Highway pavement distress evaluation: Modeling measurement error

There has been a proliferation of inspection technologies to quantify distresses on highway pavement systems. These technologies employ varying measurement principles and are subject to measurement errors. Estimates of measurement errors are therefore required in order to select among these techniques, and to get accurate assessments of pavement condition. There is abundant literature concerning techniques available for the numerical study of measurement errors. Available techniques include econometric methods of coping with errors in variables when investigating relationships between variables that have been measured with error; calibration approaches under various measurement conditions; and evaluation of measurement errors due to quantification, by a proxy, of concepts that are not directly measurable or observable. The methodologies employed in these techniques rely on certain assumptions for model development and estimation. These assumptions include the nature of error occurrence, whether systematic or random; the effect of these errors on the measured result, whether multiplicative or additive; and the level of knowledge about the true value of the measured object. Such assumptions may be violated under certain conditions. This paper identifies such situations and develops a generalized measurement error modeling approach, in which existing approaches are special cases. Existing methods are reviewed as to the specification models used to represent measurement errors; the types of errors accounted for in the suggested specification model; the type of errors not accounted for and the biases induced in estimated parameters by ignoring certain error types. The approach developed is capable of quantifying the accuracy of measurement for cases where the true value of the measured object is not known. This is the case in highway pavement distress evaluation as there is no single well-accepted technology which can be used as a proxy for the value for calibration purposes. The methodology developed in this paper explicitly estimates the true value and is applied to the calibration of new technologies for highway distress evaluation.