It is important to define at this point what we
mean by errors, especially as the term is unfortunately
used to denote two different concepts that are relevant
to a regression model. In a regression model, errors are
the difference between subjects' observed values on the
response variable and the values predicted by the true
regression model for the population as a whole. This
usage of the term error needs to be distinguished from
the concept of measurement error, which will be defined
and discussed later in this article.