A method is provided for building improved calibration models or for improving modifications to these models or validation of the models used in the non-invasive spectroscopic measurement of an analyte or attribute of tissue. The method uses a matched reference sample and measurement of that sample to ensure that the correct relationship between the spectra and analyte is made during the model building, modification or calibration process. A matched reference sample is one in which the analyte of interest or attribute of interest in the reference sample is representative of the analyte or attribute at the site being non-invasively sampled or the agreement between the reference concentration and the non-invasively sampled concentration is clinically significant.