LQAS: User Beware

Abstract

Researchers around the world are using Lot Quality Assurance Sampling (LQAS) techniques to assess public health parameters and evaluate program outcomes. In this paper, we report that there are actually two methods being called LQAS in the world today, and that one of them is badly flawed. This paper reviews fundamental LQAS design principles, and compares and contrasts the two LQAS methods. We raise four concerns with the simply-written, freely-downloadable training materials associated with the second method. The first method is founded on sound statistical principles and is carefully designed to protect the vulnerable populations that it studies. The language used in the training materials for the second method is simple, but not at all clear, so the second method sounds very much like the first. On close inspection, however, the second method is found to promote study designs that are biased in favor of finding programmatic or intervention success, and therefore biased against the interests of the population being studied. We outline several recommendations, and issue a call for a new high standard of clarity and face validity for those who design, conduct, and report LQAS studies.