MANCOVA guidance

Discuss statistics related things

by JUL » Sun Mar 03, 2019 1:55 am

Hello,

I am new to the platform, and also to any sort of programming language (though I have figured out syntax before in SPSS). I am running into limitations in SPSS for a particular model and so have been trying to work around by learning R and just moved to trying jamovi. All this to say, I'm very much a novice with this!

I have a dataset with both multiple dependent and independent variables (all continuous). I also have sex as a variable which may be having an effect. When I run it in SPSS, I cannot get a generalized linear model (or anything I can put multiple dependents into) to exclude list-wise rather than case-wise, and because most cases have some data missing, I end up with virtually no cases being included. So I believe I need MANCOVA to deal with this situation, but I can't figure out how to do it in jamovi. I saw the MANCOVA and jmv packages in R which led me here, but I am just so unfamiliar with programming that I'm still rather stumped. I can't see a MANCOVA test explicitly in the jamovi interface so think here I'll need code too? I saw the following in another question forum that you had posted, but don't know how to get from there to my own data:

# Data
npk2 <- within(npk, foo <- rnorm(24))

# Fit model using only manova function
result1 <- manova(cbind(yield, foo) ~ block + N*P*K, npk2)

# Fit model using lm + manova
model <- lm(cbind(yield, foo) ~ block + N*P*K, npk2)
result2 <- manova(model)

# Extract MANOVA test results
summary(result1)

# Extract univariate ANOVA results
summary.aov(result1)

Any thoughts / guidance would be very much appreciated.
JUL
 
Posts: 3
Joined: Sun Mar 03, 2019 1:44 am

by jonathon » Sun Mar 03, 2019 2:51 am

this won't do it for you?

Screen Shot 2019-03-03 at 13.49.44.png
Screen Shot 2019-03-03 at 13.49.44.png (11.67 KiB) Viewed 669 times


jonathon
User avatar
jonathon
 
Posts: 894
Joined: Fri Jan 27, 2017 10:04 am

by JUL » Sun Mar 03, 2019 3:35 am

Oh dear - how did I miss that earlier!? Too much staring at this screen.

Do you know how it excludes behind this? Because the cases seem to decrease across the entire test as I add variables in, rather than staying the same and being independent for each variable (if that makes sense). So it looks like it might be excluding case-wise across the whole model?
JUL
 
Posts: 3
Joined: Sun Mar 03, 2019 1:44 am

by jonathon » Sun Mar 03, 2019 3:38 am

i think that's normal behaviour for a MANCOVA (or any of 'normal' ANOVA family), they can't deal with missing values. in order to deal with missing values, i think you need to turn to mixed effects models. GAMLj might be able to help you out here, but i'm not sure.

jonathon
User avatar
jonathon
 
Posts: 894
Joined: Fri Jan 27, 2017 10:04 am

by JUL » Sun Mar 03, 2019 3:46 am

Yes, that is what I was finding in SPSS, but when I found something online that suggested you could tweak it in R. I'm just not at all near that stage with programming and hoped this might do it. GAMLj only has the one dependent variable option, but that may be the way I'll need to go. Thank you so much for your help! I am really liking jamovi so far - even if it won't quite be the smoking gun for this problem (so to speak).
JUL
 
Posts: 3
Joined: Sun Mar 03, 2019 1:44 am


Return to Statistics