If you did it in the lab, you`ll know that it`s very unlikely that the second study will produce the same result as the first. If you perform a number of tests (i.e. identical in all respects), you will probably get scattered results. The fit is significantly improved, especially at the high end of glucose. In practice, it would be enough to round them significantly, between 2.0 and 0.4 × glucose and +1.8 mmol/L. In one case, the average age of human babies, measured by two methods, was compared (caterer in 1979). If there was no significant difference in a t-test, the measurement methods were considered consistent. Dependent and independent variables are measured with errors, but the regression line of the smallest squares ignores the error in x. It evaluates the average value y for a given observed x. .