Fit model gives a point as Result, why and how to interpret?

[#1]

Hello, 

I did a fitting model and if you see in the Fresh/Frozen output I don´t get a P-value or log worth, do you know why can this be happening?. If I run the fit model without the interaction effect it works, but if I put the interaction effect it doesn´t work. Do you have a hint or explanation on why this might be happening?

[#2]

Hi @dimacapi07 ,

  If you look at your Effect Tests output, you'll see that for your Fresh/Frozen factor, there are zero (0) degrees of freedom. Since there are no degrees of freedom, it's essentially a constrained fit and an error of the estimate cannot be calculated, and therefore no F-value or LogWorth can be calculated.

  It's analogous to fitting a line between two points -- the line is perfectly fit, and the two parameters that make up the line, slope and y-intercept, cannot vary because there are only those two points and they define the line perfectly, so there is no "wiggle room" for the fit and you won't be able to calculate any errors or F-values, etc. It's undefined.

DS

[#3]

In addition to SDF1's explanation, if you look under the response Bag 2 CC, you see Singularity Details.  This means you have confounded terms in the model.  You don't have enough DF's to fit that model.

[#4]

Hi Statman,

Thank you very much for the explanation. I will check this then in more detail as I have a similar dataset with no issue at  all and I have the same parameters. But indeed without degrees of freedom it is not possible to compute.