Assessment of linearity is an important part of the verification that
an instrument or test method is in working order. Analyte levels shoul
d be chosen throughout the range of clinical interest, and the materia
l should be as close as possible to that used clinically. After data a
re plotted visual inspection often is all that is needed. Should mathe
matical analysis be required the laboratorian will check the data for
outliers and equality of variance before assessing how well data fit,
that is, how closely the data points approximate a straight line (also
known as goodness of fit). Several means of assessing goodness of fit
are described. This is the third article in a four-part continuing ed
ucation update series on statistics. Other articles discuss the use of
control charts, the use of linear regression to compare methods, and
the creation and verification of reference intervals.