What are heterogeneous treatment effects, and why can the average hide them?
PICTURE THIS: DATA SPLIT
Fit on train, tune on val, report on test once.
Simple meaning
Heterogeneous treatment effects are different lifts for different users, such as new versus power users.
WHY — A/B Testing instead of guessing?
Why interviewers care about A/B Testing:
people who only read docs from people who shipped.
and tied to Data Science work.
Name the idea, why it exists, then one short example.
End with when you use it and one common pitfall.
STEPS — What happens step by step?
Before you speak the answer, walk the interviewer through these steps:
- 1Heterogeneous treatment effects are
different lifts for different users, such as new versus power users.
- 2A near-zero average can
mix a large win and a large loss, so shipping on ATE alone can harm a key segment.
- 3Pre-specify a few segments
or use honest causal trees rather than slicing every dimension after the fact.
- 4Give an example
One tiny concrete case you can say aloud.
- 5Common mistake
What juniors usually get wrong.
- 6Close
When you pick this over the alternative.
EXAMPLE — See it in action
Here's a short line you can speak, broken into clear beats:
Note: Adapt this scaffold to your own project — keep it under 60–90 seconds.
Key takeaway
Heterogeneous treatment effects are different lifts for different users, such as new versus power users. A near-zero average can mix a large win and a large loss, so shipping on ATE alone can harm a key segment.