Reality in Statistical Modeling of Epidemiologic Data
Epidemiologic designs are used in gathering public health data for risk identification, programs planning and outcomes investigation. Often these data are not collected from random probability sampling of the study participants, and yet we quantify the random errors in these studies. Such practice should be discouraged in evidence discovery. Therefore when random variables, which are based on probability sampling of study participants are used, only then can we quantify the random error with a probability value ( p value). Without such sampling,(probability sampling method) with exception being large registry databases studies, results of epidemiologic studies should remain at a descriptive statistics level.