Jamovi Query

Discuss the jamovi platform, possible improvements, etc.

by 776655104 » Thu Oct 08, 2020 1:58 am

We are interested in using Jamovi at my University but we had some areas of interest to ask about.

Will there eventually be a test of normality added to the repeated measures ANOVA routine?

Will the ANOVA routine eventually be expanded with things such as the Welch test or Games-Howells posthoc (I think Jasp has these already)?

You have posthocs such as Tukey's HSD, but agriculture tends to prefer Fisher's LSD (I think different disciplines may or may not have preferred posthocs depending on the type of data and designs they use). Are there any plans to add the Fisher's protected LSD post hoc test?

JB
776655104
 
Posts: 7
Joined: Thu Oct 08, 2020 1:45 am

by 776655104 » Thu Oct 08, 2020 3:19 am

For the ANOVA I did see the Welch and Games-Howell. My mistake - was looking in the wrong area.
776655104
 
Posts: 7
Joined: Thu Oct 08, 2020 1:45 am

by reason180 » Mon Oct 12, 2020 4:51 pm

Doesn't jamovi already have Fisher's LSD? (The "No Correction" option.)
reason180
 
Posts: 52
Joined: Mon Jul 24, 2017 4:56 pm

by reason180 » Mon Oct 12, 2020 9:42 pm

If you select "Exploration" "Descriptives" from the top menu, you can conduct a normality test for any of the columns in the data set.
reason180
 
Posts: 52
Joined: Mon Jul 24, 2017 4:56 pm

by 776655104 » Tue Oct 13, 2020 2:16 am

Okay regarding the Fisher's LSD as the no correction option (although maybe this should be more apparent?). For the normality tests I was meaning for the assumption checks, thus normality tests of the residuals by time point (as opposed to normality checks of the variables which is what you get from descriptives). Currently for repeated measures ANOVA it has sphericity and homogeneity assumption checks, but no normality/Q-Q plots assumption check options. The only option you have is to do separate ANOVAs for each individual time point and check the residuals there.
776655104
 
Posts: 7
Joined: Thu Oct 08, 2020 1:45 am

by reason180 » Tue Oct 13, 2020 4:16 am

I'll leave it to others to respond further regarding the normality tests, since I don't see the distinction normality tests by time point and normality checks of the variables. "Descriptives" does support generating a Q-Q plot for each "time point" (which is a way to assess the normality assumption for each "time point").
reason180
 
Posts: 52
Joined: Mon Jul 24, 2017 4:56 pm

by 776655104 » Tue Oct 13, 2020 8:45 am

But the descriptives option would be working on and testing the normality the variables them-self. Not the residuals for the repeated measures ANOVA. I am asking specifically about the assumption checks for repeated measures ANOVA procedure. In that particular procedure there is no normality or Q-Q plot option for the residuals.
776655104
 
Posts: 7
Joined: Thu Oct 08, 2020 1:45 am

by reason180 » Wed Oct 14, 2020 9:46 pm

What I'm saying is, given that the normality assumption is that within each of the relevant populations the values are normally distributed, I think "testing the normality the variables themselves" is identical to testing normality of "the residuals for the repeated measures ANOVA."
reason180
 
Posts: 52
Joined: Mon Jul 24, 2017 4:56 pm

by 776655104 » Thu Oct 15, 2020 12:05 am

I don't think that this is the right approach as it does not always pan out, especially for more complex models where you can actually have normal residuals with a non normal dependent variable. It would be better to have the assumption check there, especially since the other linear type models have it.
776655104
 
Posts: 7
Joined: Thu Oct 08, 2020 1:45 am

by jonathon » Thu Oct 15, 2020 5:22 am

we've looked into add Q-Q plots and normality tests to the RM ANOVA, but we've never been able to get a clear answer as to whether these are appropriate.

i'll see about adding it.

jonathon
User avatar
jonathon
 
Posts: 1405
Joined: Fri Jan 27, 2017 10:04 am

Next

Return to General

cron