안녕하세요. 캡션에 대해서 질문드리겠습니다.

아래 첨부한 pdf 파일(pdf1.pdf)에서처럼, caption을 표의 제목으로 쓰고 있습니다. 표에 대한 긴 설명은 표의 아래 따로 넣고 있구요..

그런데, 표의 제목이 길어지면 두번째 첨부한 pdf 파일(pdf2.pdf)에서 처럼, "TABLE 6"와 캡션 이 모두 왼쪽으로 쏠려 버리는데요..

이를 해결할 방법이 있을까요? 표의 제목이 길어지더라도  "TABLE 6" 와 표의 제목을 항상 가운데로 유지시키고 싶습니다.

참고로 아래는 제가 사용하는 package들과, 표를 그리는 코드입니다. 

 =============================================================================== 

\documentclass[12pt, leqno, times, amssymb, letterpaper]{article}

\textwidth=6.5in
\textheight=9.0in
\headheight=0.0in
\headsep=0in
\oddsidemargin=0.0in
\topmargin=0.0in

\usepackage{amsmath,amssymb,amsthm}
\usepackage{graphicx}
\usepackage{natbib}
\usepackage{bm}
\usepackage{epsfig}
\usepackage{rotating}
\usepackage{longtable,ltcaption}
\usepackage{lscape}
\usepackage[tablename=TABLE, figurename=FIGURE]{caption}
\usepackage{setspace}
\usepackage{tabularx}
\usepackage{multirow}

\doublespacing

\begin{document}

\clearpage
\begin{table}[tbp]
\captionsetup{labelsep=newline, textfont=bf}
\caption{Cross-Sectional Regressions Including the Remaining Part of the Betas}
\begin{center}
{\scriptsize
\begin{tabularx}{11.5cm}{l Cc C}
\hline
\vspace{-2mm}\\
&\multicolumn{3}{c}{Unrestricted: $E(R_{i,t})=\bm{\lambda}_{1}^{\prime}\, \bm{\beta}_{i,\bullet}\,+\,\bm{\lambda}_{2}^{\prime}\, \bm{\beta}_{i,\bullet}^{\,\bot}$}\\
\vspace{-3mm}\\
\cline{2-4}
\vspace{-2mm}\\
& Restricted: $E(R_{i,t})=\bm{\lambda}_{2}^{\prime}\, \bm{\beta}_{i,\bullet}^{\,\bot}$&& Restricted: $E(R_{i,t})=\bm{\lambda}_{1}^{\prime}\, \bm{\beta}_{i,\bullet}$\\
\vspace{-3mm}\\
\cline{2-2} \cline{4-4}
\vspace{-1.5mm}\\
&\multicolumn{3}{c}{$\bullet=d_1$}\\
\emph{F}-statistic&  3.228&&  6.320 \\
\emph{p}-value&  0.046&&  0.004 \\
Bootstrap-\emph{p}&  0.080&&  0.013 \\
\vspace{-2mm}\\
&\multicolumn{3}{c}{$\bullet=d_2$}\\
\emph{F}-statistic&  1.941&&  2.439 \\
\emph{p}-value&  0.157&&  0.096 \\
Bootstrap-\emph{p}&  0.239&&  0.092 \\
\vspace{-2mm}\\
&\multicolumn{3}{c}{$\bullet=d_3$}\\
\emph{F}-statistic&  2.904&&  0.699 \\
\emph{p}-value&  0.062&&  0.564 \\
Bootstrap-\emph{p}&  0.039&&  0.538 \\
\vspace{-2mm}\\
&\multicolumn{3}{c}{$\bullet=d_4$}\\
\emph{F}-statistic&  0.411&&  0.233 \\
\emph{p}-value&  0.747&&  0.872 \\
Bootstrap-\emph{p}&  0.815&&  0.911 \\
\vspace{-2mm}\\
&\multicolumn{3}{c}{$\bullet=d_5$}\\
\emph{F}-statistic&  1.100&&  1.377 \\
\emph{p}-value&  0.373&&  0.280 \\
Bootstrap-\emph{p}&  0.477&&  0.426 \\
\vspace{-2mm}\\
&\multicolumn{3}{c}{$\bullet=d_6$}\\
\emph{F}-statistic&  6.851&&  8.927 \\
\emph{p}-value&  0.003&&  0.001 \\
Bootstrap-\emph{p}&  0.018&&  0.068 \\
\vspace{-2mm}\\
&\multicolumn{3}{c}{$\bullet=s_6$}\\
\emph{F}-statistic&  5.663&&  7.158 \\
\emph{p}-value&  0.006&&  0.002 \\
Bootstrap-\emph{p}&  0.081&&  0.306 \\
\vspace{-2mm}\\
\hline
\end{tabularx}
}
\end{center}
\label{tab6}
{\footnotesize Table \ref{tab6} implements (\ref{return-beta_decomp_exp}) in the context of the Fama-French three-factor model using the excess returns on 25 portfolios sorted by size and book-to-market. The first column reports $F$-tests for the null hypothesis that $\bm{\lambda}_1$ is the zero vector. The second column reports $F$-tests for the null hypothesis that $\bm{\lambda}_2$ is the zero vector. The first set of $p$-values is based on the $F$ distribution. The second set, indicated by Boostrap-$p$, is based on empirical distributions obtained from a bootstrap procedure designed to incorporate time-series dependence as well as estimation errors in details or smooths and the factor loadings. The data are monthly and cover the period from July 1963 to June 2008.}
\end{table}

\end{document}

 ===========================================================================