Antimicrobial Resistance (AMR) can be measured in different ways. We compare two classes of measurements (sample-level and isolate-level) from faeces samples of livestock. We estimate the agreement of both measurement types. We used simulation to extrapolate onto historic prevalence data, to predict the prevalence we would have found, had we used a different test at that time. The sample-level screening provides higher estimate of prevalence than the isolate-level of prevalence. We propose that such data offers evidence supporting a reappraisal of measuring AMR for epidemiological purposes. In particular we suspect that sample-level screening may be in some epidemiological studies more relevant than the commonly used isolate-level screening.