Missing data is a common occurrence in most medical research data collection enterprises. There is an extensive literature concerning missing data, much of which has focused on missing outcomes. Covariates in regression models are often missing, particularly if information is being collected from multiple sources. The method of weights is an implementation of the EM algorithm for general maximum-likelihood analysis of regression models, including generalized linear models (GLMs) with incomplete covariates. In this paper, we will describe the method of weights in detail, illustrate its application with several examples, discuss its advantages and limitations, and review extensions and applications of the method.
Get full access to this article
View all access options for this article.
References
1.
Little RJA.Regression with missing X's: a review . Journal of the American Statistical Association1992; 87: 1227-1237 .
2.
Jones MP.Indicator and stratification methods for missing explanatory variables in multiple linear regression . Journal of the American Statistical Association1996; 91: 222-230 .
3.
Robins JM , Rotnitzky A , Zhao LP.Estimation of regression coefficients when some regressors are not always observed . Journal of the American Statistical Association1994; 89: 846-866 .
4.
Little RJA , Schluchter MDMaximum likelihood estimation with mixed continuous and categorical data with missing values . Biometrika1985; 72: 497-512 .
5.
Ibrahim JG.Incomplete data in generalized linear models . Journal of the American Statistical Association1990; 85: 765-769 .
6.
Fienberg SE.The analysis of cross-classified categorical data. Boston, MA: Massachusetts Institute of Technology , 1980.
7.
McCullagh P , Nelder JA.Generalized linear models. London: Chapman & Hall , 1989.
8.
Dempster AP , Laird NM , Rubin DB.Maximum likelihood from incomplete data via the EM algorithm . Journal of the Royal Statistical Society, Series B1977; 39: 1-22 .
9.
Wood A , Hinde J.Binomial variance component models with a non-parametric assumption concerning random effects . In: Longitudinal data analysis: Surrey conference on sociological theory and method, Vol 4. Avebury: Aldershot, 1987: 110-128 .
10.
Louis TA.Finding the observed information matrix when using the EM algorithm . Journal of the Royal Statistical Society, Series B1982; 44: 226-233 .
11.
Schluchter MD , Jackson KL.Log-linear analysis of censored survival data with partially observed covariates . Journal of the American Statistical Association1989; 84: 42-52 .
12.
Laird NM , Olivier D.Covariance analysis of censored survival data using log-linear analysis techniques . Journal of the American Statistical Association1981; 76: 231-240 .
13.
Lipsitz SR , Ibrahim JG.Using the EM-algorithm for survival data with incomplete categorical covariates . Lifetime Data Analysis1996; 2: 5-14 .
14.
Lipsitz SR , Ibrahim JG.Estimating equations with incomplete categorical co-variates in the Cox model . Biometrics1998; 54: 1002-1013 .
15.
Zahner GEP , Daskalakis C.Factors associated with mental health, general health and school-based service use for psychopathology . American Journal of Public Health1997; 87: 1440-1448 .
16.
Zahner GEP , Jacobs JH , Freeman DH , Trainor K.Rural-urban child psychopathology in a northeastern U.S. state: 1986-1989 . Journal of the American Academy of Child Adolescent Psychiatry1993; 32: 378-387 .
17.
Zahner GEP , Pawelkiewicz W , DeFrancesco JJ , Adnopoz J.Children's mental health service needs and utilization patterns in an urban community . Journal of the American Academy of Child Adolescent Psychiatry1992; 31: 951-960 .
18.
Achenbach TM.Manual for the teacher's report form and 1991 profile. University of Vermont: Department of Psychiatry , 1991.
19.
Boyle MH , Offord DR , Racine YA , Fleming JE , Szatmari P , Links PS.Predicting substance use in early adolescence based on parent and teacher assessments of childhood psychiatric disorder: results from the Ontario child health study follow-up . Journal of Child Psychology and Psychiatry1993; 34: 535-544 .
20.
Lipsitz SR , Ibrahim JG.A conditional model for incomplete covariates in parametric regression models . Biometrika1996; 83: 916-922 .
21.
Ibrahim JG , Weisberg S.Incomplete data in generalized linear models with continuous covariates . The Australian Journal of Statistics1992; 34: 461-470 .
22.
Ibrahim JG , Chen M-H , Lipsitz SR.Monte Carlo EM for missing covariates in parametric regression models . Biometrics, 1999, in press.
23.
Cox DR , Wermuth N.Response models for mixed binary and quantitative variables . Biometrika1992; 79: 441-461 .
24.
Fitzmaurice GM , Laird NM.Regression models for a bivariate discrete and continuous outcome with clustering . Journal of the American Statistical Association1995; 90: 845-852 .
25.
Little RJA , Rubin DB.Statistical analysis with missing data. New York: John Wiley , 1987.
26.
Leong T , Lipsitz SR , Ibrahim JG.Using missing data methods in genetic studies with missing mutation status . Statistics in Medicine1999; 18: 473-485 .
27.
Vach W.Some issues in estimating the effect of prognostic factors from incomplete covariate data . Statistics in Medicine1997; 16: 57-72 .
28.
Vach W , Blettner M. . Logistic regression with incompletely observed categorical covariates - investigating the sensitivity against violation of the missing at random assumption . Statistics in Medicine1995; 14: 1315-1329 .
29.
Vach W , Illi S.Biased estimation of adjusted odds ratios from incomplete covariate data due to violation of the missing at random assumption . Biometrical Journal1997; 39: 13-28 .
30.
Ibrahim JG , Lipsitz SR , Chen M-H.Missing covariates in generalized linear models when the missing data mechanism is nonignorable . Journal of the Royal Statistical Society, Series B1999; 61: 173-190 .
31.
Lipsitz SR , Ibrahim JG , Chen M-H , Peterson H.Non-ignorable missing covariates in generalized linear models . Statistics in Medicine, 1999, in press.
32.
Baker SG , Laird NM.Regression analysis for categorical variables with outcome subject to nonignorable nonresponse . Journal of the American Statistical Association1988; 83: 62-69 .
33.
Bashir SA , Duffy SW.The correction of risk estimates for measurement error . Annals of Epidemiology1997; 7: 154-164 .
34.
Carroll RJ , Ruppert D , Stefanski LA.Measurement error in nonlinear models. London: Chapman & Hall , 1995.
35.
Formann AK , Kohlmann T.Latent class analysis in medical research . Statistical Methods in Medical Research1996; 5: 179-211 .
36.
Kaldor J , Clayton D.Latent class analysis in chronic disease epidemiology . Statistics in Medicine1985; 4: 327-335 .
37.
Kosinski AS , Flanders WD.Regression model for estimation of odds ratios with misclassified exposure. Technical Report 96-5, Department of Biostatistics, Rollins School of Public Health, Emory University , 1996.
38.
Brenner H.Use and limitations of dual measurements in correcting for nondifferential exposure misclassification . Epidemiology1992; 3: 216-222 .
39.
Mantel H , Singh A , Kinack M , Rowe G.Statistical matching: use of auxiliary information to avoid the conditional independence assumption . In: Proceedings of the Bureau of the Census Annual Research Conference, 1991: 688-711 .
40.
Singh AC , Mantel HJ , Kinack MD , Rowe G.Statistical matching: use of auxiliary information as an alternative to the conditional independence assumption . Survey Methodology1993; 19: 59-79 .
41.
Efron B , Tibshirani RJ.An introduction to the bootstrap. London: Chapman & Hall , 1993.
