← Meta Interview Insights

Meta·Data Scientist·Technical Phone Screen·Senior

SeniorPrefer not to say
Jul 2026Remote

Summary

Meta DS interview with a heavy stats focus. The whole session was basically one long A/B testing problem that kept branching into harder territory. Not a vibe check at all, they wanted to see if you actually know the math.

Questions Asked (3)

Q1

Given an A/B test where both control and treatment have 10 million impressions and CTRs of 1.20% vs 1.23%, calculate the absolute and relative lift, pooled standard error, z-statistic, two-sided p-value, and a 95% confidence interval for the relative lift.

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

The arithmetic itself is fine but I fumbled the CI for relative lift.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clearly defining the metrics: absolute lift is the difference in CTRs, relative lift is the ratio of that difference to the control CTR. Then compute the pooled standard error using the pooled proportion, calculate the z-statistic, find the two-sided p-value, and construct a 95% confidence interval for the relative lift using the delta method or normal approximation. Finally, interpret the results in the context of statistical and practical significance.

Pro tip: Always state your assumptions (e.g., independence, large sample) and mention that with such large samples, even tiny lifts can be statistically significant but may not be practically meaningful. Also, consider using the delta method for the confidence interval of the relative lift to account for the ratio's non-linearity.

1. Calculate absolute and relative lift

Compute the absolute lift as the difference between treatment and control CTRs. Then compute the relative lift as the absolute lift divided by the control CTR, often expressed as a percentage.

2. Compute pooled standard error

Calculate the pooled CTR by combining successes from both groups. Then compute the pooled standard error using the formula sqrt(p_pool*(1-p_pool)*(1/n_control + 1/n_treatment)).

3. Calculate z-statistic and p-value

Compute the z-statistic as the absolute lift divided by the pooled standard error. Then find the two-sided p-value using the standard normal distribution.

4. Construct 95% confidence interval for relative lift

Use the delta method or normal approximation to compute the standard error of the relative lift, then construct the confidence interval as relative_lift ± 1.96 * SE_relative.

5. Interpret results

Discuss statistical significance (p-value < 0.05) and practical significance (effect size). Mention any caveats such as multiple testing or novelty effects.

Key Points to Mention

  • Absolute lift = treatment CTR - control CTR
  • Relative lift = (treatment CTR - control CTR) / control CTR
  • Pooled standard error formula for two proportions
  • Z-statistic = absolute lift / pooled standard error
  • Two-sided p-value from standard normal distribution
  • 95% confidence interval for relative lift using delta method or normal approximation
  • Interpretation of statistical vs practical significance

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

Q2

If you peeked at the experiment results daily over 14 days without any alpha spending correction, how much does the Type I error inflate, and how would you fix it using either a sequential testing plan with alpha spending or a Bayesian approach with a prior on lift?

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

Knew peeking was bad, knew it inflated Type I error, but quantifying it precisely under 14 peeks was where I got vague.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by explaining that peeking without correction inflates Type I error, then quantify the inflation using the multiple testing framework (e.g., 14 peeks at α=0.05 can inflate to ~20-30%). Finally, describe two solutions: sequential testing with alpha spending (e.g., O'Brien-Fleming) and Bayesian approach with a prior on lift, highlighting trade-offs.

Pro tip: Mention that the inflation depends on the correlation between interim looks and the number of peeks; in practice, daily peeking over 14 days can inflate Type I error to around 20-30%, but exact value requires simulation. Also, note that Bayesian methods with a skeptical prior can control false positives while providing interpretable posterior probabilities.

1. Quantify the inflation

Explain that each peek is a hypothesis test, so with 14 independent tests at α=0.05, the family-wise error rate is 1 - (0.95)^14 ≈ 51%. However, since tests are correlated, the actual inflation is lower, typically around 20-30% for daily peeking. Suggest simulation to get exact value.

2. Explain why correction is needed

Highlight that without correction, the false positive rate increases, leading to more false discoveries and undermining trust in experiments. This is especially critical in tech companies where many experiments run concurrently.

3. Sequential testing with alpha spending

Describe alpha spending functions (e.g., O'Brien-Fleming, Pocock) that allocate alpha across interim analyses. Explain that the overall Type I error is controlled at the desired level (e.g., 5%) by adjusting thresholds at each peek. Mention that this requires pre-specifying the number and timing of looks.

4. Bayesian approach with prior on lift

Explain that a Bayesian approach with an informative prior (e.g., skeptical prior centered at zero) can regularize the lift estimate and control false positives. The posterior probability of superiority can be monitored continuously without inflating error rates, as long as the prior is pre-registered.

5. Compare and recommend

Discuss trade-offs: sequential testing is frequentist and requires planning, while Bayesian is more flexible but sensitive to prior choice. Recommend based on context, e.g., for Meta's scale, sequential testing is common, but Bayesian is gaining traction for its interpretability.

Key Points to Mention

  • Type I error inflation due to multiple testing (peeking)
  • Family-wise error rate and Bonferroni correction (though not ideal for correlated tests)
  • Alpha spending functions (O'Brien-Fleming, Pocock) and sequential testing
  • Bayesian A/B testing with prior on lift (e.g., normal prior centered at zero)
  • Simulation to estimate actual inflation and power
  • Trade-offs between frequentist and Bayesian approaches in industry

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

Q3

If you're simultaneously tracking 8 guardrail metrics alongside your primary metric, how do you control for familywise error rate or false discovery rate?

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

Mentioned Bonferroni for FWER and Benjamini-Hochberg for FDR.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by acknowledging the multiple comparisons problem and its impact on false positives. Then, discuss methods to control familywise error rate (e.g., Bonferroni, Holm-Bonferroni) and false discovery rate (e.g., Benjamini-Hochberg), and explain how to choose based on the goal (strict control vs. discovery). Finally, emphasize practical considerations like pre-registration, hierarchical testing, and the trade-off between power and error control.

Pro tip: At Meta, guardrail metrics are often used to detect harm, so a conservative approach like Bonferroni or Holm is preferred to minimize false negatives, but you can also mention that if the goal is exploratory, FDR methods like Benjamini-Hochberg are more powerful. Also, highlight that guardrails are typically one-sided tests for harm, which affects adjustment.

1. Acknowledge the multiple comparisons problem

Explain that testing 8 guardrails plus the primary metric inflates the chance of false positives, and that without correction, the familywise error rate (FWER) can exceed the nominal alpha.

2. Choose an error control method

Decide whether to control FWER (e.g., Bonferroni, Holm-Bonferroni) or FDR (e.g., Benjamini-Hochberg). FWER is stricter and suitable when any false positive is costly; FDR is more powerful when some false discoveries are acceptable.

3. Apply the method correctly

For Bonferroni, divide alpha by the number of tests (e.g., 0.05/9 ≈ 0.0056). For Holm, order p-values and compare sequentially. For Benjamini-Hochberg, rank p-values and compare to (i/m)*alpha.

4. Consider practical adjustments

Mention that guardrails are often one-sided (testing for harm), so adjust alpha accordingly. Also, pre-register the testing plan to avoid p-hacking, and consider hierarchical testing where primary metric is tested first.

5. Discuss trade-offs and alternatives

Acknowledge that corrections reduce power, especially with many metrics. Suggest alternatives like combining guardrails into a composite or using Bayesian methods to mitigate the issue.

Key Points to Mention

  • Familywise error rate (FWER) vs. false discovery rate (FDR)
  • Bonferroni correction and its conservativeness
  • Holm-Bonferroni step-down procedure
  • Benjamini-Hochberg procedure for FDR control
  • One-sided tests for guardrail metrics (harm detection)
  • Pre-registration and hierarchical testing to control error rates

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