Missing Data Methods for Partial Correlations

Journal of Biometrics & Biostatistics

article

Gina M D’Angelo¹  JingqinLuo¹  ChengjieXiong¹ 

[1] Division of Biostatistics, Washington University School of Medicine

关键词: Partial correlation;    Fisher-z transformation;    Missing data;    Missing at random;    Expectation-maximization algorithm;    Alzheimer?s disease;   

DOI  :  10.4172/2155-6180.1000155

来源: Hilaris Publisher

【 摘 要 】

In the dementia area it is often of interest to study relationships among regional brain measures; however, it is often necessary to adjust for covariates. Partial correlations are frequently used to correlate two variables while adjusting for other variables. Complete case analysis is typically the analysis of choice for partial correlations with missing data. However, complete case analysis will lead to biased and inefficient results when the data are missing at random. We have extended the partial correlation coefficient in the presence of missing data using the expectation-maximization (EM) algorithm, and compared it with a multiple imputation method and complete case analysis using simulation studies. The EM approach performed the best of all methods with multiple imputation performing almost as well. These methods were illustrated with regional imaging data from an Alzheimer’s disease study.

【 授权许可】

Unknown