Statistical tools for the analysis of nutrition effects on the survival of cohorts

We discuss various methods which can be employed for the comparative analysis of samples of response curves. In the application discussed here, these curves are hazard functions, each generated by the survival data obtained for a cohort of experimental subjects which are fed a specific diet. It is demonstrated how comparisons of the effects of different diets on survival can be carried out by employing statistical techniques for inference on samples of curves. The methods are illustrated with data on the survival of large cohorts of male and female Mediterranean fruit flies under full diet and under protein deprivation. These statistical methods allow one to investigate differences between the samples of hazard functions generated by the four groups defined by combinations of sex and diet.