Development of a guide to statistics for maintenance quality assurance programs in transportation

Abstract

This report provides maintenance managers and practitioners with knowledge of how to apply statistics in MQA programs. Literature were reviewed and MQA manuals from 10 states were synthesized to understand state-of-practice for managing statistics in MQA programs. It was observed among lead states that a wide range of measured elements and threshold definitions exist for roadway, roadside and vegetation, drainage, traffic control, and rest areas. The role of statistics in MQA was described, and key statistical terms were defined. Actual MQA data were collected and analyzed to illustrate how fundamental statistical procedures and applications to describe features of MQA data. Examples were demonstrated for basic summary statistics, confidence intervals, data stratification, analysis of variance, sample size determination, precision, sensitivity analysis, and power level of statistics when evaluating the QA process itself. A new ranking procedure was developed that applies the concepts of percent defective (PD) or percent within limits (PWL), and computes confidence limits for the statewide system or any strata (functional class, division, county, or 10-mile roadway segment), and then ordering them by their lower confidence limits. If those limits do not include the desired level of the particular measure, then that particular section of pavement would be considered noncompliant.