23
Step 6: Evaluate Linearity
Objective:
Determine the difference between the measured value and the
reference value when using one measurement instrument over the
expected characteristic process window.
Process:
Select three to five sample standards which encompass the entire
expected measurement range.
If sample standards are not available, then use production units from
the expected measurement range. Have the samples measured 15
times by the calibration laboratory and compute the average of the
15 readings. Use this average as the “reference value”.
Measure the sample standard 15 to 25 times in as short a period of
time as possible using the same operator, same equipment, and
same set-up to obtain the data.
Compute the average of the readings.
Compute the bias for each of the samples. See Step 5 for bias
equation.
Make an x-y graph of reference value (as the abscissa, x-axis) vs.
bias (as the ordinate, y-axis).
On the graph, overlay a “best fit” linear regression line.
Compute the slope of the linear regression line which compares the
bias values to the reference values. Use the equation on the
following page:
