Walrus Package

[Andrew_Govus]

HI All,

Thanks for porting the Walrus package to Jamovi from R.

I have a few questions/requests that I wondered if someone might know the answer to:

Robust one-way ANOVA with trimmed means does not seem to provide the df in the Robust ANOVA table, nor the explanatory measure of effect size as the WRS2 package does in R. I'm not that worried about the effect size, but the DF probably should be reported with the F statistic. Is there are reason for this that I'm missing?

Example below:

Jamovi syntax

walrus::ranova(
data = data,
dep = HEIGHT,
factors = COHORT,
ph = TRUE)

Yields:

#
# Robust ANOVA
# ─────────────────────────────
# F p
# ─────────────────────────────
# COHORT 5.349 0.0032
# ─────────────────────────────
# Note. Method of trimmed
# means, trim level 0.2
#

The same analysis using the WRS2 package in R gives me:

R output

#Call: t1way(formula = HEIGHT ~ COHORT, data = ON_DATA)
#
#Test statistic: F = 5.5156
#Degrees of freedom 1: 3
#Degrees of freedom 2: 43.39
#p-value: 0.00268
#
#Explanatory measure of effect size: 0.35
#Bootstrap CI: [0.13; 0.69]

The DF 1 and 2, as well as the explanatory measures of effect size are missing from the Jamovi output.

Is this a functional limitation or something that can be implemented for consistency with R. Sorry if I've made an error, happy to be corrected here.

Regards,

Andy Govus

[jonathon]

i'm not 100% on this, because i wrote this a long time ago, but i may have omitted the df, just because none of the other robust ANOVA functions provided dfs.

here's the code:

https://github.com/jamovi/walrus/blob/master/R/ranova.b.R

you'd be welcome to submit a PR.

cheers

jonathon

[MAgojam]

Hi Andy,
can you take a look at this screenshot, to see if there is what you need?

walrus_upd.png (104.97 KiB) Viewed 17780 times

Some time ago I modified the walrus module locally to make it available to some of my students in physiotherapy.
If Jonathon deems it useful I can make a "Pull request".

Cheers,
Maurizio

[Andrew_Govus]

Thank you both for your reply.

Maurizio - yes, that is what I was after. Many thanks if you could please make that a pull request.

Keep us the excellent work Jonathon - many converts to Jamovi at my uni (La Trobe) so far!

Regards,

Andrew Govus

[jonathon]

ah great. always glad to hear of australian uni's running with us.

cheers

[samdbkk]

Hi, I was just wondering how to actually install this updated version of the Walrus module as currently only version 1.0.2 appears to be available.

Thank you.

[jonathon]

1.0.5 is available from cran:

install.packages('walrus')

cheers

[samdbkk]

Hi Jonathan,

Thank you for your very prompt response. As an R and Jamovi novice, I naively thought that having the most updated Walrus package would mean that the degrees of freedom and effect size information would be automatically displayed but this was not the case. I would greatly appreciate it if any guidance could be offered regarding how to have this extra information shown on R or Jamovi (specifically for a robust two-way ANOVA).

Thank you again in advance.

[EdVanSmack]

Hi, I have tried to install version 1.0.5 via the R module. Unfortunately I always get the message "you cannot download packages with the jamovi R". If I install the module walrus directly via the function in jamovi, I get the version 1.0.2. Can you help how I get the new version?

I have a question on the side: I am used to always report SD. In the robust descriptiva table, only SE is reported. Can you explain me why?

Thank you in advance
Eddy

[jonathon]

i think you'll find there are no changes between walrus 1.0.2 and 1.0.5 ... the newer versions were only created to placate CRAN.

but i've pushed the 1.0.5 to the library now, and it should pop out in the few hours.

jonathon