Estimating the fraction of ‘non-genuine’ artwork by Henry Moore for sale on eBay: application of latent class screening test methodology


Joseph L. Gastwirth, Department of Statistics, George Washington University, Washington DC 20052, USA.


Summary.  Although the Internet has provided consumers with new ways to purchase goods, it also has allowed less scrupulous businesses and individuals to offer poor quality or mislabelled items. In particular, the operators of Internet auction sites may not be required to guarantee the quality or genuineness of the items listed. A latent class approach that was originally developed to estimate the accuracy of medical tests and the prevalence of an infection is adapted to provide estimates of the prevalence of ‘questionable’ artwork offered on eBay. The method requires the existence of two subpopulations with a different prevalence of the trait and each subject is classified by two different tests. As Henry Moore produced less expensive lithographs and etchings, in addition to small sculptures and original drawings, two subpopulations of artwork were available. Two knowledgeable individuals independently assessed all the works by Moore offered on eBay for a period of 18 months. After discussing the issue of non-genuine versions of Moore's art with experts on Henry Moore, a Bayesian approach was adopted. The method estimates that about 90% of the drawings and small sculptures that are listed on eBay are not genuine whereas most, about 90–95%, of signed prints are genuine. Because the methodology does not require a ‘gold standard’ it can be adopted by consumer protection agencies or producers of expensive items to monitor the authenticity of those objects offered to the public on Internet auction sites.