CSP 2017 Online Program

Additive P-Value Combinations and an Application in Consumer Product Research (303433)

Keywords: p-value, correlated p-value, beta distribution, inverse normal

Combining multiple hypothesis into one overall hypothesis is a technique from the early 20th century that has emerged as a 21st century method for large scale hypothesis testing such as in 'omics'. The two main methods are 'ordinal' and additive. Ordinal works best for the large number of tests such as gene arrays but additive techniques that rely on transforming the p-values and adding to get a test statistic can be used with less numerous tests. The poster will show popular additive techniques, discuss correlations and weighting, R programming approaches and show an application from consumer product development. It might include a live software application of p-value combining.