Academic Journals Database
Disseminating quality controlled scientific knowledge

Binomial Distribution Sample Confidence Intervals Estimation 6. Excess Risk

Author(s): Sorana BOLBOACĂ | Andrei ACHIMAŞ CADARIU

Journal: Leonardo Electronic Journal of Practices and Technologies
ISSN 1583-1078

Volume: 3;
Issue: 4;
Start page: 1;
Date: 2004;
Original page

Keywords: Confidence interval estimation | Excess risk | Risk factors assessments

We present the problem of the confidence interval estimation for excess risk (Y/n-X/m fraction), a parameter which allows evaluating of the specificity of an association between predisposing or causal factors and disease in medical studies. The parameter is computes based on 2x2 contingency table and qualitative variables. The aim of this paper is to introduce four new methods of computing confidence intervals for excess risk called DAC, DAs, DAsC, DBinomial, and DBinomialC and to compare theirs performance with the asymptotic method called here DWald.In order to assess the methods, we use the PHP programming language and a PHP program was creates. The performance of each method for different sample sizes and different values of binomial variables were assess using a set of criterions. First, the upper and lower boundaries for a given X, Y and a specified sample size for choused methods were compute. Second, were assessed the average and standard deviation of the experimental errors, and the deviation relative to imposed significance level α = 5%. Four methods were assessed on random numbers for binomial variables and for sample sizes from 4 to 1000 domain.The experiments show that the DAC methods obtain performances in confidence intervals estimation for excess risk.
     Live and Work Abroad