Represents a serious risk associated with conflating correlation and causation when evaluating a road safety program?

• A. Random measurement error
• B. Sample bias
• C. Lack of reliability
• D. Crash reductions may be attributed to the road safety program, when they may be due to other factors

Answer: (D)

When evaluating a road safety program, it’s common to see crashes drop after implementation and assume the program caused the improvement. The real danger here is confusing a nearby association with a true cause. Crashes can change for many reasons unrelated to the program— broader traffic trends, seasonal effects, weather, changes in enforcement, economic shifts, or other concurrent road safety efforts. Without a proper comparison or counterfactual, you can’t separate the program’s effect from these other factors. That’s why this option highlights the key risk: crash reductions might be attributed to the program when, in fact, they result from something else. To avoid this, use evaluation designs that create a fair comparison group or account for confounders, so you can more confidently infer whether the program really caused the change. The other data-quality issues—random measurement error, sample bias, and lack of reliability—affect data quality, but they don’t capture the specific pitfall of inferring causation from mere correlation in program outcomes.