Microarray Studies - Assessing Statistical Significance

Doctor Anil Malhotra discusses the problem of deciding what meets the criteria for statistical significance for microarray studies.

I think it is fair to say that different research groups have different ideas and theories about how to achieve or what meets the criteria for statistical significance. Many groups apply a very simple correction called the Bonferroni correction, which is somewhat conservative and leads to a p-value which is being required on the order of 10-7 or 10-8 for a whole genome association study. Other groups have taken the approach of simply examining the best 20, 50, or 100 results in any individual data set and then comparing those to another data set perhaps across the top 20, 50, or 100 and looking for commonalities between these kinds of studies.