Advanced contrast options

Discuss the jamovi platform, possible improvements, etc.
Post Reply
TPalma
Posts: 24
Joined: Fri Jul 14, 2017 9:04 am

Advanced contrast options

Post by TPalma »

Hi guys,

First of all, congratulations for the amazing work!
I just wanted to ask whether do you have plans for improving the contrast options for ANOVA's? By "advanced" I don't mean the options we can find in SPSS (at least in previous versions) or JASP. I would like to see something similar to what we can find in STATISTICA, namely, to have the option to run all sorts of comparisons between all cells in any design (between, within, and mixed). If you had this option, I would probably just start using Jamovi.

Keep up with the good work.

Best,
Tomás
User avatar
Ravi
Posts: 194
Joined: Sat Jan 28, 2017 11:18 am

Re: Advanced contrast options

Post by Ravi »

Hi Tomás,
Could you describe or send a screenshot of the contrast functionality in statistica (I don't have a copy)?

Cheers,
Ravi
TPalma
Posts: 24
Joined: Fri Jul 14, 2017 9:04 am

Re: Advanced contrast options

Post by TPalma »

Hi Ravi,

I did it via email.

Thanks!
Best,
Tomás
pandelis
Posts: 5
Joined: Thu Sep 28, 2017 7:16 am

Re: Advanced contrast options

Post by pandelis »

Hi,

I confirm that this feature (advanced contrast options) would be very useful !

For instance I teach statistics with a focus on ANOVA to 3rd year students and that's the only reason why I cannot use, for the moment, Jamovi...

Best regards
User avatar
reason180
Posts: 268
Joined: Mon Jul 24, 2017 4:56 pm

Re: Advanced contrast options

Post by reason180 »

FYI: I notice that jamovi can already do post-hoc comparison of each cell mean to every other cell mean in a within-subject, between-subject, or mixed design. Does the current request for advanced contrast options go beyond that?
User avatar
reason180
Posts: 268
Joined: Mon Jul 24, 2017 4:56 pm

Re: Advanced contrast options

Post by reason180 »

On a related note: I'm not aware of how a Tukey's, single error term, could be employed for a within-subject factorial, or for a mixed factorial ( I'm pretty sure that why SPSS won't do Tukey post-hoc tests in such circumstances). How do we find out what formulas jamovi uses?
pandelis
Posts: 5
Joined: Thu Sep 28, 2017 7:16 am

Re: Advanced contrast options

Post by pandelis »

FYI: I notice that jamovi can already do post-hoc comparison of each cell mean to every other cell mean in a within-subject, between-subject, or mixed design. Does the current request for advanced contrast options go beyond that?
Yes it goes beyond.

For instance if you have 2 variables such as Sex2 and Looks3 and you want for Males only to see if there is a difference betwween Look1 and Look2 combined vs Look3

Here are screenshots from Statistica with a contrast in an analysis with 3 variables
2017-10-01 11_53_39-Specify Contrasts for this Factor_ Spreadsheet1.png
2017-10-01 11_53_39-Specify Contrasts for this Factor_ Spreadsheet1.png (18.03 KiB) Viewed 8606 times
2017-10-01 11_54_00-Enter Contrasts for this Factor     _ Spreadsheet1.png
2017-10-01 11_54_00-Enter Contrasts for this Factor _ Spreadsheet1.png (26.18 KiB) Viewed 8606 times
2017-10-01 11_54_17-Enter Contrasts for this Factor     _ Spreadsheet1.png
2017-10-01 11_54_17-Enter Contrasts for this Factor _ Spreadsheet1.png (25.06 KiB) Viewed 8606 times
User avatar
Ravi
Posts: 194
Joined: Sat Jan 28, 2017 11:18 am

Re: Advanced contrast options

Post by Ravi »

Just an update on this functionality. At the moment we are working on data recoding facilities which technically lets you do all these contrasts. Of course, it would be nicer to have this functionality directly in the analysis but I'm still figuring out the best way to do this UI-wise.

Maybe make a github issue so I won't forget about it: https://github.com/jamovi/jamovi/issues
pandelis
Posts: 5
Joined: Thu Sep 28, 2017 7:16 am

Re: Advanced contrast options

Post by pandelis »

Thank you for your answer.
The github issue is done !

Best
Post Reply