This report describes the results of a project to investigate the quality of road survey data collected from road networks in New Zealand.
The objective was to establish a statistical technique which could be used to identify whether data were inconsistent with data from previous years' surveys.
Seven databases were assembled with data in the format suitable for time series analyses from different road controlling authorities. This proved to be a complicated process due to problems with how the data were stored and referenced.
The data were evaluated on 100-m road sections and major variations in condition between years were found which could not be ascribed to pavement deterioration. In many instances these were due to maintenance applied to the pavement that was not always recorded in the databases.
The roughness data were considered to be the most suitable for time series analysis, and they were analysed using a Box-Jenkins approach with the software application Autobox. While Autobox proved suitable for identifying some of the deficiencies, it was considered that an alternative approach incorproating the spatial component and more suitable for small sample sizes would be likely to give better results.
Keywords: condition rating, data quality, databases, exception reporting, RAMM, outlier, pavements, roads, roughness, statistics, surveys, time series, variability