Activity Number:
|
531
- SPEED: Statistical Computing: Methods, Implementation, and Application, Part 2
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, July 31, 2019 : 11:35 AM to 12:20 PM
|
Sponsor:
|
Section on Statistical Computing
|
Abstract #307946
|
|
Title:
|
Exact Inference for Analyzing Contingency Tables in Finite Populations
|
Author(s):
|
Shiva Dibaj* and Gregory Wilding and Graham Warren
|
Companies:
|
UT MD Anderson Cancer Center and SUNY at Buffalo and University of Kentucky
|
Keywords:
|
Exact test;
Hypergeometric;
Categorical data;
Computation;
Finite population;
Survey sample
|
Abstract:
|
With recent developments in computer power the application of exact inferential methods has become more feasible which has resulted in increasing popularity of these approaches. However, there is a lack of such methodology for populations with more complex structure, such as finite populations. When a small sample is drawn from a finite population, the number of individuals with a specific characteristic of interest follows hypergeometric distribution. In order to test for the comparison of two proportions in finite populations we develop an exact unconditional test. To calculate the exact p-value, all possible realizations of the test statistic along with their probabilities are determined. We utilize the information gained from the sample to restrict our search for the maximum p-value. Our proposed test has power equal to its competitors while maintains the pre-specified nominal significance level.
|
Authors who are presenting talks have a * after their name.