A <a href="search.pl?p=nonparametric test">nonparametric test</a> used to determine whether a difference between two sets of paired data has <a href="search.pl?p=statistical significance">statistical significance</a>. This test gives more emphasis to larger differences and is almost as powerful as the <I>t</I>-test. Siegel and Castellan (1988) give details on this and other nonparametric tests that can be used to compare forecasts from two methods. <o:p></o:p></P>
 <ul>
   <li><SPAN Plous, 
S. (1993), <I>Siegel, S. &amp; N. J. Castellan, Jr. (1988), 
<I>Nonparametric Statistics for the Behavioral Sciences</I>, 2<SUP>nd</SUP> ed. 
New York: McGraw-Hill.</SPAN></li>
 </ul>