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

Discuss the jamovi platform, possible improvements, etc.
divineguthix
Posts: 10
Joined: Thu Jan 20, 2022 6:53 am

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

Post by divineguthix »

Hi. Do you know any module where we can analyze RCBD(randomized complete block design) and Latin Square type of experiments? I kinda need it for my students' researches. Thank you so much.

Also if you know fractional factorial designs analysis I would be happy to use that too.
User avatar
jonathon
Posts: 2613
Joined: Fri Jan 27, 2017 10:04 am

Re: Design of Experiment Analysis (CRD, RCBD, Latin Squares,

Post by jonathon »

does the Randomizer module do what you want?

jonathon
divineguthix
Posts: 10
Joined: Thu Jan 20, 2022 6:53 am

Re: Design of Experiment Analysis (CRD, RCBD, Latin Squares,

Post by divineguthix »

No it just gives me the randomized experiment runs, but not the anova (the one I need)
User avatar
reason180
Posts: 274
Joined: Mon Jul 24, 2017 4:56 pm

Re: Design of Experiment Analysis (CRD, RCBD, Latin Squares,

Post by reason180 »

'Randomized complete block design,' ' Latin Square type of experiments,' and 'fractional factorial designs,' have to do with how an experiment is designed. They are not analyses.

Such designs are typically analyzed using routines such as t tests and ANOVAS, which are built into jamovi.
divineguthix
Posts: 10
Joined: Thu Jan 20, 2022 6:53 am

Re: Design of Experiment Analysis (CRD, RCBD, Latin Squares,

Post by divineguthix »

Then how do I analyze these? The way it provides me the proper model and anova I need?
User avatar
reason180
Posts: 274
Joined: Mon Jul 24, 2017 4:56 pm

Re: Design of Experiment Analysis (CRD, RCBD, Latin Squares,

Post by reason180 »

Example: A Latin square was used counterbalance the order in which each of six participants experienced Condition A, Condition B, and Condition C. Each participant produced a response in each of the three conditions. The data are shown below.
Untitled65.png
Untitled65.png (11.8 KiB) Viewed 2853 times
An appropriate analysis to assess the relationship between Condition (A, B, or C) and Response, would be a one-factor repeated-measures ANOVA.
divineguthix
Posts: 10
Joined: Thu Jan 20, 2022 6:53 am

Re: Design of Experiment Analysis (CRD, RCBD, Latin Squares,

Post by divineguthix »

So I will just use one factor? Isn't that erroneous? I want blocking to be considered in my calculations. I know CRD can just do one way anova, but RCBD and Latin Squares can't be done with jamovi's features.
User avatar
reason180
Posts: 274
Joined: Mon Jul 24, 2017 4:56 pm

Re: Design of Experiment Analysis (CRD, RCBD, Latin Squares,

Post by reason180 »

*If* you want to include an assessment of the effect of Condition Order (i.e., an assessment of the effect of the "blocking" variable), then for the present example, an appropriate analysis would be a 3 (Condition: A, B, C) by 6 (Condition Order: ABC, ACB, BAC, BCA, CAB, CBA) repeated-measures analysis of variance, with Condition varying within subjects and Condition Order varying between subjects. See the illustration below.
Untitledz.png
Untitledz.png (21 KiB) Viewed 2781 times
divineguthix
Posts: 10
Joined: Thu Jan 20, 2022 6:53 am

Re: Design of Experiment Analysis (CRD, RCBD, Latin Squares,

Post by divineguthix »

Will the linear model generate be the same as if I will calculate the rcbd in let's say R? Because I don't want them to be very distinct, and drawing wrong conclusions
User avatar
reason180
Posts: 274
Joined: Mon Jul 24, 2017 4:56 pm

Re: Design of Experiment Analysis (CRD, RCBD, Latin Squares,

Post by reason180 »

Again, there's no such thing as an "RCBD analysis."

If you are going to analyze the results of a "randomized complete block" experiment (or if you're going to do what I did here, and employ Latin square counterbalancing rather than randomization) then you're probably going to conduct an analysis of variance, regression, or perhaps a linear mixed effects analysis. See https://stat.ethz.ch/~meier/teaching/an ... signs.html for examples of such ANOVAs (for analyzing the results of a randomized complete block design) in R. And note that jamovi does indeed do analysis of variance (as well as regression and linear mixed-effects analysis).
Post Reply