Binomial Distribution Sample Confidence Intervals Estimation 5. Odds Ratio
|
 |
|
Post a Comment
|
|
|
|
|
|
ABSTRACT:
Evaluation of the strength of association between predisposing or causal factors and disease can be express as odds ratio in case-control studies. In order to interpret correctly a point estimation of odds ratio we need to look also to its confidence intervals quality. The aim of this paper is to introduce three new methods of computing the confidence intervals, R2AC, R2Binomial, and R2BinomialC, and compare the performances with the asymptotic method called R2Wald.
In order to assess the methods a PHP program was develop. First, the upper and lower confidence boundaries for all implemented methods were computes and graphically represented. Second, the experimental errors, standard deviations of the experimental errors and deviation relative to the imposed significance level á = 5% were assessed. Estimating the experimental errors and standard deviations at central point for given sample sizes was the third criterion. The R2Wald and R2AC methods were assessed using random binomial variables (X, Y) and sample sizes (m, n) from 4 to 1000.
The methods based on the original method Binomial adjusted for odds ratio (R2Binomial, R2BinomialC functions) obtain systematically the lowest deviation of the experimental errors percent relative to the expected error percent and the R2AC method, the closest average of the experimental errors percent to the expected error percent.
|
|
|
|
STATISTICS
|
|
Click on # to view
|
|
Citations
|
|
4
|
|
References
|
|
2
|
|
Comments
|
|
0
|
|
Quality
|
|
0/0.00
|
|
Interest
|
|
0/0.00
|
|
View(er)s
|
|
1/672
|
|
|
|
|
|
|
| Prev |
Next |
|