297 – General Methodological Advances
A Databased Measure for Statistical Inference
Walid Nuri
This paper presents an alternative to the probabilistic approach in measuring events of a sample space and in defining a method making a decision based on a specific statistical decision rule. It starts by introducing the concept of a population image that is created using the measurements of the provided random sample (selected for inference about a population parameter). Using the population image, this approach employs a computer simulation technique to select (with replacement) a large number m, say, of random samples from the measurements Then, for an event E in a sample space, the function m(E) is defined as the number of cases within these samples that support the event E. The measure is defined as the limiting value of m(E) as m → ∞. It is called a measure of favorability. Statistical inference problems frequently encountered in applications are discussed using this new approach to measuring events. Actual examples are given to show how to apply this approach to these inference problems.
