|
Activity Number:
|
109
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 7, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
| Abstract - #307030 |
|
Title:
|
Estimating Correlation with Multiply Censored Data
|
|
Author(s):
|
Elizabeth Newton*+ and Ruthann Rudel
|
|
Companies:
|
Silent Spring Institute and Silent Spring Institute
|
|
Address:
|
29 Crafts Street, Newton, MA, 02458,
|
|
Keywords:
|
correlation ; censored ; detection limit ; environmental data
|
|
Abstract:
|
Environmental data frequently are left censored due to detection limits of laboratory procedures. This presents difficulties in statistical analysis of the data. Here we examine methods for estimating the correlation between two variables each of which is multiply censored. We introduce a maximum likelihood estimator (MLE) that, instead of estimating all parameters simultaneously, relies on more accurate estimates of mean and variance. We compare ML methods with Kendall's tau-b, a modification Kendall's tau adjusted for censoring, and several commonly used ad-hoc methods: correlations estimated with non-detects set to DL/2 and correlations of detects only (DET). With increasing levels of censoring most methods are highly biased. The ad-hoc methods in general tend toward zero if singly censored and one if multiply censored. Based on RMSE, DET performs the worst and MLE the best.
|
