Causal Treatment Models
Estimate Causal Effects Using Observational Data
The Causal Treatment personality enables you to estimate causal effects using observational data. Causal questions are most directly answered by analyzing experimental data that are collected from randomized controlled trials. However, researchers often cannot randomly assign observations to receive a certain intervention because of ethical concerns or logistic constraints.
The Causal Treatment personality provides an estimate of the average treatment effect by leveraging insights from models that describe the treatment (X) and outcome (Y) mechanisms to mitigate confounding bias. These models enable you to answer the following questions:
• Does X cause Y?
• How would changing X affect Y?
• What would be the expected value of Y if no observations had received the treatment (X = 0), or if every observation had (X = 1)?
The Causal Treatment personality can be used to analyze the effects of binary or continuous treatments on binary or continuous outcomes. The personality also provides visualizations to help assess causal assumptions.
Figure 14.1 Propensity Score Density Plot for Evaluating the Positivity Assumption
