25.10.2023 Tools II#

Randomized Control Trials#

Gold Standard to find out about causation

  • 2 random groups

  • one treatment group, one with placebo

  • solvest the problem of bias

Bias: source of difference between treatment and control group that is correlated with the treatment but not due to the t.

Example: SAT Courses

  • Students taking SAT Prep Courses

  • have lower grades in the end

  • but due to lower starting point for course takers

Problems of RCT:

  • external validity: not applicable to other contexts

  • Attrition: reduction of sample size over time leads to bias

  • Expense

RCT in TANF context#

  • Experiment in Califarion in 1992

  • 1/3 families to existing program

  • 2/3 treatment group with 15% lower benefits

Results

  • Employment rates: 49% vs 44.5%

  • Elasticity of Labor: \(\frac{ -10\% }{15\%}=0.67\)

Observational Data#

Obersavtional Data: Data generated by individual behavior in real world

Problem: Bias

Time Series Analysis#

Analysis of the co-movement of two trends

  • can be used to analyze correltaion

  • to support theory

TANF Example

img

Problems:

  • separation from correlation to causation

  • excldued variables (often macro labor market)

Regression Analysis#

= best fitting linear-relationship between two variables

Cross-sectional regression analysis: Statistical analysis of the relationship between two or more variables exhibited by many individuals at one point in time.

find best fitting linear elationship between two variables

Regression line: The line that measures the best linear approximation to the relationship between any two variables

Example:

img

Problem of Biases can be solved with control variables

Control Variables: Variables that are included in cross-sectional regression models to account for differences between treatment and control groups that can lead to bias

Quasi Experiments#

Quasi Experiments: Changes in economic environment that create nearly identical treatment and control groups for studying the effect of that environmental change

Example: Card minimum wage Research uses DiD

Difference in Difference (DiD): The difference between the changes in outcomes for the treatment group and control group

Hypothetical Example: one state increases state aid, other does not

img

Assumption:

  • external effects (financial crisis) affect both states in the same way

  • both groups would develop the same (common trend assumption)

Tutorial#

internal validity: how valid for the population being studied

external validity: how valid for the general population

Threats to external validiry

  • non representative sample

  • non represenattive treatment

  • general equilibrium

    • scale and duration might change economic environment

    • ex: training unemployed => higher expectations from employers

    • changes baseline

Regression#

= best fitting linear-relationship between two variables

  • just correlation!

  • selection bias, omitted-variable bias