Hi -
I'm running a non-rotated structural analysis with 4 groups, 3 conditions. I'm trying to examine 2 contrasts (group and condition)
Here is the contrast file that's created:
(i.e. contrast 1 = condition 1 vs. condition 3; contrast 2 = group 1 & 2 versus group 3 & 4)
But when the analysis is run:
- there is only 1 LV
- The design LV (and in the contrast window), only displays contrast # 2.
- The task pls / brain scores for the single LV that emerges reflects contrast # 1.
So two questions:
- Why would there be only one LV when two contrasts were entered?
- Why is only contrast # 2 reflected in the single design LV; and contrast # 1 in the brain scores?
Any ideas or advice would be most appreciated.
Many thanks -
Gary
Gary... are you entering just one datamat?
n
Hi Nancy -
I'm entering a datamat for each group (N=4). And the coniditons all appear so the whole design is reflected in the contrast builder screen.
Best,
Gary
OK.. so the contrast file makes sense.. but this doesn't:
- there is only 1 LV
- The design LV (and in the contrast window), only displays contrast # 2.
- The task pls / brain scores for the single LV that emerges reflects contrast # 1.
if you load your contrast text file directly into matlab, do you have two columns?
i made a fake contrast set and it works in the older version of the gui for 2 contrasts:, 2 conditions x 2 groups gives me
1 1
0 1
0 0
-1 0
for the contrast, and 2 LVs, each reflecting the proper contrast.
- btw, which version are you using (we recently upgraded)..
Hi again -
So I've loaded the contrast file in matlab and it gives the two columns as shown above. I'm still using the older gui - i.e. the one before the most recent release in the last few weeks. I'm perplexed.
I have the images from the results file but can't seem to post them in here though. I can email them if that would help.
Thanks,
Gary
Hi again -
Problem solved! I was running the non-rotated analysis by altering a batch file I had saved. Once I ran the analysis again from scratch, it all worked. Two LVs as expected.
Seems like there was something about altering the btach file.
Thanks,
Gary
First, you should use the latest version of PLS applications (published on March 1, 2013). The old version will no longer be supported, since all the new features will have to be built on the latest version rather than the old version. In other words, if you keep using the old version you will no longer get new features that I will publish in the future. Please always keep an eye on what's new in PLS applications under:
http://research.baycrest.org/pls/whatsnew.txt
Second, you should get the same result no matter you use the latest version or use any old version. I believe that you made some mistakes when you create batch file manually. As a matter of fact, it could have been much easier and error free if you let PLS applications to create batch file for you. Here's how: In main PLS window, click Run PLS Analysis on Structural data, the analysis window will pop up. In analysis window, please fill out all necessary fields, and then click "Save to Batch" instead of "Run". Using batch file to run PLS analysis has many benefits. You can load the batch file again to the GUI, you can modify the batch file by yourself, you can also keep all the information that you entered and save it in the batch file, so the parameter won't get lost. All the details have been described clearly under:
http://research.baycrest.org/pls/UserGuide.htm
Finally, I just want to let you know that I have tested the following case with both old version and latest version, GUI and batch, they are all the same, with the correct result (i.e. 2 LV, etc.). I have 4 groups, 2 conditions (I don't have 3 conditions), and 2 contrasts. There is 1 subject for each group.
Here's the batch file automatically created from old version:
%------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Result File Name Start %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Note: Result file must be listed first, and must follow the file
% name format of xxxx_yyyyresult.mat, where xxxx stands for
% "any result file name prefix" and yyyy stands for the name
% of PLS module (either PET ERP fMRI, BfMRI, or STRUCT). File
% name is case sensitive on Unix or Linux computers.
result_file g1_STRUCTresult.mat
%%%%%%%%%%%%%%%%%%%%%%%%%%
% Result File Name End %
%%%%%%%%%%%%%%%%%%%%%%%%%%
%------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%
% Group Section Start %
%%%%%%%%%%%%%%%%%%%%%%%%%
group_files g1_STRUCTdatamat.mat
group_files g2_STRUCTdatamat.mat
group_files g3_STRUCTdatamat.mat
group_files g4_STRUCTdatamat.mat
% ... following above pattern for more groups
%%%%%%%%%%%%%%%%%%%%%%%
% Group Section End %
%%%%%%%%%%%%%%%%%%%%%%%
%------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%
% PLS Section Start %
%%%%%%%%%%%%%%%%%%%%%%%
% Notes:
% 1. Mean-Centering PLS
% 2. Non-Rotated Task PLS (please also fill out contrast data below)
% 3. Regular Behav PLS (please also fill out behavior data & name below)
% 4. Multiblock PLS (please also fill out behavior data & name below)
% 5. Non-Rotated Behav PLS (please also fill out contrast data and
% behavior data & name below)
pls 2 % PLS Option (between 1 to 5, see above notes)
num_perm 0 % Number of Permutation
num_boot 0 % Number of Bootstrap
clim 95 % Confidence Level for Behavior PLS
save_data 0 % Set to 1 to save stacked datamat
intel_system 0 % Set to 0 for non-intel system
%%%%%%%%%%%%%%%%%%%%%
% PLS Section End %
%%%%%%%%%%%%%%%%%%%%%
%------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Condition Selection Start %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
selected_cond 1 1
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Condition Selection End %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%
% Contrast Data Start %
%%%%%%%%%%%%%%%%%%%%%%%%%
% Notes: only list selected conditions (selected_cond)
contrast_data 1.000000000000000 1.000000000000000
contrast_data -1.000000000000000 1.000000000000000
contrast_data -1.000000000000000 1.000000000000000
contrast_data 1.000000000000000 1.000000000000000
contrast_data 1.000000000000000 -1.000000000000000
contrast_data -1.000000000000000 -1.000000000000000
contrast_data -1.000000000000000 -1.000000000000000
contrast_data 1.000000000000000 -1.000000000000000
% ... following above pattern for more groups
%%%%%%%%%%%%%%%%%%%%%%%
% Contrast Data End %
%%%%%%%%%%%%%%%%%%%%%%%
%------------------------------------------------------------------------
Here's the batch file automatically created from latest version:
%------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Result File Name Start %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Note: Result file must be listed first, and must follow the file
% name format of xxxx_yyyyresult.mat, where xxxx stands for
% "any result file name prefix" and yyyy stands for the name
% of PLS module (either PET ERP fMRI BfMRI STRUCT or SmallFC).
% File name is case sensitive on Unix or Linux computers.
result_file g2_STRUCTresult.mat
%%%%%%%%%%%%%%%%%%%%%%%%%%
% Result File Name End %
%%%%%%%%%%%%%%%%%%%%%%%%%%
%------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%
% Group Section Start %
%%%%%%%%%%%%%%%%%%%%%%%%%
group_files g1_STRUCTsessiondata.mat
group_files g2_STRUCTsessiondata.mat
group_files g3_STRUCTsessiondata.mat
group_files g4_STRUCTsessiondata.mat
% ... following above pattern for more groups
%%%%%%%%%%%%%%%%%%%%%%%
% Group Section End %
%%%%%%%%%%%%%%%%%%%%%%%
%------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%
% PLS Section Start %
%%%%%%%%%%%%%%%%%%%%%%%
% Notes:
% 1. Mean-Centering PLS
% 2. Non-Rotated Task PLS (please also fill out contrast data below)
% 3. Regular Behav PLS (please also fill out behavior data & name below)
% 4. Multiblock PLS (please also fill out behavior data & name below)
% 5. Non-Rotated Behav PLS (please also fill out contrast data and
% behavior data & name below)
% 6. Non-Rotated Multiblock PLS (please also fill out contrast data and
% behavior data & name below)
pls 2 % PLS Option (between 1 to 6, see above notes)
% Mean-Centering Type:
% 0. Remove group condition means from conditon means within each group
% 1. Remove grand condition means from each group condition mean
% 2. Remove grand mean over all subjects and conditions
% 3. Remove all main effects by subtracting condition and group means
mean_type 0 % Mean-Centering Type (between 0 to 3, see above)
% Correlation Mode:
% 0. Pearson correlation
% 2. covaraince
% 4. cosine angle
% 6. dot product
cormode 0 % Correlation Mode (can be 0,2,4,6, see above)
num_perm 0 % Number of Permutation
num_split 0 % Natasha Perm Split Half
num_boot 0 % Number of Bootstrap
boot_type strat % Either strat or nonstrat bootstrap type
clim 95 % Confidence Level for Behavior PLS
save_data 0 % Set to 1 to save stacked datamat
%%%%%%%%%%%%%%%%%%%%%
% PLS Section End %
%%%%%%%%%%%%%%%%%%%%%
%------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Condition Selection Start %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Notes: If you don't need to deselect conditions, just leave
% "selected_cond" and "selected_bcond" to be commented.
% First put k number of 1 after "selected_cond" keyword, where k is the
% number of conditions in sessiondata file. Then, replace with 0 for
% those conditions that you would like to deselect for any case except
% behavior block of multiblock PLS. e.g. If you have 3 conditions in
% sessiondata file, and you would like to deselect the 2nd condition,
% then you should enter 1 0 1 after selected_cond.
%
selected_cond 1 1
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Condition Selection End %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%
% Contrast Data Start %
%%%%%%%%%%%%%%%%%%%%%%%%%
% Notes: only list selected conditions (selected_cond)
contrast_data 1.000000000000000 1.000000000000000
contrast_data -1.000000000000000 1.000000000000000
contrast_data -1.000000000000000 1.000000000000000
contrast_data 1.000000000000000 1.000000000000000
contrast_data 1.000000000000000 -1.000000000000000
contrast_data -1.000000000000000 -1.000000000000000
contrast_data -1.000000000000000 -1.000000000000000
contrast_data 1.000000000000000 -1.000000000000000
% ... following above pattern for more groups
%%%%%%%%%%%%%%%%%%%%%%%
% Contrast Data End %
%%%%%%%%%%%%%%%%%%%%%%%
%------------------------------------------------------------------------
Baycrest is an academic health sciences centre fully affiliated with the University of Toronto
Privacy Statement - Disclaimer - © 1989-2024 BAYCREST HEALTH SCIENCE. ALL RIGHTS RESERVED