← Disney Interview Insights

Disney·Data Scientist·Technical Phone Screen·Senior

Senior
May 2026

Summary

Disney data scientist interview that went deep into experiment design, like really deep. Six questions all centered on one A/B test scenario for a paywall copy change, covering sample sizing, variance reduction, clustering, sequential testing, multiple comparisons, and noncompliance. Not a vibe check round at all.

Questions Asked (6)

Q1

Starting from scratch: compute the per-arm sample size for a two-sided test with 80% power and alpha 0.05, given an 18% baseline conversion rate and a 7% relative minimum detectable effect. Walk through the formulas.

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

This is the foundation question and I spent too long second-guessing whether the 7% was relative or absolute.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by converting the relative MDE to an absolute effect size, then apply the standard two-proportion sample size formula for a two-sided test. Clearly state each formula and plug in the values, showing intermediate calculations to demonstrate transparency.

Pro tip: Always sanity-check your result against a known rule of thumb (e.g., ~16 * p(1-p)/delta^2 for 80% power and 5% alpha) and mention that real-world factors like traffic allocation or multiple testing may require adjustments.

1. Define parameters and convert relative MDE to absolute

Identify baseline conversion rate (p1 = 0.18), relative MDE (7%), and compute absolute MDE (delta = 0.18 * 0.07 = 0.0126). Set p2 = p1 + delta = 0.1926.

2. State the sample size formula for two proportions

Use n = (Z_{α/2} + Z_β)^2 * (p1(1-p1) + p2(1-p2)) / (p2 - p1)^2 for equal allocation. For α=0.05 two-sided, Z_{α/2}=1.96; for 80% power, Z_β=0.84.

3. Plug in values and compute

Calculate p1(1-p1)=0.18*0.82=0.1476, p2(1-p2)=0.1926*0.8074≈0.1555, sum=0.3031. (1.96+0.84)^2=7.84. Delta^2=0.0126^2=0.00015876. n = 7.84 * 0.3031 / 0.00015876 ≈ 14968 per arm.

4. Round up and interpret

Round up to the nearest whole number (e.g., 14,968 per arm). Explain that this is the required sample size per variant to detect the specified effect with 80% power and 5% significance.

Key Points to Mention

  • Two-sided test vs one-sided: use Z_{α/2} for two-sided.
  • Power (1-β) and its corresponding Z-value (0.84 for 80% power).
  • Formula for sample size in two-proportion test: n = (Z_{α/2} + Z_β)^2 * (p1(1-p1) + p2(1-p2)) / (p2 - p1)^2.
  • Conversion of relative MDE to absolute effect size.
  • Assumption of equal allocation and independence.
  • Consideration of practical adjustments (e.g., multiple comparisons, non-compliance).

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

Q2

If you apply CUPED using a pre-experiment covariate with R-squared of 0.25, how much does that reduce variance and what does the new required sample size look like?

A/B Testing & Experimentation
Author's notes

CUPED variance reduction is just 1 minus R-squared, so 25% variance reduction here.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by explaining the CUPED variance reduction formula: variance reduction equals R-squared of the pre-experiment covariate with the outcome. Then apply it to the given R-squared of 0.25 to conclude a 25% reduction in variance, which translates to a 25% reduction in required sample size (since sample size is proportional to variance). Finally, discuss practical implications and assumptions.

Pro tip: Mention that the R-squared must be estimated from historical data and that CUPED is most effective when the covariate is highly correlated with the outcome; also note that the reduction in sample size is approximate and assumes the covariate is properly used.

1. Recall CUPED variance reduction formula

State that CUPED reduces variance by a factor of (1 - R^2), where R^2 is the squared correlation between the pre-experiment covariate and the outcome metric.

2. Apply the given R-squared

Plug in R^2 = 0.25 to get variance reduction = 1 - 0.25 = 0.75, meaning variance is reduced by 25%.

3. Translate to sample size reduction

Since required sample size is proportional to variance, a 25% reduction in variance leads to a 25% reduction in required sample size (e.g., from N to 0.75N).

4. Discuss assumptions and practical considerations

Note that this assumes the covariate is pre-experiment, properly implemented, and that the R-squared estimate is accurate; also mention that CUPED can be combined with other variance reduction techniques.

Key Points to Mention

  • CUPED variance reduction formula: Var_reduction = 1 - R^2
  • R-squared of 0.25 implies 25% variance reduction
  • Sample size reduction is proportional to variance reduction, so 25% fewer samples needed
  • Assumption: covariate is pre-experiment and not affected by treatment
  • R-squared must be estimated from historical data or pilot study
  • CUPED is most effective when covariate is highly correlated with outcome

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

Q3

Adjust the sample size for household-level randomization using the design effect formula DE = 1 + (m-1) times ICC, where average household size is 1.8 and ICC is 0.06. Give the adjusted per-arm sizes with and without CUPED.

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

Design effect comes out to 1 + (0.8)(0.06) = 1.048.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, compute the design effect (DE) using the given formula with m=1.8 and ICC=0.06, yielding DE = 1 + (1.8-1)*0.06 = 1.048. Then, multiply the standard per-arm sample size (without clustering) by DE to get the adjusted per-arm size for household-level randomization. For CUPED, apply the variance reduction factor (1 - ρ²) to the adjusted sample size, where ρ is the correlation between the pre-experiment covariate and the outcome; if ρ is not provided, state the formula and note that the CUPED-adjusted size is the household-adjusted size times (1 - ρ²).

Pro tip: Always clarify that the design effect inflates the sample size to account for intra-cluster correlation, and that CUPED further reduces it by leveraging pre-experiment data; if the baseline sample size isn't given, express the answer as a multiplier (e.g., 'multiply the standard per-arm size by 1.048 for household clustering, and by 1.048*(1-ρ²) with CUPED').

1. Identify given parameters and baseline

Note the average household size (m=1.8), ICC=0.06, and that the baseline per-arm sample size (n) is not provided. State that you will express the adjusted sizes as multipliers of n.

2. Compute the design effect

Apply the formula DE = 1 + (m-1)*ICC = 1 + (1.8-1)*0.06 = 1.048. Explain that this factor accounts for the correlation among individuals within the same household.

3. Calculate adjusted per-arm size without CUPED

Multiply the baseline per-arm size by DE: n_adj = n * 1.048. If n is known, provide the numeric value; otherwise, state the multiplier.

4. Incorporate CUPED adjustment

Apply the CUPED variance reduction factor (1 - ρ²), where ρ is the correlation between the pre-experiment covariate and the outcome. The CUPED-adjusted per-arm size is n_adj * (1 - ρ²). If ρ is not given, present the formula and note that typically ρ ranges from 0.3 to 0.7, leading to further sample size reduction.

5. Summarize and interpret

Conclude that household-level randomization increases the required sample size by 4.8%, while CUPED can reduce it by a factor of (1 - ρ²). Emphasize that both adjustments are essential for accurate power analysis in clustered randomized experiments.

Key Points to Mention

  • Design effect formula DE = 1 + (m-1)*ICC and its role in adjusting for intra-cluster correlation.
  • Calculation: DE = 1.048 for m=1.8 and ICC=0.06.
  • Adjusted per-arm sample size without CUPED = baseline n * 1.048.
  • CUPED variance reduction factor (1 - ρ²) and its application to further adjust sample size.
  • Need for baseline sample size to provide numeric answer; otherwise express as multiplier.
  • Practical implication: household randomization requires larger sample, but CUPED can mitigate the increase.

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

Q4

How do O'Brien-Fleming spending boundaries change how Type I error is allocated across interim looks, and what does that mean practically for your experiment timeline and power?

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

The key point is that O'Brien-Fleming is conservative early and spends very little alpha at the first looks, which protects you from stopping too soon on noise.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining O'Brien-Fleming boundaries as a group sequential design that spends very little alpha early and more later, preserving overall Type I error. Then explain the practical implications: you need larger effect sizes to stop early, but you gain flexibility to monitor without inflating false positives, which affects timeline and power calculations.

Pro tip: Mention that O'Brien-Fleming is conservative early, so you rarely stop for efficacy at the first interim, but it's excellent for safety monitoring. Also, highlight that you must pre-specify the number and timing of interim looks to control alpha properly.

1. Define O'Brien-Fleming boundaries

Explain that O'Brien-Fleming is a group sequential method where the alpha spent at each interim look is very small early and increases toward the final analysis, following a Lan-DeMets spending function.

2. Explain Type I error allocation

Describe how the overall Type I error rate (e.g., 5%) is partitioned across interim looks, with minimal alpha spent at early looks and most reserved for the final look, thus controlling the family-wise error rate.

3. Discuss impact on stopping boundaries

Note that early stopping requires a very large treatment effect (stringent boundary), making it hard to stop early for efficacy, but easy to stop for harm. Later looks have less stringent boundaries.

4. Relate to timeline and power

Explain that because early stopping is unlikely, the experiment often runs to the planned end, so timeline is more predictable. Power is slightly reduced compared to a fixed-sample design if you don't stop early, but the loss is minimal with few interim looks.

5. Provide practical recommendations

Suggest using O'Brien-Fleming when early stopping for efficacy is not critical, but safety or futility monitoring is important. Pre-specify interim looks and adjust sample size to maintain desired power.

Key Points to Mention

  • Group sequential design and alpha spending functions
  • Lan-DeMets spending function approximation
  • Conservative early stopping boundaries for efficacy
  • Impact on maximum sample size and power
  • Pre-specification of interim analysis timing
  • Trade-off between early stopping and resource savings

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

Q5

With 12 secondary metrics, how would you control false discovery rate at 10% and how do you communicate what a discovery actually means in that context?

A/B Testing & ExperimentationProduct Analytics & MetricsStakeholder Management
Author's notes

Benjamini-Hochberg: rank p-values, compare each to (rank/total) times FDR threshold, reject up to the largest one that passes.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by explaining that with 12 secondary metrics, you need to control the false discovery rate (FDR) to avoid false positives. Describe using a method like Benjamini-Hochberg to control FDR at 10%, and then discuss how to communicate that a 'discovery' means a metric is statistically significant after FDR control, but still requires practical significance and business context.

Pro tip: Emphasize that FDR control is about managing the expected proportion of false discoveries among all discoveries, not the probability of any false positive. This shows a nuanced understanding that impresses interviewers.

1. Define the multiple testing problem

Acknowledge that testing 12 metrics inflates the chance of false positives. Explain that without correction, the family-wise error rate would be too high.

2. Choose FDR control method

Propose using the Benjamini-Hochberg procedure to control FDR at 10%. Explain that it ranks p-values and compares them to thresholds based on rank and total tests.

3. Apply to secondary metrics

Detail that you would apply BH to the 12 secondary metrics, possibly after confirming primary metric significance. Note that FDR control allows some false discoveries but limits their proportion.

4. Interpret and communicate discoveries

Explain that a 'discovery' means the metric's p-value passed the BH threshold, so it's statistically significant after FDR control. But stress that it's not definitive proof; consider effect size, confidence intervals, and business relevance.

5. Stakeholder communication

Describe how to convey results to non-technical stakeholders: use plain language, avoid jargon, and emphasize that discoveries are hypotheses for further validation, not final truths.

Key Points to Mention

  • False discovery rate (FDR) vs family-wise error rate (FWER)
  • Benjamini-Hochberg procedure and its assumptions
  • Adjusted p-values or q-values
  • Practical significance vs statistical significance
  • Effect size and confidence intervals
  • Pre-registration of metrics and analysis plan

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

Q6

Given 10% noncompliance in treatment and 3% contamination in control, compute the ITT effect versus the complier average causal effect (CACE). How would you present both numbers to a product team without misleading them?

A/B Testing & ExperimentationStakeholder ManagementProduct Analytics & Metrics
Author's notes

ITT uses everyone as assigned regardless of what they saw, CACE adjusts for the actual compliance rate.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, clarify that ITT measures the effect of assignment to treatment, while CACE measures the effect among compliers. Then, compute both using the given noncompliance and contamination rates, and discuss how to present them to a product team by emphasizing the trade-off between real-world impact and efficacy.

Pro tip: Always present ITT as the conservative, policy-relevant estimate and CACE as the per-protocol effect, but caution that CACE relies on assumptions like exclusion restriction and monotonicity. Use a simple analogy, such as comparing the effect of offering a coupon (ITT) versus the effect on those who actually use it (CACE).

1. Define ITT and CACE

Explain that ITT estimates the effect of treatment assignment, regardless of compliance, while CACE estimates the effect among those who comply with their assigned treatment.

2. Compute ITT effect

Given 10% noncompliance in treatment and 3% contamination in control, calculate the ITT effect as the difference in outcomes between the treatment and control groups as assigned.

3. Compute CACE

Use the formula CACE = ITT / (compliance rate in treatment - contamination rate in control) to estimate the effect among compliers.

4. Interpret and present to product team

Present ITT as the expected impact if the treatment is rolled out to the entire population, and CACE as the potential impact if perfect compliance were achieved. Highlight that CACE is an estimate with assumptions and may not reflect real-world scalability.

5. Address assumptions and limitations

Mention key assumptions for CACE (e.g., exclusion restriction, monotonicity) and caution that violations could bias the estimate. Recommend using ITT for decision-making unless there's a strong reason to focus on compliers.

Key Points to Mention

  • ITT is conservative and reflects real-world effectiveness; CACE reflects efficacy under perfect compliance.
  • CACE requires assumptions: exclusion restriction (treatment assignment affects outcome only through treatment received) and monotonicity (no defiers).
  • The formula for CACE: ITT divided by the difference in compliance rates between treatment and control groups.
  • Noncompliance and contamination dilute the ITT effect, making it smaller than CACE.
  • Present both numbers with clear labels and explain the trade-offs: ITT for policy, CACE for understanding mechanism.
  • Use visual aids or simple examples to communicate the difference to non-technical stakeholders.

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