Hi, all

Just new here.

I had a question about the linearity study.

After finishing the numerical calculation, I found the R-sq value is almost zero, which means it was not a good linear model. I know the higher the R-sq, the better the linear model. But my question is, if the gage is perfect, we should have a straight line with slope=0, at this time the R-sq is also zero, from one extreme point to the other.

My study shows everything is good except the R-sq. Zero bias line is lying between confidence intervals, t statistics is less than the t value. But R-sq is almost zero. Around 0.2%

Any suggestion on it?

Thanks.

This is as it should be. If a gage is perfectly linear with a slope = 0 and y-intercept = 0, all points on the line should be random variation around the master values. The model is y = 100% noise variation, so R^2 must be zero. None of the variation is explained by the model.