Skip to content
ML Academy

Analytics

Interview question

  • seniorlevel
  • 15 minsuggested time
  • 2Data analyst, Data scientist

The airline's post-flight survey was sent to 50,000 passengers (hypothetical): 10,000 business-cabin and 40,000 economy. Response rates were 30% and 6%, so 3,000 business and 2,400 economy answered. The mean score is 8.1 among business respondents and 7.0 among economy respondents. The dashboard shows 7.61 and a rising trend, and the head of customer experience wants it as the company-wide figure. What is wrong, what would you publish instead, and how confident can you be in it?

Tests: Correcting for unequal response across known groups (post-stratification or inverse-probability weights), computing the effective sample size and the variance cost of weighting, and stating what weighting cannot fix.

day 4day 5day 6

Sign in to practise: answers are saved before model answers are shown.