We consider partially observed binary matched-pair data. We assume that the incomplete subjects are missing at random. Within this missing framework, we propose an EM-algorithm based approach to construct an interval estimator of the proportion difference incorporating all the subjects. In conjunction with our proposed method, we also present two improvements to the interval estimator through some correction factors. The performances of the three competing methods are then evaluated through extensive simulation. Recommendation for the method is given based on the ability to preserve type-I error for various sample sizes. Finally, the methods are illustrated in two real-world data sets. An R-function is developed to implement the three proposed methods.
OsobaDErlichmanCWillanAR, et al. Superiority of methylprednisolone sodium succinate over low dose metoclopramide hydrochloride in the prevention of nausea and vomiting produced by cancer chemotherapy. Clin Invest Med1986; 9: 225–231.
2.
McNemarQ. Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika1947; 12: 153–157.
3.
MostellerF. Some statistical problems in measuring the subjective response to drugs. Biometrics1952; 8: 220–226.
4.
RubinDB. Inference and missing data. Biometrika1976; 63: 581–592.
5.
ChoiSStableinD. Practical tests for comparing two proportions with incomplete data. Journal of the Royal Statistical Society: Series C (Applied Statistics)1982; 31: 256–262.
6.
ChoiSStableinD. Comparing incomplete paired binomial data under non-random mechanisms. Stat Med1988; 7: 929–939.
7.
GokhaleDSirtonikB. On tests for correlated proportions in the presence of incomplete data. Psychometrika1984; 49: 147–152.
8.
FuchsC. Maximum likelihood estimation and model selection in contingency tables with missing data. J Am Stat Assoc1982; 77: 270–278.
9.
TangM-LLingM-HTianG-L. Exact and approximate unconditional confidence intervals for proportion difference in the presence of incomplete data. Stat Med2009; 28: 625–641.
10.
IbrahimJG. Incomplete data in generalized linear models. J Am Stat Assoc1990; 85: 765–769.
11.
IbrahimJGChenM -HLipsitzSR, et al. Missing-data methods for generalized linear models: A comparative review. J Am Stat Assoc2005; 100: 332–346.
12.
FirthD. Bias reduction of maximum likelihood estimates. Biometrika1993; 80: 27–38.
13.
HeinzeGSchemperM. A solution to the problem of separation in logistic regression. Stat Med2002; 21: 2409–2419.
14.
HeinzeG. A comparative investigation of methods for logistic regression with separated or nearly separated data. Stat Med2006; 25: 4216–4226.
15.
MaitiTPradhanV. A comparative study of the bias corrected estimates in logistic regression. Stat Methods Med Res2008; 17: 621–634.
KosmidisIFirthD. Bias reduction in exponential family nonlinear models. Biometrika2009; 96: 793–804.
18.
KosmidisIFirthD. Multinomial logit bias reduction via the poisson log-linear model. Biometrika2011; 98: 755–759.
19.
KosmidisIPaguiE. C. K.SartoriN. Mean and median bias reduction in generalized linear models. Statistics and Computing2020; 30: 43–59.
20.
KosmidisIFirthD. Jeffreys-prior penalty, finiteness and shrinkage in binomial-response generalized linear models. Biometrika2021; 108: 71–82. https://doi.org/10.1093/biomet/asaa052.
21.
AgrestiACaffoB. Simple and effective confidence intervals for proportions and differences of proportions result from adding two successes and two failures. Am Stat2000; 54: 280–288.
22.
AgrestiAMinY. Simple improved confidence intervals for comparing matched proportions. Stat Med2005; 24: 729–740.
23.
HarelOZhouXH. Multiple imputation for the comparison of two screening tests in two phase Alzheimer studies. Stat Med2007; 26: 2370–2388.
24.
RubinD. B.Multiple imputation for survey nonresponse. New York: John Wiley & Sons, 2004.
25.
EverittBS. The analysis of contingency tables. New York: Chapman and Hall/CRC, 1992.
26.
LouisT. A.Finding the observed information matrix when using the EM algorithm. Journal of the Royal Statistical Society. Series B (Methodological), 1982; 44(2): 226–233.
27.
JeffreysH. An invariant form for the prior probability in estimation problems. In Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, The Royal Society 1946; 186(1007): 453–461.
28.
RuppertD. Statistics and data analysis for financial engineering, volume 13. Springer, 2011.
29.
TannerMA. Tools for statistical inference, volume 3. Springer, 1991.
30.
SimonCPBlumeL. Mathematics for economists, volume 7. New York: Norton, 1994.
31.
WhiteIRRoystonPWoodAM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med2011; 30: 377–399.
32.
R Core Team. A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria,2015.
33.
NewcombeRG. Measures of location for confidence intervals for proportions. Communications in Statistics-Theory and Methods2011; 40: 1743–1767.
34.
MorrisTPWhiteIRCrowtherMJ. Using simulation studies to evaluate statistical methods. Stat Med2019; 38: 2074–2102.
35.
HeinzeGPlonerMBeyeaJ. Confidence intervals after multiple imputation: combining profile likelihood information from logistic regressions. Stat Med2013; 32: 5062–5076.
36.
WangW. An inductive order construction for the difference of two dependent proportions. Stat Probab Lett2012; 82(8): 1623–1628.
37.
FayM. P.LumbardK. Confidence intervals for difference in proportions for matched pairs compatible with exact mcNemar’s or sign tests. Statistics in Medicine2021; 40(5): 1147–1159. https://doi.org/10.1002/sim.8829.
38.
WuCJ. On the convergence properties of the em algorithm. The Annals of statistics, pages 95–103, 1983.
