The plot with the fitted line shows the response and the predictor data. The plot includes the orthogonal regression line, which represents the orthogonal regression equation.
You can also choose to display the least squares fitted line on the plot for comparison. Larger differences between the two lines show how much the results depend on whether you account for uncertainty in the values of the predictor variable. The least squares values equal the predicted values for the orthogonal regression, so that you can also use the least squares line to examine the predicted values.
Use the plot and the fitted line to assess whether the orthogonal regression equation is a good fit for the data. When the model fits the data, the points fall closely to the regression line. In particular, you can examine the fitted line plot for these criteria:
- The sample contains an adequate number of observations throughout the entire range of all predictor values.
- The sample contains no curvature that the model does not fit.
- The sample contains no outliers, which can have a strong effect on the results. Try to identify the cause of any outliers. Correct any data entry or discernible measurement errors. Consider removing data values that are associated with abnormal, one-time events (special causes). Then, repeat the analysis.
You often use orthogonal regression in clinical chemistry or a laboratory to determine whether two instruments or methods provide comparable measurements.
This plot shows an example of measurements from two instruments or methods that are comparable. The points follow the fitted line with minimal scatter and without any pattern that reveals systematic differences between the methods.
In the results below, the confidence intervals for the coefficients do not provide evidence that the measurements of the two instruments differ. However, the plot shows that points do not fall close to the line, which indicates that the measurements from the two instruments are not comparable. Because the data do not fit the equation, the usual conclusion is that the instruments differ.
Orthogonal Regression Analysis: Current versus New
Predictor Coef SE Coef Z P Approx 95% CI
Constant -0.00000 0.215424 -0.0000 1.000 (-0.422224, 0.42222)
New 1.00000 0.517586 1.9320 0.053 (-0.014450, 2.01445)