Katkat
Back

Explain confounding with an Uber example

PayPalData Scientist · Onsite
EasyData ScientistOnsiteStatistics & Math

Last updated

This Statistics & Math interview question for a Data Scientist tests understanding of confounding and causal inference in observational data. Candidates must define a confounder, give a concrete non-demographic Uber example identifying the exposure, outcome, and confounder with the direction of bias, and describe at least two mitigation methods along with the assumptions each requires.

Question

You are interviewing for a Data Scientist role and are given access to Uber / Uber Eats data. Answer the following about confounding in causal inference:

  1. Define confounding in the context of estimating causal effects from observational data. Explain what a confounder is and why it can bias an observed relationship between an exposure and an outcome.
  2. Give a concrete Uber-related example (avoid generic demographic examples like age/sex). Your example should clearly identify:
  • the treatment / exposure (X),
  • the outcome (Y), and
  • the confounder (Z) that affects both X and Y. Explain intuitively the direction of the bias (how it could manufacture a false effect or hide a real one).
  1. Describe at least two practical ways you would detect and/or mitigate confounding in an analysis (in the design or the modeling), and state what assumptions each method requires.