Skip to content

IAI Actuarial Core Principles · Actuarial Statistics · Purpose and function of data analysis

An insurer has 2,000 motor policy records. During the cleaning step, 80 records have a missing vehicle age. The analyst finds that missingness is unrelated to any variable. Which action is most appropriate and least likely to bias the estimate of mean claim cost?

Use the 1,920 complete records, or impute in a documented way, and record the treatment. Since the data are missing at random, this causes little bias. Filling with zero or the maximum creates artificial values that distort results, and discarding everything is wasteful.

  1. ADelete the whole dataset and recollect all data
  2. BSet every missing vehicle age to zero
  3. CReplace the 80 missing ages by the maximum observed age
  4. DAnalyse the 1,920 complete records, or impute using a documented method, and record the treatment usedCorrect
  5. Remove the vehicle age field and the claim cost field

Explanation

With missingness unrelated to any variable (completely at random), removing the 80 records or imputing sensibly does not bias the results. Setting ages to zero or the maximum introduces artificial values. Whatever treatment is used should be documented so the analysis can be reproduced.

Did you get it right without looking?

One question tells you little. A timed set on Purpose and function of data analysis shows your real accuracy, how long you take and where you lose marks.

Practise Purpose and function of data analysisFree diagnostic mock

More Purpose and function of data analysis questions