Design of Experiment Analysis (CRD, RCBD, Latin Squares,etc)

[divineguthix]

But there's literally an anova linear model specifically for RCBD. That is why I am calling it RCBD analysis because that's essentially built on that. And I know that is just interaction term dissolved from factorial design.

The problem with anova in base jamovi is that I cannot put a blocking factor. If I did a normal anova, it will read the additional factor as indeed another factor and will compute for interaction effects when I wanna see the RCBD model and p values for it.

[reason180]

(1) Where? [Re: "But there's literally an anova linear model specifically for RCBD"]

(2) In principle, a 'blocking factor' can indeed interact with another factor. Thus, while there might be a rationale--in a given situation--for excluding from the model any interaction term that includes the blocking factor, such a rationale won't always exist, and so the interaction term(s) shouldn't *necessarily* be excluded.

(3) In jamovi, you can use the GAMLj module to construct a general linear model that allows you to remove any main-effect term or any interaction term you want.

[divineguthix]

(1) Look at this:

(2) Then it is not a blocking factor but it is a another factor. That would make my design factorial instead of RCBD. Even textbooks know this distinction.

(3) The problem with GAMLj is that it doesn't count the proper df in my factors. It gives different test statistic and p value as compared to me using JMP instead (which does this analysis perfectly.) I would use JMP if its free but it isn't, so I am tryna find a work around that is free and not programming based like R.

[reason180]

If you think there are errors in jamovi's general linear modeling--gamlj in particular--you might consider reporting it as a bug, or at least posing a question in this forum about how to use it.

[divineguthix]

I will do. That's my next step.

[RHainez]

Hello,

Any news on this subject?

Have a nice day