Re: Prüfung von Residuen auf Normalverteilung (Diagramm Normalquantile)The plots look the same to me except for the red line and the confidence bands. The line from Fit Model looks to be off, but the data points appear the same to me.
I have not seen this behavior before and am not able to reproduce it. Could you possibly share some details of the model that you fit? Do you have any excluded data points? Are there any "errors" or "warnings" as part of the model fit? What is the model form that you used? Are all of the factors continuous?
Any information that you provide might help determine why that line is different.
Re: Checking residuals for normal distribution (normal quantile diagram)Thanks for the answer.
The points themselves seem to be the same, but unfortunately the different red lines and confidence bands confuse me a bit.
The model was created according to an RSM design (45 attempts), adjusted using least squares and not yet further reduced or edited.
So it still contains all the main effects, quadratic effects and 2-factor interactions. A total of 4 continuous and one 5 step factor (discontinuous) were used.
During the adjustment, neither errors nor warnings were displayed to me and I used all 45 values of the target size (no missing or excluded points).
By the way, I have two targets. And the second has the same problem. It makes no difference whether I adjust the target values together, separately or just one.
Checking residuals for normal distribution (normal quantile diagram)Hi,
I have created a test plan using DoE and now want to evaluate it. When fitting the model, I would of course like to check whether the residuals are normally distributed. To do this, I have displayed the "diagram of residual normal quantiles" in the model window.
In addition, I saved the residuals as a column and displayed them as a histogram via "Distributions". There is also the option of creating a "diagram of the normal quantiles".
Now to my question: is there a difference between the "diagram of the residual normal quantiles" and the "diagram of the normal quantiles" when using the stored residuals? I actually thought that both digrams should be identical but the lines and intervals shown in the diagrams look different (see pictures).
Does anyone happen to have an idea why?
Many greetings
AnalyticsDOEGraphicsRe: Prüfung von Residuen auf Normalverteilung (Diagramm Normalquantile)I don't know why the lines look the way they do. I can tell you that it is related to your 5-level categorical factor. I think at this point it would be best to have you contact JMP Technical Support to see if they can determine why the line is drawn the way that it is.
