Lecture 3 - Part I And II Flashcards
What is the null hypothesis?
A statement of no relationship among the variables being tested
What is the research / Alternative Hypothesis?
The predicted relationship between the variables being tested
What is research validity?
Truthfulness of a causal inference that is made from the results of a research study
What is internal validity?
Degree of confidence that the causal relationship being tested is trustworthy and not influenced by other factors or variables within the context of the present study
What is external validity?
The extent to which results from a study can be generalized to other
1. People
2. Settings
3. Treatments
4. Outcomes
5. Times
What are the 5 types of external validities?
- Population validity
- Ecological validity
- Treatment population validity
- Outcome validity
- Temporal validity
What are the 3 conditions for making the claim of causation?
- A must be correlated with B
- Changes in A must precede changes in B
- No plausible alternative explanation exists for the relationship between A and B
In the ABCD framework, what does each letter represent?
A: Independent Variable
B: Dependent Variable
C: Mechanism for how A affects B
D: Alternative Explanation
What is an observational study?
Researchers collect data by observing and making inferences based on that data. No interference from researchers
What is an experimental study?
A study where researchers control the assignment of treatments to subjects using a chance mechanism
What are the benefits of the experimental approach?
- Causal inferences can be drawn from the data collected
- Experimental approach grants researchers greater control over the other variables
What are the cons of the experimental approach?
- Does not test the effect of non-manipulated variables
- There may be ethical or practical problems with variable control
- The environment does not allow for generalization to real-life situations
- Cost
Why can’t lab findings be generalized to real life situations?
- Subjects know they are observed by researchers
- Subjects are placed in an unfamiliar situation and asked to make unfamiliar decisions
What is a laboratory experiment?
A controlled scientific investigation in a laboratory environment, often situated in a university computer lab or classroom often with college students participating.
Participants typically make decisions within an artificial context.
What is a field experiment?
Research using some controlled elements of traditional laboratory experiments, but takes place in natural/real world settings
What is an artefactual field experiment?
Same as conventional lab experiments but with a non-standard subject pool
What is a framed field experiment?
Similar to artefactual field experiments but with field context in either the commodity, task or information that the subjects use.
What is a natural field experiment?
Same as framed field experiment but subjects do not know they are participants in an experiment
What are the 2 properties of good measurement?
- Reliablity : consistency and stability of scores
- Validity : accuracy of inferences, interpretation, or actions made on the basis of test scores
What are 2 rules for economic-specific validity?
- Decisions should be properly incentivized
- earnings based on participant’s decisions
- paid no lower than average hourly rate - Use of deception is not allowed unless strictly necessary
What is the name for the experimental design involving 2 treatment variables?
2x2 factorial design
What is a 2 level and 3 level experimental design?
2 level: 1 control 1 treatment condition
3 level: 1 control 2 treatment conditions
What is a recommended minimum number of observations in each cell for a 2x2 design?
30
What are the 2 principles of manipulating a variable?
1) the manipulation indeed varies variable A
2) the manipulation only varies variable A
What are the 4 steps for manipulating variables?
- Determining the number of treatment/manipulated variables
- Determine the manipulation of the variables
- Manipulation check
- Control techniques
What are the 5 control techniques?
- Between subject design or within subject design
- Randomization
- Order effect
- Demand effect
- Wealth/income effect
What is the difference beween within-subject and between-subject?
Between: each individual exposed to only 1 treatment condition
Within: each individual exposed to more than one treatment condition
What are the benefits of within-subject experiment?
- Can control for many factors which may disturb your results
- Requires a smaller sample size than between subjects, leading to increased statistical power
What are the negatives of a within-subject design
- Carryover effect
- effect of prior experimental conditions - Demand effect
- participants try to interpret experimenter’s intentions and act accordingly
Advantage of between subject experience
Each participant enters the study fresh and naive with respect to the procedures to be tested
What is the effect of randomization on groups
It provides indirect control of uncontrolled variables by ensuring their eventual independence of treatment variables.
What is a common technique for creating equivalent groups?
Random assignment
Everyone has an equal chance of being placed in any of the groups, typically using a chance mechanism to determine which condition a subject is assigned to.
What is the order effect?
An influence on a particular trial that arises from its position in a sequence of trials.
In a session with multiple decision making tasks, where should the key task of interest be positioned?
At the outset
What are solutions to the demand effect?
- Use a between subject design
- Create incentives that make it costly to deviate from true preferred choices
What is the wealth effect?
In the incentive to exercise paper, one group exercised more and this reason could be due to experiment payout being higher than other groups, creating a wealth effect
Solution to wealth effect
All subjects receive an equal amount of payment regardless of treatment condition
How does wealth effect come into play in experiments with multiple rounds?
Participants earnings in prior rounds may influence their subsequent decisions
Solution to wealth effect in experiments with multiple rounds?
Subjects are paid according to results from a randomly selected period