Skip to main content

Effects analysis

Since version 10

Introduction

The Effects Analysis tool calculates the Causal Effect on one outcome Y, based on the values of one or more treatments X.

This is equivalent to setting an intervention on each treatment value, but this tool provides a way to automatically record the outcome Y for each treatment intervention, and also allows for comparison of different treatments in a chart.

For example, in the image below, we are calculating the causal effect of Education on Salary and also the causal effect of Experience on Salary.

info

Values on the X-axis are normalized, so when multiple treatments are present, they can be compared.

Effects analysis

Treatments

Discrete Treatment

When a treatment X is discrete, the causal effect on the outcome Y, is calculated for each state in X, given the current evidence.

Continuous Treatment

When a treatment X is continuous, the treatment X is discretized given the current evidence, then the causal effect on the outcome Y is calculated for each discretized interval of X.

Outcome

Discrete Outcome

When the outcome Y is discrete, and the outcome state y is specified, P(Y=y | Do (X=x)) is calculated for each treatment value.

Discretized Outcome

When the outcome Y is discretized, and the outcome state is not specified, , the mean and the variance of Y is calculated for each treatment value.

info

If the outcome state is specified, the probability of the outcome is calculated instead.

Continuous Outcome

When the outcome Y is continuous, the mean and the variance of Y is calculated for each treatment value.