Limits of Agreement Method for Comparison of Pavement Friction Measurement

Abstract

Previous comparisons of pavement friction equipment focused on the harmonization of devices that measure pavement friction, but did not assess the agreement of the measurements between them. Harmonization was accomplished with normalization of the measurements through adjustment to a standard slip speed and use of macrotexture measurements in models of the friction–speed relationship. A problem with this procedure was that the true value of friction on which the comparisons were made was assumed. When true values remain unknown, valid comparisons between two devices require the use of agreement assessments. Limits of agreement (LOA) were used to compare two pavement friction devices. The first comparison was made with one run to illustrate LOA. A second comparison with two runs was made to gain understanding of their repeatability and use in LOA. A more complex LOA procedure was conducted with multiple runs of the two devices. The results of another study that used LOA between another pair of pavement friction devices showed a better agreement. The obtained LOA results showed a 95% confidence level, which was considered not adequate. A comparison between the 95% repeatability coefficients and the 95% LOA computed for the two devices indicated that factors other than repeatability were responsible for the lack of agreement. Use of the LOA method of analysis should be considered before harmonization of different devices. Predictive models based on correlation are not recommended for comparing pavement friction measuring devices.