A relationship between a predictor variable (<I>X<SUB>1</SUB></I>) and the dependent variable (<I>Y</I>) that depends upon the level of another predictor variable (<I>X<SUB>2</SUB></I>). (There may be main effects as well.) To address problems containing interaction, consider a program such as <a href="search.pl?p=AID">AID</a>. It is difficult to find evidence that interaction terms in <a href="search.pl?p=regression analysis">regression analysis</a> contribute to forecast accuracy.<B><o:p></o:p></B></P>