Section 19 of 24 · 4 lessons

Causal Inference and Experiments

Knowing which plants grew well does not tell us which intervention helped them. We will examine how treatment groups were formed, design a randomised comparison, and state the assumptions needed to learn from observational records. The calculations will keep those assumptions visible.

0 of 4 finished

Asking What a Change Would Do

Predicting an outcome and changing it are different questions. Examine why groups differ before using a random assignment to make a treatment comparison.

  1. 01Separating a Useful Prediction From a Causal ClaimPlants receiving extra water may grow poorly because they were already struggling. Compare the question a prediction answers with the question an intervention asks.
    Build with Botie →Read the lesson
  2. 02Making a Fair Comparison by AssignmentRandom assignment separates a treatment decision from the reasons a gardener might otherwise choose it. Estimate an average effect while keeping the experiment’s unit and uncertainty clear.
    Build with Botie →Read the lesson

Working With Observational Evidence

An old dataset cannot be rerun as an experiment. Make the assumptions behind adjustment and weighting explicit, including where the data provide no comparison.

  1. 03Comparing Treatments in the Same PopulationWhen starting conditions affect treatment choice, standardisation can compare both treatments using one target population, if the measured conditions suffice to control confounding.
    Build with Botie →Read the lesson
  2. 04Reweighting Records Without Inventing Missing EvidenceInverse-probability weighting changes which records count most. Understand the treatment probabilities and overlap assumptions before interpreting the resulting contrast.
    Build with Botie →Read the lesson