Estimation of the correlation coefficient using the Bayesian Approach and its applications for epidemiologic research

BMC Medical Research Methodology

Malla Rao²  Lucinda J England²  Kirsten B Moysich¹  Enrique F Schisterman² 

[1] Department of Cancer Control, Epidemiology and Biostatistics, Roswell Park Cancer Institute, Buffalo, NY, USA;Division of Epidemiology, Statistics and Prevention National Institute of Child Health and Human Development / National Institute of Health, Bethesda, MD, USA

关键词: Meta-Analysis and Transformations;    Low Birthweight;    Correlation Coefficients;    Bayesian Analysis;   

Others  :  1143396
DOI  :  10.1186/1471-2288-3-5

 received in 2002-09-06, accepted in 2003-03-25,  发布年份 2003

【 摘 要 】

Background

The Bayesian approach is one alternative for estimating correlation coefficients in which knowledge from previous studies is incorporated to improve estimation. The purpose of this paper is to illustrate the utility of the Bayesian approach for estimating correlations using prior knowledge.

Methods

The use of the hyperbolic tangent transformation (ρ = tanh ξ and r = tanh z) enables the investigator to take advantage of the conjugate properties of the normal distribution, which are expressed by combining correlation coefficients from different studies.

Conclusions

One of the strengths of the proposed method is that the calculations are simple but the accuracy is maintained. Like meta-analysis, it can be seen as a method to combine different correlations from different studies.

【 授权许可】

   
2003 Schisterman et al; licensee BioMed Central Ltd. This is an Open Access article: verbatim copying and redistribution of this article are permitted in all media for any purpose, provided this notice is preserved along with the article's original URL.

【 参考文献 】

• [1]Liu K: Measurement Error and its impact on partial correlation and multiple linear regression analyses. Am J Epidemiol 1988, 127:864-874.
• [2]Hakstian AR, Schroeder ML, Rogers WT: Inferential theory for partially disattenuated correlation coefficients. Psychometrika 1989, 54:397-407.
• [3]Millsap RE: Sampling variance in attenuated correlation coefficients: a Monte Carlo Study. J Appl Psychol 1988, 73:316-319.
• [4]Bashir SA, Duffy SW: The Correction of Risk Estimates for Measurement Error. Ann Epidemiol 1997, 7:154-164.
• [5]Dear KBG, Puterman ML, Dobson AJ: Estimating correlations from epidemiological data in the presence of measurement error. Statistics in Medicine 1997, 16:2177-2189.
• [6]Spiegelman D, Cassella M: Fully parametric and semi-parametric regression models for common events with covariant measurement error in main study/validation study designs. Biometrics 1997, 53:395-409.
• [7]Lee PM: Bayesian Statistics: An Introduction. New York, Oxford University Press 1989.
• [8]Berry DA, Stangl DK: Bayesian Biostatistics. New York, Marcel Dekker 1996.
• [9]Box GEP, Tao DR: Bayesian Inference in Statistical Analysis. Reading, MA, Addison-Wesley 1973.
• [10]Fisher RA: Frequency distributions of the values of the correlation coefficient in samples from an indefinitely large population. Biometrika 1915, 10:507-521.
• [11]Mendola P, Vena J, Buck GM: Exposure Characterization, Reproductive and Developmental Health in the New York Angler Cohort Study. Great Lakes Research Review 1995, 1:2.
• [12]Pickett KE, Abrams B, Selvin S: Maternal height, pregnancy weight gain, and birth weight. Am J Hum Biol 2000, 12:682-687.