Lecture 2 Flashcards
Explanatory variables
variables observed in an experiment
Function of ordinary least squares (OLS)
chooses all betas s.t sum of squared errors between observed and estimate observations are minimized
unbiasedness
on average, the estimate beta exactly equals the true beta
representative sample
when on average, a regression line matches the population line well
condition for obtaining a causal effect
x must be uncorrelated with the error term
omitted variables bias
when treatment x is correlated with omitted variables (error term)
error term
contains all the variables that determine an outcome, y, but was excluded from the model
controls
independent variables included in a model to address biases
quasi-experimental conditions
different types of experiments meant to mimic experimental conditions in observational data
difference in diffirences experiment
type of experiment where a treatment group is compared to a control group that is not exposed to the treatment group
parallel trends assumption
assumption that treatment and control group would follow similar trajectories in the absence of intervention
fixed effects
attributes held constant in experiments with data over both time and space
panel fixed affects
quasi experimental design that holds attributes constant
Matching
quasi-experimental design in which a treatment group is matched to a control group that is similar based on observable factors
Regression discontinuity
quasi-experimental design that evaluates the treatment effect only at the cutoff score
