\documentstyle[fleqn,11pt]{article}  % fleqn just indents the displays
\parindent 0in                       % .2in or .3 in or any indent you want
\pagestyle{plain}                    % plain gives numbering of pages
\setlength{\mathindent}{18pt}
\setlength{\oddsidemargin}{0in}      % -.25in makes wider magrins, e.g.
\setlength{\topmargin}{5pt}          % -.5in makes top margin higher
\headheight 0pt
\headsep 0pt
\textheight 9in                      % Change this when changing topmargin
\textwidth 6.5in                     % Change this when changing oddsidemargin
\parskip 8pt
\renewcommand{\baselinestretch}{1.5} % Change this 1.5 or whatever

\begin{document}
\thispagestyle{empty}

\begin{center}
\section*{Mantel-Haenszel Test Statistics   \\
\vspace*{1ex}
for Correlated Binary Data}

by \\

\vspace*{2ex}

Jie Zhang and Dennis D. Boos \\
Department of Statistics, North Carolina State University \\
Raleigh, NC 27695-8203 \\
tel: (919) 515-1918 \\
fax: (919) 515-7591 \\
e-mail: jzhang@stat.ncsu.edu, boos@stat.ncsu.edu \\
\vspace*{2ex}

March, 1996
\end{center}
\newpage
\thispagestyle{empty}


\begin{center}
\section*{Mantel-Haenszel Test Statistics   \\
% \vspace*{1ex}
for Correlated Binary Data}
\section*{Summary}
\end{center}

This paper proposes two new Mantel-Haenszel test statistics for
correlated binary data in $2 \times 2$ tables that are asymptotically valid in both
sparse data (many strata) and large-strata limiting models. Monte Carlo experiments
show that the statistics compare favorably to previously proposed test statistics,
especially for 5-25 small to moderate-sized strata.
Confidence intervals are also obtained and compared to those obtained from
the test of Liang (1985).

\newpage
\setcounter{page}{1}
  \begin{center}
\section{Introduction}
  \end{center}

Multicenter randomized clinical trials are frequently
used to test the efficacy of
new medical regimens. When the response variables are binary,
and there
are only two treatment groups (drug and placebo), the data often
can be  summarized in a series of
$2 \times 2$ tables based on some covariates or strata
such as the clinical centers.
We will illustrate this by the following two examples.

\noindent {\bf Example 1. A psoriasis multicenter clinical trial} \\
The first example is from a clinical trial comparing a new drug
and a placebo for treatment of psoriasis. In each of the sixteen centers,
patients were randomly assigned to one of the two treatments and
were evaluated on a 0-1
scale (0 = no improvement, 1 = improvement) at four following
visits. The summarized data are in Table A1 in the Appendix.
The full data for 11 of the 277 patients (from Center
2) are in Table A2.

\end{document}