Sample Size and Precision for Pavement Inspection in a Maintenance Quality Assurance Program

Considering the large amount of highway asset items, a statistically valid data sampling method is required for the level of service inspection of highways. The presented paper discussed the determination of sample size for the pass/fail highway inspection approach. The historical highway inspection data from Tennessee were collected for demonstration. It was found that passing percent is the most significant factor for sample size while population size has little influence. An effective approach to improve the precision at lower management levels without increasing sample size is to use an average sample size for all different subgroups. To investigate the necessity of stratified sampling, statistical paired t-tests were conducted to identify if the passing percent of subgroups are significantly different with the whole state. The results showed no significant difference between interstates and state routes. Although sampling at district levels could potentially improve the significance level of half of the districts, the sample size would be multiplied and is not cost-effective.