← Meta Interview Insights

Meta·Data Scientist·Technical Phone Screen·Intermediate

Intermediate
May 2026

Summary

Meta data scientist screen focused on probability and Bayes' rule applied to ad conversion scenarios. Pretty straightforward if you've brushed up on conditional probability, but the follow-up about reversing the conditioning tripped me up a bit.

Questions Asked (1)

Q1

You're given a prior conversion rate of 2%, with conditional rates of 3% for male users and 1% for female users. After learning the user is male, what's the updated conversion probability and how do you get there? Then flip it: if a conversion happened, what's the probability the user was male?

Product Analytics & MetricsA/B Testing & Experimentation
Author's notes

The first part I handled fine, it's just plugging into the conditional directly since gender is observed.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, clarify that the overall conversion rate is a weighted average of the gender-specific rates, so the male rate (3%) is the updated probability after learning the user is male. Then, for the reverse, apply Bayes' theorem using the prior gender distribution implied by the overall rate, or state that it cannot be determined without the proportion of male users.

Pro tip: Always check if the prior gender mix is given; if not, explicitly state the assumption (e.g., 50/50) and show how the answer changes. This demonstrates rigor and avoids silent errors.

1. Clarify the given rates

Identify the overall conversion rate (2%) and the conditional rates for male (3%) and female (1%). Recognize that the overall rate is a weighted average of these conditional rates.

2. Compute updated probability after learning gender

Since the user is male, the updated conversion probability is simply the male conversion rate, 3%. No further calculation is needed because the condition is directly given.

3. Determine prior gender distribution

Use the overall rate and conditional rates to solve for the proportion of male users. Let p be the proportion male: 0.02 = 0.03p + 0.01(1-p). Solve to find p = 0.5 (assuming the rates are accurate).

4. Apply Bayes' theorem for reverse probability

Compute P(Male | Conversion) = P(Conversion | Male) * P(Male) / P(Conversion) = (0.03 * 0.5) / 0.02 = 0.75. So there is a 75% chance the user was male given a conversion.

5. Discuss assumptions and limitations

Note that the calculation relies on the overall rate being exactly the weighted average and on the gender proportions being 50/50. If the proportions are unknown, the reverse probability cannot be determined without additional information.

Key Points to Mention

  • Conditional probability and Bayes' theorem
  • Weighted average relationship between overall and subgroup rates
  • Importance of base rates (prior gender distribution)
  • Assumption of 50/50 gender split derived from the given rates
  • Calculation: P(Male|Conversion) = (0.03 * 0.5) / 0.02 = 0.75
  • Caveat: if gender proportions are not given, the reverse probability is indeterminate

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