Multiple-dose design and bias-reducing methods for limiting dilution assays

This paper gives an overview of several (mostly recent) statistical contributions to the theory of Limiting and Serial Dilution Assays (LDA's, SDA's). A simple and useful method is presented for the setup of a design for an LDA or an SDA. This method is based on several user-supplied design parameters, consisting in the researcher's advance information and other parameters inherent to the particular problem. The commonly used Maximum Likelihood (ML) and Minimum Chi-square methods for the estimation of the unknown parameter in an LDA or an SDA are described and compared to several bias-reducing estimation methods, e.g. jackknife and bootstrap versions of the ML method. One particular jackknife version is recommended.