Activity Number:
|
50
|
Type:
|
Contributed
|
Date/Time:
|
Sunday, August 11, 2002 : 4:00 PM to 5:50 PM
|
Sponsor:
|
Section on Statistical Education*
|
Abstract - #300908 |
Title:
|
Appropriateness of the Paired T-Test from Poisson Data: A Case Study of the Effect of Surgery and Streptococcus Pneumoniae and Pseudomonas Aeruginosa on Hearing Loss
|
Author(s):
|
Stacia Merkel*+ and Byron Gajewski+
|
Affiliation(s):
|
St. Cloud State University and St. Cloud State University
|
Address:
|
720 4th Ave S, St. Cloud, Minnesota, 56301-4498, U.S. 720 4th Avenue South, St. Cloud, Minnesota, 56301, USA
|
Keywords:
|
Power ; Method of Moments ; Simulation
|
Abstract:
|
Undergraduate statistics texts indicate that the paired t-test is inappropriate for non-normal data with small sample sizes. Robustness of the paired t-test to these assumptions is assessed with paired Poisson data motivated from an actual medical study. The data in the medical study are discrete, violating the assumption of the paired t-test. The Method of Moments was used to generate parameters for the Poisson distribution. From simulated data, we were able to show that as the sample size increases, power increases at an acceptable rate. Therefore, we argue that the paired t-test is appropriate in this paired Poisson setting.
In the medical study, the goal was to quantify the significance of middle ear infection and surgery on hearing loss. Thirty guinea pigs were randomly assigned into three groups based on the type of infection induced. Two days after middle ear injection, the guinea pigs underwent surgery. Hearing was measured immediately before, and seven days following surgery. Electrocochleography was performed to measure hearing threshold. Clicks were measured at different discrete decibel levels.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |