

Question 14

Conjunction Fallacy + representativeness heuristics

Answer 14

likelihood of event is assessed as more likely bc of how it represents something
*If you can’t assess probabilities correctly, you can’t make optimal decisions

Question 15

Economic theory

Answer 15

*how indiv. evaluate outcomes, reason, and reach decisions
provides framework and tools for describing and analyzing economic situations
- involves description of social situation, specifying actors, choices they face, and how they evaluate each possible outcome

Question 16

behavioral assumptions (3)

Answer 16

1. specifies how indiv. evaluate each possible outcome
2. explains individuals cognitive abilities + how they form beliefs about uncertain states of the world
3. specifies how individuals behave

Question 17

standard economic theory assumptions (3)

Answer 17

1. econ decision makers are cognitively sophisticated= assumed to be rational
2. risk preferences obey expected utility principle; time pref. exhibit exponential discounting; social pref are assumed to be 0; ppl have cobb-douglas utility functions
3. solution concepts are typically equilibrium concepts, like competitive equil. or nash equil.

Question 18

economic experiments?

Answer 18

a controlled data generating process= “control” means factors which influence behavior are held constant and only one fact of interest is varied at a time
*natural exp= controlled data generating process happens naturally, BUT normally researcher who controls process

Question 19

What do theories do (re: experiments)

Answer 19

experiments used to test a theory…
Theory tells the researcher the relevant variables which need to be controlled or manipulated, and makes predictions of what will
happen in the experiment.

Question 20

What do experiments do (re: theory)

Answer 20

Experiments make contributions to scientific research in multiple fields

• enable us to address theories
• allow us to examine regularities from the field in controlled setting
• • test whether observed behavior connects to predictions of a particular model

Question 21

Reasons testing predictions of theories can be difficult

Answer 21

joint hypothesis tests

Question 22

joint hypothesis tests

Answer 22

whether assumption of theory holds in the field, and whether predictions of theory hold in the field
*problems w using data, that may not be completely applicable

Question 23

theories developed to explain observations in world (2)

Answer 23

• outcome-based theories: assume indiv. care about inequality outcomes
• intention-based theories: assume indiv care about intentions of others

Question 24

theories developed to explain observations in world (2)

Answer 24

• outcome-based theories: assume indiv. care about inequality outcomes
• intention-based theories: assume indiv care about intentions of others

Question 25

experiments testing predictions

Answer 25

- observational data will involve omitted variables which can be important to the theory (can involved testing comparative statics of a theory) - sometimes there are conditions where theories make different predictions that would not naturally occur (importance of distinguishing bw competing theories) (p. 126) - can limit domains where theories may or may not apply

Question 26

experiments providing behavioral models

Answer 26

models describe observed behavior in a particular framework

Question 27

How to evaluate theories?

Answer 27

1. offer unanticipated prediction, different from predictions of others 2. theories are falsifiable 3. theories need to make predictions in a large set of circumstances (scope of theory)

Question 28

purpose of linda experiment

Answer 28

- documenting empirical regularity - testing theory - measuring indiv characteristics - pedagogical purpose

Question 29

Dictator game experiment Q addressed?

Answer 29

Q addressed: Do individuals have social preferences, do they care about the payoffs that another person receives or own payoff?

Question 30

Dictator game results

Answer 30

- if player gains utility from own payoff and monoticity: dictator will give nothing to other player - giving more than 0, consistent w altruism or inequality aversion

Question 31

Dictator game purpose of experiments

Answer 31

- documenting empirical regularity - measuring indiv. characteristics - testing theory * external validity: how much indiv give in dictator game may be culturally dependent

Question 32

Ultimatum game

Answer 32

- test of theory (SPNE): backward induction standard assumption in spne: proposer allocated at most 1 to responder and responder accepts - tests altruism and fairness (responder could reject bc of unfair)

Question 33

ultimatum game purpose

Answer 33

- understanding social preferences - learning about simple bargaining - testing theory - pedagogical

Question 34

Principles of economics experiments

Answer 34

- degree of realism: replicate real world as closely as possible + replicate assumptions of a formal model of interest as closely as possible - creating an econ environment:individual economic agents + institution through which they interact - induced value theory - practical implications

Question 35

induced value theory

Answer 35

Key idea: induce pre-specified characteristics in experimental subjects Subjects’ innate characteristics become largely irrelevant

Question 36

Conditions to induce agents characteristics:

Answer 36

Monotonicity: prefer more to less, no satiation Salience of reward: reward depends on agents actions, those of other agents, and the institutional rules Dominance: changes in participants utility come predominantly from reward and other influences

Question 37

what to aim for in exp. econ

Answer 37

- make Qs of interest precise and find best design to answer - effective design is usually simpler than reality and than relevant formal models - find simplest possible design (ultimatum game as an abstraction of bargaining) - laboratory experiment should not try to replicate reality, but enhance understanding of problem

Question 38

implications of exp econ

Answer 38

- motivate subjects by paying cash - use more abstract not realistic phrasing - find subjects whose opp costs are low and learning curves are steep - create simplest economic env - promote salience, avoid loaded words (no dictator label) - maintain privacy of subjects - do not deceive subjects, unless necessary

Question 39

Holt and Laury (2002) main Qs

Answer 39

- do indiv. exhibit risk aversion and if so, to what extent? - does risk aversion depend on size of stakes? - do indiv. make diff choices when facing hypothetical and real payoffs?

Question 40

H&L 2002 type of experiment

Answer 40

- measuring indiv characteristics - documenting empirical regularities - testing theory - testing methodology

Question 41

H&L 2002 testing

Answer 41

- Hypothesis1: does size of stakes matter - hypothesis 2: real monetary incentives matter? - hypothesis 3: being exposed to a high real payoff affects choices in low real payoff treatment

Question 42

controlled variables opposite

Answer 42

happenstance data/observational data/naturally occurring data

Question 43

How to control variable

Answer 43

- hold it constant - Choose two more different levels that may produce sharply different outcomes, control variable at each chosen level; variables controlled at two or more levels are called treatment variables - Creates cheaper and simple experiments - Vary treatment variables independently to obtain clearest possible evidence on their effects

Question 44

Nuisance variablesn

Answer 44

*uncontrolled nuisance can cause inferential errors if confounded w focus variables--> EX/ H&L2002 wealth effects and solutions to these variables→ make nuisance variable independent of controlled variable

Question 45

How to avoid confounding problems

Answer 45

- Independence among controlled variables - Uncontrolled variables need to be independent of treatment variables - Randomization gives indirect control of uncontrolled variables, bc ensures independence of treatment variables - assign chosen levels of treatment variable in random order

Question 46

Types of experimental design

Answer 46

- Between-subject design: | - Within-subjects design

Question 47

Between-subject design:

Answer 47

vary focus variable only across subjects; when you have more ppl available for experiment

Question 48

Within-subjects design

Answer 48

use several different levels of the focus variable for each subject → potential issues: experimenter demand effects and order effects (have to randomize order)

Question 49

How to control nuisances

Answer 49

- Experience and learning: use experienced subjects - Noninstitutional interactions: monitor breaks - Fatigue and boredom - Selection biases - Subject or group idiosyncrasies: replication is necessary

Question 50

How to conduct experiment

Answer 50

1. Formulate a research q 2. Choose design that answers research q 3. Write down hypothesis 4. Choose treatment variables 5. How are subjects allocated to groups- randomly