← Meta Interview Insights

Meta·Data Scientist·Technical Phone Screen·Senior

Senior
Jul 2026

Summary

A Meta DS interview that went deep into experiment statistics, way deeper than I expected. The whole session was basically one long clustered experiment design problem broken into six parts, and I was sweating through most of it.

Questions Asked (6)

Q1

Given a baseline exploration rate, a target minimum detectable effect, and cluster-level randomization with a known intra-cluster correlation, derive the required number of users and clusters per arm. Show the formulas including the design effect adjustment.

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

The design effect formula itself isn't hard once you remember DE = 1 + (m-1)*ICC, but I fumbled the algebra when converting the inflated variance back to cluster counts.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by deriving the sample size formula for individual randomization using the baseline rate and minimum detectable effect, then adjust for cluster randomization by multiplying by the design effect (1 + (m-1)*ICC). Finally, compute the number of clusters per arm by dividing the total required users by the cluster size, ensuring it's rounded up.

Pro tip: Always clarify whether the ICC is for the outcome or the effect, and consider using the coefficient of variation to account for varying cluster sizes, as Meta often deals with unequal cluster sizes in practice.

1. Derive individual sample size

Use the standard formula for comparing two proportions: n = (Z_{α/2} + Z_{β})^2 * (p1(1-p1) + p2(1-p2)) / (p2 - p1)^2, where p1 is baseline rate and p2 = p1 + MDE.

2. Calculate design effect

Compute the design effect (DE) = 1 + (m - 1) * ICC, where m is the average cluster size and ICC is the intra-cluster correlation.

3. Adjust sample size for clustering

Multiply the individual sample size by the design effect to get the required number of users per arm: N_users = n * DE.

4. Compute clusters per arm

Divide the required users per arm by the cluster size m and round up to get the number of clusters per arm: k = ceil(N_users / m).

5. Validate and refine

Check assumptions (e.g., normality, known ICC) and consider adjustments for unequal cluster sizes or finite population correction if needed.

Key Points to Mention

  • Design effect formula: 1 + (m - 1) * ICC
  • Sample size formula for proportions: n = (Z_{α/2} + Z_{β})^2 * (p1(1-p1) + p2(1-p2)) / (p2 - p1)^2
  • Cluster size (m) and number of clusters per arm (k) relationship: N_users = k * m
  • Intra-cluster correlation (ICC) and its impact on variance inflation
  • Rounding up clusters to ensure sufficient power
  • Potential need for unequal cluster size adjustment using coefficient of variation

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

Q2

What exact hypothesis test would you use for the primary metric in a clustered experiment, and when would you prefer a nonparametric alternative?

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

I went with a cluster-robust z-test on cluster means, which felt right, but I got a bit tangled explaining when to swap to something nonparametric.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the clustering structure and the primary metric's distribution, then recommend a cluster-robust test like the cluster-robust t-test or a mixed-effects model. Discuss when nonparametric alternatives like the cluster bootstrap or permutation tests are preferable, emphasizing trade-offs between assumptions and power.

Pro tip: Mention that at Meta, cluster-robust standard errors are often the default for clustered experiments, but always check the number of clusters—if it's small, use a nonparametric method like the cluster bootstrap to avoid inflated Type I error.

1. Clarify clustering and metric

Identify the clustering unit (e.g., user, session) and the primary metric's type (continuous, binary, count). Confirm whether clusters are balanced and the number of clusters.

2. Choose parametric test

For large number of clusters, use a cluster-robust t-test (e.g., with CR2 standard errors) or a mixed-effects model if you need to model random effects. For binary metrics, consider a cluster-robust logistic regression or GEE.

3. Consider nonparametric alternatives

When the number of clusters is small (<50), the metric is skewed, or there are outliers, prefer nonparametric methods like the cluster bootstrap or permutation tests (randomization at cluster level).

4. Evaluate assumptions and power

Check assumptions of parametric tests (normality of cluster-level means, homoscedasticity). If violated, nonparametric methods may be more reliable, but they can have lower power.

5. Recommend and justify

State your preferred test based on the context, and explain the trade-offs (e.g., robustness vs. power, computational cost).

Key Points to Mention

  • Cluster-robust standard errors (e.g., CR2) to account for intra-cluster correlation.
  • Mixed-effects models or GEE for clustered data with binary/count outcomes.
  • Cluster bootstrap for nonparametric inference, especially with few clusters.
  • Permutation tests at the cluster level to maintain exchangeability.
  • Trade-offs: parametric tests assume normality of cluster-level statistics; nonparametric tests are robust but may have lower power.
  • Importance of the number of clusters: small number of clusters can lead to inflated Type I error with cluster-robust methods.

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

Q3

Define the confidence interval you would report for this experiment and explain how you'd interpret it alongside practical significance thresholds.

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

Pretty standard but the 'jointly with practical significance' part is where people slip up.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the experiment design, primary metric, and whether the metric is a mean or proportion, then specify the confidence level (typically 95%) and how you'd compute the interval (e.g., using the delta method or bootstrap for ratio metrics). Explain that you'd interpret the interval by checking whether it excludes zero and whether the entire interval lies above or below the pre-defined practical significance threshold (e.g., minimum detectable effect or business-relevant lift).

Pro tip: Meta often cares about practical significance over statistical significance, so emphasize that a statistically significant result with a confidence interval that barely crosses zero may not be worth shipping. Also, mention that you'd consider the confidence interval's width to assess precision and potentially run the experiment longer if the interval is too wide to make a decision.

1. Clarify experiment and metric

Confirm the experiment's primary metric (e.g., click-through rate, revenue per user), the unit of analysis (user-level), and the confidence level (usually 95%).

2. Choose the appropriate method

For simple metrics like proportions, use the standard normal approximation; for ratio metrics or complex designs, use the delta method or bootstrap to account for variance.

3. Compute and report the interval

Calculate the confidence interval for the difference between treatment and control (absolute or relative), and report it with the point estimate and confidence level.

4. Interpret against practical thresholds

Compare the interval to the pre-defined practical significance threshold (e.g., a 1% lift). If the entire interval is above the threshold, the effect is practically significant; if it includes zero, the result is inconclusive.

5. Make a decision recommendation

Based on the interval and thresholds, recommend whether to ship, iterate, or abandon the change, considering business impact and risk.

Key Points to Mention

  • Confidence level (e.g., 95%) and its interpretation as repeated sampling coverage
  • Difference between statistical significance (interval excludes zero) and practical significance (interval exceeds threshold)
  • Handling of ratio metrics or non-normal data using delta method or bootstrap
  • Importance of pre-registering the practical significance threshold to avoid post-hoc rationalization
  • Consideration of confidence interval width for decision-making and potential need for longer runtime
  • Meta's emphasis on actionable metrics and avoiding false positives in large-scale testing

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

Q4

You're peeking at the primary metric weekly for four weeks. Design a sequential testing plan and show the adjusted alpha at each look.

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

I knew O'Brien-Fleming style spending was the answer they wanted.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the goal: control the family-wise error rate (FWER) across four weekly looks while maintaining power. Then choose a sequential testing method (e.g., Pocock, O'Brien-Fleming, or alpha spending) and compute adjusted alpha at each look, explaining trade-offs. Finally, discuss practical considerations like peeking bias, stopping rules, and communication with stakeholders.

Pro tip: Mention that in practice, you'd pre-register the sequential plan and use a tool like gsDesign or sequential package to compute boundaries, and emphasize that the adjusted alpha depends on the chosen spending function and the correlation between test statistics.

1. Clarify objectives and constraints

Confirm the primary metric, hypothesis (two-sided vs one-sided), desired overall alpha (e.g., 0.05), power (e.g., 80%), and expected effect size. Discuss whether early stopping is for efficacy, futility, or both.

2. Choose a sequential testing method

Select a method such as Pocock (constant alpha at each look), O'Brien-Fleming (conservative early, liberal late), or a flexible alpha spending function (e.g., Lan-DeMets). Explain the trade-offs in terms of Type I error control and power.

3. Compute adjusted alpha at each look

For four equally spaced looks, calculate the adjusted alpha levels. For example, with Pocock and overall alpha=0.05, each look uses alpha≈0.0182; with O'Brien-Fleming, the first look might use alpha≈0.005, increasing to ≈0.048 at the final look. Show the calculations or reference software.

4. Address practical considerations

Discuss how to handle unplanned peeks, missing data, and multiple metrics. Mention that the sequential plan should be pre-registered and that any deviation requires correction.

5. Summarize and communicate

Conclude with a clear recommendation, emphasizing that the adjusted alpha preserves the overall Type I error rate and that the choice of method depends on the risk tolerance for early stopping.

Key Points to Mention

  • Family-wise error rate (FWER) control and the inflation of Type I error with multiple looks.
  • Alpha spending functions (e.g., Lan-DeMets) and how they allocate alpha over time.
  • Pocock vs. O'Brien-Fleming boundaries: constant vs. increasing alpha levels.
  • The need for pre-registration and avoiding ad-hoc peeking.
  • Impact on power and sample size: sequential designs often require a larger maximum sample size.
  • Software tools (e.g., gsDesign, R packages) for computing boundaries.

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

Q5

You're tracking three guardrail metrics alongside the primary. How do you control for multiple testing while preserving power on the primary, and what's your decision logic for combining primary and guardrail results?

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

Went hierarchical: primary first, guardrails only evaluated conditionally.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by distinguishing the primary metric from guardrails, then explain how to control family-wise error rate (FWER) or false discovery rate (FDR) across the guardrails without inflating the primary's alpha. Emphasize that guardrails are typically non-inferiority checks, so use one-sided tests with a pre-specified margin, and describe a decision framework that prioritizes the primary while treating guardrail violations as blockers or escalation triggers.

Pro tip: Frame guardrails as 'canaries'—they are not meant to be optimized, but to detect harm; using a non-inferiority margin and a gatekeeping procedure (e.g., fixed-sequence) preserves power on the primary and avoids unnecessary multiplicity adjustments.

1. Clarify roles and hypotheses

Define the primary metric as the sole confirmatory endpoint and guardrails as safety checks with non-inferiority hypotheses. State that guardrails are not for claiming improvement, only for detecting degradation.

2. Choose a multiplicity control strategy

Use a gatekeeping or hierarchical testing procedure: test the primary at full alpha first; only if significant, test guardrails with alpha control (e.g., Bonferroni or Holm) to preserve power on the primary. Alternatively, use a fixed-sequence procedure where guardrails are tested only after the primary wins.

3. Set non-inferiority margins and power

Pre-specify acceptable degradation margins for each guardrail based on business impact, and power the experiment for the primary while ensuring guardrails have sufficient power to detect meaningful harm. Consider that guardrails often need larger sample sizes for non-inferiority.

4. Define decision logic

If primary is significant and all guardrails pass non-inferiority, ship. If primary is significant but a guardrail fails, escalate for review or consider a follow-up experiment. If primary is not significant, do not test guardrails for superiority; only check for harm if pre-registered.

5. Communicate trade-offs and sensitivity

Discuss how the chosen procedure affects power and false positive rates, and provide sensitivity analyses (e.g., different margins, FDR vs FWER) to show robustness of conclusions.

Key Points to Mention

  • Family-wise error rate (FWER) vs false discovery rate (FDR) and when to use each
  • Gatekeeping procedures (e.g., fixed-sequence, hierarchical testing) to preserve alpha on primary
  • Non-inferiority testing for guardrails with pre-specified margins
  • Power considerations: guardrails may need larger sample sizes; primary power should not be compromised
  • Decision rules: primary success + guardrail pass = ship; primary success + guardrail fail = escalate; primary fail = no ship
  • Pre-registration of analysis plan to avoid p-hacking and ensure valid inference

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

Q6

If 10% of control users are accidentally exposed to the treatment, what direction does this bias the intent-to-treat estimate and how would you correct for it?

A/B Testing & ExperimentationRoot Cause Analysis
Author's notes

Contamination biases ITT toward the null, attenuating the treatment effect estimate.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, explain that contamination dilutes the treatment effect, biasing the ITT estimate toward the null. Then, describe how to correct for it using instrumental variables or by estimating the complier average causal effect (CACE), while noting the assumptions required.

Pro tip: Mention that while ITT is the gold standard for causal inference, it estimates the effect of assignment, not treatment receipt. For a more precise effect, consider CACE, but be transparent about its assumptions and limitations.

1. Identify the bias direction

Explain that contamination in the control group dilutes the difference between treatment and control, biasing the ITT estimate toward zero (null effect).

2. Quantify the contamination

Determine the proportion of control users exposed to treatment (10%) and assess how it reduces the observed effect size.

3. Choose a correction method

Use instrumental variables (IV) with random assignment as the instrument to estimate the Complier Average Causal Effect (CACE), or apply a simple adjustment if the exposure rate is known.

4. Validate assumptions

Check that the instrument is valid (random assignment, exclusion restriction) and that the exposure is accurately measured. Discuss potential violations.

5. Interpret results

Compare the corrected estimate (CACE) with the ITT estimate, and discuss the implications for decision-making, noting that ITT remains relevant for policy.

Key Points to Mention

  • ITT estimates the effect of treatment assignment, not treatment receipt.
  • Contamination dilutes the effect, biasing toward the null.
  • Instrumental variables (IV) can correct for non-compliance.
  • CACE (Complier Average Causal Effect) estimates the effect for those who comply.
  • Assumptions: random assignment, exclusion restriction, monotonicity.
  • ITT is still valuable for policy decisions as it reflects real-world effectiveness.

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