Insurance Research & Evidence-Based Practice 2 — Questions and Answers
Question 1: In insurance research, what does a 'confidence interval' represent?
- The range within which the true population parameter likely falls (Correct answer)
- The probability that a claim will be approved
- The maximum coverage limit for a policy
- The percentage of policyholders who file claims annually
Correct answer: The range within which the true population parameter likely falls
A confidence interval defines the range where the true population value is expected to fall with a specified level of certainty (e.g., 95%).
Question 2: Which study design is considered the gold standard for establishing causality in insurance outcome research?
- Cross-sectional survey
- Randomized controlled trial (Correct answer)
- Retrospective cohort study
- Case series report
Correct answer: Randomized controlled trial
Randomized controlled trials minimize confounding by randomly assigning subjects to treatment and control groups, establishing the strongest causal evidence.
Question 3: An insurer uses a loss development triangle to estimate future claim payments. What is the primary purpose of this actuarial tool?
- To compare premiums across competitors
- To project ultimate losses from immature claim data (Correct answer)
- To calculate the probability of a policyholder lapsing
- To determine agent commission structures
Correct answer: To project ultimate losses from immature claim data
Loss development triangles show how reported claims mature over time, allowing actuaries to estimate incurred but not yet fully developed losses.
Question 4: What is 'selection bias' in the context of insurance research?
- Choosing a deductible that is too high for the insured
- Systematic error from non-random differences between study and comparison groups (Correct answer)
- An insurer's preference for certain risk types
- The tendency of adjusters to favor claimants
Correct answer: Systematic error from non-random differences between study and comparison groups
Selection bias occurs when the study sample is not representative of the population, leading to skewed or invalid conclusions.
Question 5: A meta-analysis in insurance research primarily serves to:
- Interview a large panel of policyholders simultaneously
- Pool and synthesize results from multiple independent studies (Correct answer)
- Create a single large prospective cohort
- Replace actuarial judgment with automation
Correct answer: Pool and synthesize results from multiple independent studies
Meta-analysis statistically combines findings from several studies to produce a more precise overall estimate of an effect.
Question 6: Which metric best measures the predictive accuracy of a binary insurance underwriting model?
- Standard deviation of premiums
- Area Under the ROC Curve (AUC) (Correct answer)
- Average claim frequency
- Combined ratio
Correct answer: Area Under the ROC Curve (AUC)
The AUC (ROC curve area) quantifies how well a binary classifier distinguishes between two outcomes, making it standard for evaluating underwriting models.
Question 7: When an insurance researcher reports a p-value of 0.03, it means:
- There is a 3% chance the hypothesis is correct
- There is a 3% probability of observing results this extreme if the null hypothesis is true (Correct answer)
- The study has 97% statistical power
- The finding is practically significant for pricing purposes
Correct answer: There is a 3% probability of observing results this extreme if the null hypothesis is true
A p-value of 0.03 means there is only a 3% probability of obtaining the observed results (or more extreme) if the null hypothesis were true.
Insurance Research & Evidence-Based Practice 2
In insurance research, what does a 'confidence interval' represent?