← Meta Interview Insights

Meta·Data Scientist·Technical Phone Screen·Senior

Senior
May 2026

Summary

Meta data scientist technical screen, heavy on stats and experimentation theory. The question was essentially one big multi-part problem about variance estimation and A/B testing design for a ratio metric, and it went pretty deep pretty fast.

Questions Asked (3)

Q1

Your team's KPI is ARPU, defined as Revenue divided by Active Users, computed per experiment bucket. Given that the correlation between Revenue and Active Users is 0.6, use the delta method to derive an approximate variance for ARPU, including the covariance term. How would you estimate the required inputs from sample data?

A/B Testing & ExperimentationProduct Analytics & Metrics
Author's notes

This is where I started sweating.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Define ARPU as the ratio of sample means R̄ and Ā, then apply the delta method to approximate its variance using the first-order Taylor expansion. Include the covariance term because Revenue and Active Users are correlated (ρ=0.6). Finally, explain how to estimate the required inputs (means, variances, covariance) from the experimental data.

Pro tip: Emphasize that the delta method variance is only valid for large samples; for small samples or heavy-tailed revenue, consider bootstrap or other robust methods. Also, clarify that the covariance term can be positive or negative, and its sign affects the variance estimate.

1. Define ARPU and its components

State that ARPU = R / A, where R is the sample mean revenue and A is the sample mean active users per bucket. Note that both are random variables with means μ_R and μ_A.

2. Apply the delta method

Use the first-order Taylor expansion around (μ_R, μ_A) to approximate Var(ARPU) ≈ (1/μ_A^2) Var(R) + (μ_R^2/μ_A^4) Var(A) - 2(μ_R/μ_A^3) Cov(R, A).

3. Incorporate correlation

Replace Cov(R, A) with ρ * sqrt(Var(R) * Var(A)), where ρ = 0.6, to explicitly include the correlation in the variance formula.

4. Estimate inputs from sample data

Compute sample means (R̄, Ā), sample variances (s_R^2, s_A^2), and sample covariance (s_{RA}) from the experiment bucket data. Use these as plug-in estimates for the delta method variance.

5. Compute and interpret the variance

Plug the estimates into the formula to get an approximate variance for ARPU. Discuss how the correlation term affects the variance and any assumptions (e.g., large sample, finite variance).

Key Points to Mention

  • Delta method: first-order Taylor expansion for ratio of means.
  • Variance formula includes terms for Var(R), Var(A), and Cov(R, A).
  • Cov(R, A) = ρ * σ_R * σ_A, with ρ = 0.6.
  • Sample estimates: R̄, Ā, s_R^2, s_A^2, s_{RA}.
  • Assumptions: large sample, finite variance, independent observations within bucket.
  • Alternative methods (e.g., bootstrap) if assumptions violated.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q2

Compare two analysis approaches for this experiment: testing ARPU directly versus modeling log-Revenue with log-Active Users as a covariate, something like an ANCOVA setup. What are the tradeoffs?

A/B Testing & ExperimentationTechnical Trade-offs
Author's notes

Honestly did not see this pivot coming.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the goal: ARPU is a ratio metric, so direct testing can be noisy and sensitive to outliers. Then compare the two approaches in terms of statistical power, interpretability, and assumptions, highlighting when each is preferable. Conclude with a recommendation based on the experiment's context and practical considerations.

Pro tip: Mention that log-transforming revenue and using log-active users as a covariate can improve normality and reduce variance, but it changes the interpretation to a multiplicative effect on ARPU. Also note that ANCOVA can adjust for baseline differences in active users, but it requires the covariate to be unaffected by treatment.

1. Define the metric and goal

Clarify that ARPU is revenue divided by active users, and the goal is to detect a treatment effect on ARPU. Discuss the implications of ratio metrics.

2. Direct ARPU testing

Explain that testing ARPU directly uses a t-test or similar on the ratio. It's simple and interpretable, but can have high variance and be sensitive to outliers in revenue or active users.

3. Log-revenue with log-active users as covariate (ANCOVA)

Describe modeling log(revenue) with log(active users) as a covariate. This can improve normality, reduce variance, and adjust for baseline differences. However, it assumes a linear relationship on the log scale and that the covariate is not affected by treatment.

4. Compare tradeoffs

Contrast the two approaches: direct ARPU is straightforward but may lack power; ANCOVA can be more powerful but requires stronger assumptions and careful interpretation (e.g., treatment effect on log-revenue given active users).

5. Recommendation and considerations

Suggest choosing based on data characteristics, business context, and whether active users is a post-treatment variable. If active users is affected by treatment, ANCOVA may be invalid; consider alternative methods like delta method or bootstrap.

Key Points to Mention

  • Ratio metrics like ARPU can have high variance and non-normal distributions.
  • Log transformation can stabilize variance and make the distribution more normal.
  • ANCOVA with log-active users as covariate can increase power by explaining some of the variance in revenue.
  • The covariate must be measured before treatment or be unaffected by treatment for ANCOVA to be valid.
  • Interpretation of treatment effect differs: direct ARPU gives absolute change; log-linear model gives multiplicative change.
  • Consider using the delta method or bootstrap for inference on ARPU if direct testing is problematic.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q3

When is a standard t-test acceptable for testing ARPU, and when would you switch to a bootstrap? Also, how do you handle heavy-tailed revenue distributions, for example through winsorization or robust estimators?

A/B Testing & ExperimentationTechnical Trade-offs
Author's notes

This part I felt better about.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by outlining the assumptions of the standard t-test (independence, normality, homoscedasticity) and explain when they hold for ARPU, such as with large samples due to the CLT or when the metric is not heavily skewed. Then discuss when to switch to a bootstrap (e.g., small samples, extreme skewness, or when the CLT is unreliable) and how to handle heavy tails via winsorization or robust estimators, emphasizing the trade-offs.

Pro tip: Mention that at Meta, ARPU is often heavy-tailed due to power users, so you'd typically use a bootstrap or a robust method like trimmed means, but always validate with a simulation or sensitivity analysis to ensure the chosen method controls Type I error and maintains power.

1. State assumptions of the t-test

Explain that the t-test assumes independent observations, normally distributed data (or large sample for CLT), and equal variances. For ARPU, these may be violated if the distribution is skewed or has outliers.

2. Determine when t-test is acceptable

The t-test is acceptable when sample sizes are large enough (e.g., >30 per group) for the CLT to apply, and the data is not extremely skewed. Also, if ARPU is roughly symmetric and variances are similar, it's fine.

3. Identify when to switch to bootstrap

Switch to bootstrap when sample sizes are small, the distribution is highly skewed or heavy-tailed, or when you need to estimate other statistics (e.g., median). Bootstrap makes fewer assumptions and can provide more reliable inference.

4. Handle heavy-tailed distributions

Discuss options: winsorization (capping extreme values at a percentile) to reduce influence of outliers, or using robust estimators like trimmed mean or median. Also consider transformations (e.g., log) but note interpretation challenges.

5. Evaluate trade-offs and validate

Acknowledge that winsorization introduces bias but reduces variance; robust estimators may be less efficient if data is normal. Recommend simulation or sensitivity analysis to choose the best method for the specific context.

Key Points to Mention

  • Central Limit Theorem and its role in t-test validity for large samples
  • Bootstrap resampling for non-normal or small-sample scenarios
  • Winsorization: definition, typical percentiles (e.g., 1st and 99th), and impact on bias-variance trade-off
  • Robust estimators: trimmed mean, median, and their properties
  • Power and Type I error considerations when choosing a method
  • Practical implementation: using simulation to compare methods and ensure results are reliable

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.