PS3: Introduction to Empirical Analysis and Quantitative Reasoning

Ongoing flashcard set for PS3 as taught by Prof. David Broockman at UC Berkeley (updated for MIDTERM)

Reverse causation

Thinking A → B but actually B ← A

A is related to B and someone argues that A causes B, but B might cause A

Omitted Variable Bias

Thinking A → B but actually C → A and B

a third factor, C, which causes A and B makes A related to be even though it doesn’t cause it

in observational data, those who get treatment usually differ in other ways

Selection Bias

Groups differ in other ways than the variable being measured

• Ex: self-selection

Potential outcomes

What the outcome would be for the same person/country/unit if the do/do not receive a treatment?

Average treatment effect

the difference on average if everyone gets the treatment

how much of a difference it makes on average

Experiments

• systematic collection of data, impose a treatment on one group with a control

What do experiments do?

• eliminate selection bias (random assignment)

• eliminate omitted variable bias (random assignment)

• randomly sample from potential outcomes (randomization takes random sample of treatment potential outcomes and control potential outcomes

Randomization

• corrects for selection bias and omitted variable bias

• ensures no third variable impacts correlation