The theoretical probability distribution of forecast errors. It is often assumed to be normal. In the social sciences, this assumption is generally reasonable for short-interval time-series data (say, monthly or less), but not for annual data.<B> <o:p></o:p></B></P>