Why is error contribution higher than factor contribution in ANOVA? I used Taguchi L9 orthogonal array design for my experiment to optimise the tool geometry parameters for better surface roughness. I have conducted experiments as per DOE and I used Minitab to find percentage contribution of each factor on surface roughness. I ended with a 50.21% error contribution. Can anyone please give your valuable suggestions on why I am getting 50.21% error contribution which is higher than factor contribution?
Please take a look in the picture and share your valuable suggestions.
Thanks in Advance

 A: You got a high error contribution because your model does not account for that much of the variance in surface roughness.
There are lots of possible reasons for this, here are some (not an exhaustive list):

*

*Measurement error. If your variables are measured imperfectly, that will be part of the error contribution.


*Missing variables. There could be other things that contribute to surface roughness; in fact, there surely are. Models aren't meant to be perfect representations of reality.


*Nonlinear relationships: Perhaps the relationship between surface roughness and some of your variables isn't linear. In your sample, it surely isn't perfectly linear, but it might not even be close to linear.


*Sampling error. Even if your model is perfect in the population, it will not be perfect in any random sample. It might even be off by a lot.
All of these contribute to error in linear models. How much do they contribute in your case? You have to do a lot of investigating.  Also, 50% error contribution might not be a lot. What is a "lot" here is field dependent.
