In this short communication, we discuss alternative ways to estimate accordance and concordance, which are, for qualitative methods, the analogues of the well-known concepts repeatability and reproducibility. We argue that estimators in a random framework appear to be more appropriate than the estimators in a fixed framework as they were originally introduced by Langton et al. [International Journal of Food Microbiology 79 (2002) 171]. A simple example of their application is given.