Back to PLS HelpPosted on 09/15/14 01:17:53
Number of posts: 26
Dear experts,
What is the best way to display the brain score graph from PLS? Can I download the raw data and put into excel or spss etc?
Regards
Andrew
Replies:
Posted on 09/15/14 07:34:51
Number of posts: 394
Sure. If you are using the latest version, in the matlab workspace with your results, the field results.usc has the scores for all subjects and conditons. You export that to your favorite program for plotting
Posted on 09/16/14 20:46:45
Number of posts: 26
Thanks for the advice,
I have tried doing this and whilst the graph looks simialr the CIs are not the same as those given within the program. Is this due to the bootstrapping? Can I recreate this in excel or SPSS?
Regards
Andrew
Posted on 09/17/14 14:29:05
Number of posts: 394
Thanks for the advice,
I have tried doing this and whilst the graph looks simialr the CIs are not the same as those given within the program. Is this due to the bootstrapping? Can I recreate this in excel or SPSS?
Regards
Andrew
The CI's are from bootstrapping so would be hard to recreate in SPSS. However, the upper and lower bounds of the CIs are stored in the results array, so you can extract them for plotting
I'm Online
Posted on 09/17/14 14:53:05
Number of posts: 229
The CI's are from bootstrapping so would be hard to recreate in SPSS. However, the upper and lower bounds of the CIs are stored in the results array, so you can extract them for plotting
..and to note - the CIs are not typically symmetric about the mean, so you cannot (easily) use Excel to make plots - not sure whether SPSS can import/plot asymmetric CIs or not
nancy
Posted on 09/17/14 15:52:18
Number of posts: 26
If I want to say two conditions differ, do I just look at CIs to see whether they overlap or do I need to do further analysis in SPSS etc?
Posted on 09/23/14 00:32:17
Number of posts: 26
Any tips on exporting the graph. Do people just screen print or save graph and use in manuscript?
Regards
Andrew
I'm Online
Posted on 09/23/14 08:12:05
Number of posts: 229
You have multiple options to export the figure from the file menu.
If you want to modify the figure (fonts/colours, etc), you can temporarily turn the default menu/toolbars back on by using the following commands:
set(gcf,'toolbar','figure')
set(gcf,'menubar','figure')
nancy
Posted on 09/23/14 11:47:25
Number of posts: 394
Any tips on exporting the graph. Do people just screen print or save graph and use in manuscript?
Regards
Andrew
most journals want high-res images so you probably want to save as EPS
Posted on 05/13/16 03:58:47
Number of posts: 26
The CI's are from bootstrapping so would be hard to recreate in SPSS. However, the upper and lower bounds of the CIs are stored in the results array, so you can extract them for plotting
result.boot_result.ulcorr_adj, result.boot_result.llcorr_adj
Are these the bootstrapped adjusted upper and lower limits of the CI's?
Posted on 05/13/16 07:11:01
Number of posts: 394
result.boot_result.ulcorr_adj, result.boot_result.llcorr_adj
Are these the bootstrapped adjusted upper and lower limits of the CI's?
yes, but don't use them for plotting unless your bootstrap distribution is severly skewed. The adjusted CI's Gaussianize the distribution and centre the point estimate. If the task mean and CI plot looks ok in the GUI use the result,boot_result.ulcorr and llcorr
I'm Online
Posted on 05/13/16 07:12:07
Number of posts: 229
result.boot_result.ulcorr_adj, result.boot_result.llcorr_adj
are ulcorr and llcorr adjusted for skewness (if present) in the bootstrap results - normally you should not need to report/use these - they are primarily there for trouble-shootin
nancy
Login to reply to this topic.
