Dataford
Popular roles
Software EngineerData AnalystData ScientistData EngineerBusiness AnalystAI EngineerMachine Learning EngineerProduct Manager
Browse
Browse All RolesEvery role hub, from analyst to MLBrowse All CompaniesCompany-specific interview loopsAll Interview GuidesThe full guide library
Top questions by role
Software EngineerData AnalystData ScientistData EngineerBusiness AnalystAI EngineerMachine Learning EngineerProduct Manager
Top questions by skill
SQLPythonStatisticsMachine LearningA/B TestingSystem DesignGenerative AIProduct SenseMetricsBehavioral
Browse all questions →Try a mock interview
ExperiencesCoaching
New
Practice
Mock InterviewsTimed interview simulations with feedbackSuccess PathYour 6-week structured planModulesCurated lessons by topicWebinarsTalks from ex-Big Tech data leadsPlaygroundA free-form scratch editor
Learn
BlogInterview strategy and career adviceTech Job Market ReportHiring trends across data and AI rolesFor UniversitiesDataford for career centersAbout DatafordWho we are and how we build
Pricing
Get started

Interpreting P-Values in Fraud Screening

MediumStatistics & Probability00:00
Practice interviewer
In session
5 left
00:00

Whenever you're ready, walk me through how you'd approach this.

You need to log in / sign up to chat or submit.

Problem

Business Context

PayWave, a fintech company, launched a new fraud-screening rule and tested whether it increases the false-positive rate above the historical benchmark. A product manager sees a small p-value and says, “So there’s only a 2% chance the null hypothesis is true.”

Problem Statement

Use the test results below to explain quantitatively why a p-value is not the probability that the null hypothesis is true. Show both the frequentist interpretation of the p-value and a separate Bayesian calculation for the probability that the null is true after seeing the data.

Given Data

PayWave tested whether the false-positive rate is higher than the historical rate of 5.0%.

QuantityValue
Historical false-positive rate under null, p0p_0p0​0.050
Number of reviewed transactions, nnn400
Observed false positives, xxx30
Observed sample rate, p^\hat{p}p^​0.075
Significance level, α\alphaα0.05
Prior probability null is true, P(H0)P(H_0)P(H0​)0.80
Prior probability alternative is true, P(H1)P(H_1)P(H1​)0.20
Likelihood of seeing data this extreme under H1H_1H1​, P(D∣H1)P(D \mid H_1)P(D∣H1​)0.25

Assume a one-sided hypothesis test: the team only cares whether the false-positive rate increased.

Requirements

  1. State the null and alternative hypotheses.
  2. Compute the z-statistic and one-sided p-value.
  3. Explain in words what the p-value means.
  4. Compute P(H0∣D)P(H_0 \mid D)P(H0​∣D) using Bayes' theorem with the provided priors and likelihood under H1H_1H1​.
  5. Compare the p-value to P(H0∣D)P(H_0 \mid D)P(H0​∣D) and explain why they are different quantities.
  6. State the correct business interpretation for the fraud-rule decision.

Assumptions

  • The 400 reviewed transactions are independent.
  • The normal approximation for the sample proportion is acceptable.
  • The Bayesian prior and P(D∣H1)P(D \mid H_1)P(D∣H1​) are provided for illustration; they are not derived from the z-test itself.