Causal Inference and Experimental Design

These flashcards cover vocabulary and key concepts related to causal inference, experimental design, and the methodologies used in social research.

Internal Validity

A study that provides convincing evidence of cause and effect, indicating that treatment and comparison groups are truly comparable.

External Validity

The extent to which the study's setting and participants reflect the broader population of interest.

Causal Inference

A process used to determine if an observed association indicates a cause-and-effect relationship.

Endogeneity

A threat to internal validity where the variable is influenced by the error term and other factors, causing omitted variable bias, simultaneous causality bias, or errors-in-variables bias.

Exogeneity

A condition where the independent variable is not correlated with the error term, allowing for valid causal inferences.

Randomized Controlled Trials (RCTs)

Experiments where subjects are randomly assigned to treatment and control conditions to test causal effects.

Intent-to-Treat (ITT) Effect

A method of analyzing participants in the groups they were originally assigned to, regardless of what happens post-randomization.

Treatment-on-the-Treated (TOT)

Measures the causal effect of actually receiving the treatment, rather than just being assigned to it.

Difference-in-Differences (DiD)

A statistical technique that compares the differences in outcomes before and after treatment between a treatment group and a control group.

Propensity Score Matching

A statistical technique used to reduce selection bias by creating comparable groups based on observed characteristics.

Regression Discontinuity Design (RDD)

A method for estimating causal effects by assigning treatment based on a cutoff in a running variable.

Fixed Effects (FE)

A method of controlling for unobserved factors that do not change over time, focusing on within-entity changes to estimate effects; comparing to itself

Instrumental Variables (IV)

Variables used to estimate causal relationships when the independent variable is correlated with the error term.

F-statistic

A value used to test the relevance of an instrument in instrumental variable analysis.

Quasi-Experimental Design

A research method that estimates causal impacts without random assignment, using pre-existing groups.

random sampling

the process of selecting participants from a larger population in a way that every individual has an equal chance of being chosen

random assignment

occurs after participants are selected, ensuring each has an equal chance of being assigned to any group in a study.

observational study

• No control over variables 

• Selection bias

• Independent variables are self-selected or otherwise endogenous

matching observational study

a type of observational study where researchers pair or group similar subjects based on specific characteristics to reduce bias.

case-control studies

involve comparing subjects with a specific condition (cases) to those without it (controls) to identify potential causes or risk factors.

Propensity score matching

A statistical technique used to reduce selection bias by equating groups based on covariates. It estimates the effect of a treatment by comparing outcomes between treated and control groups with similar propensity scores.

sharp RD

individuals receive treatment only if they are on the other side of the cutoff

Fuzzy RD

the probability of receiving treatment changes discontinuously at the cutoff but not from 0 to 1. (Treatment is correlated, but the cutoff doesn't ideally determine it.