ISI - AIM

Abstract

During the last decades, closed population capture-recapture models have been widely used to estimate the number of faults in a system. Models that jointly account for heterogeneity in faults detection probabilities, differences in reviewers’ detection probabilities, and simplifications of these models have been proposed in the literature (i.e Mth, Mh, Mt and M0). A special class of models, that has been largely neglected, is one that account for differences between first and subsequent fault detection probabilities. One may think that there is no reason to believe that after detecting a given fault by a reviewer the first time, its detection probability would be different for other reviewers. This sort of heterogeneity, known as “behavioral heterogeneity” in the wildlife literature, could be associated to a potential lack of independence among reviewers. Suppose for instance, that after first detection of a fault by any inspector, other reviewers became some how aware o the type of fault. In this study, we reanalyzed a data set from a control experimental setting by Thelim et al. (2004), and noticed that models of the type Mbh, Mtb were also possible. We question the assumption of independent detection among reviewers and proposed that Mb–type models could be used as an indicator potential violation of this assumption. We tried out different approaches for data analyses to assess this hypothesis.