Session 1 Flashcards
price elasticity of demand
the percentage change in quantity demanded resulting from a 1% increase in price
simultaneous causality
when causality runs both ways. for example, low taxes lead to high demand of cigarettes, but if there are lots of smokers in a state then politicians may try to keep taxes low
Phillips curve
low rate of unemployment is associated with an increase in the rate of inflation over the next year
causality
specific action leads to a specific, measurable consequence
causal effect
effect on an outcome of a given action or treatment as measured in an ideal, randomized, controlled experiment
internal validity
analysis has internal validity if the statistical inferences about causal effects are valid for the population being studied
Internal validity asks the question “did the experimental stimulus make some significant difference in this specific instance? (Is there a reasonable causal relationship?)”.
external validity
analysis has external validity if its inferences and conclusions can be generalized from the population and setting studied to other populations and settings
it relates to the question of whether the causal relationship holds for persons, settings, treatments and outcomes that were not in the experiment.
Example: finding similar results about the effect of student-teacher ratio on test performance in CA and MA data would be evidence of the external validity of the findings in CA. finding diff results would signal questions to both int. and ext. validity of at least 1 of the studies.
threats to internal validity
Issues may include selection bias, pre-study history that may impact results, maturation of individuals (internal trends), regression toward mean, attrition, the testing process itself and changing the measuring instrument over time.
(1) OLS estimators will be biased and inconsistent if the regressors and error terms are correlated
(2) Confidence intervals and hypothesis tests are not valid when the standard errors are incorrect
(1)
- omitted variables
- functional form (??)
- errors in variables (e.g taking average district income from 1990 while other data refers to 1998)
- selection
- simultaneous causality
(2)
- heteroskedasticity (???)
- correlation of the error term across observations
threats to external validity
differences in:
population (mice vs. men)
settings (public vs. private unis, San Diego vs. Anchorage)
Threats to external validity include: Validity only holds for some units Interaction with treatment variables Interaction with outcomes Interaction with settings
threats to internal validity
1-estimator of the causal effect should be unbiased and consistent (that is, OLS estimator of effect of test scores of a unit change in student-teacher ratio should be unbiased estimator of true popln causal effect or change in student-teacher ratio) (b-hat vs. b)
2-hypothesis test should have desired signif. level & CI should have desired confidence level
one ex: Omitted Variable Bias
Similar findings in 2 or more studies bolster claims to _____________ while differences in their findings that aren’t readily explained cast doubt on their __________.
Similar findings in 2 or more studies bolster claims to external validity while differences in their findings that aren’t readily explained cast doubt on their external validity.
causal vs. associate variables
Causal: Changing the value of one variable (X) causes a change in the value of another variable (Y)
Association: Two variables “move together” in the data but may not be causally linked
counterfactual
The counterfactual refers to the question of what would have happened if a situation did not exist.
Compare outcome Y when X occurs and when X does not occur.
For example, if someone did not have stress or poor diet, could they still get ulcers? If so, then the causal connection between stress/diet and ulcers is questionable.
While the counterfactual cannot be directly observed (can’t have the same individual in the test and control group at the same time), the goal of empirical analysis aimed at uncovering causal relationships is to “mimic” the counterfactual using data and statistical techniques.
o Ideal situation – randomized controlled experiment (to be discussed more)
Post hoc ergo propter hoc
Post hoc ergo propter hoc: Logical fallacy of “Since that event followed this one, that event must have been caused by this one.” For example, on mornings when more people arrive at work with umbrellas, it rains in the afternoon…doesn’t mean that the umbrella carrying caused the rain.
John Henry Effect
Hawthorne Effect
Rosenthal Effect
Control group subjects exert more effort than normal (“John Henry Effect”)
Control and Treatment group subjects behave differently (“Hawthorne Effect”)
Behavior influenced by the researcher itself (“Rosenthal effect” or “Pygmalion effect”). Also called an observer-expectancy effect
