Back to PLS HelpPosted on 05/15/12 06:22:59
Number of posts: 3
Dear All,
I just want to understand what does a negative brain score within a condition means and whether one needs to take the absolute of those scores in order to associate them to some other variables like age?
Lets assume that we run task PLS and end up with a significant LV showing condition A vs. condition B.Let's further assume that condition A includes some subjects with negative brain score (although the mean BS is positive). Does that mean those subjects less reliably contributt to the pattern since their BS are negative?
Thanks
Kami
Replies:
I'm Online
Posted on 05/15/12 08:33:51
Number of posts: 291
BrainScore = datamat * BrainLV
Both positive and negative BrainScore mean strong contribution to the pattern. If the values are close to the zero, they mean less contribution.
Posted on 05/15/12 09:22:22
Number of posts: 3
Thanks for the reply.
I know the formula and I know that two sides of a contrast should be interpreted (A vs. B and B vs. A). But my question concerns whether I need to take the absolute value of a brain score within a condition (condition A) if it happened to be negative? That makes a big difference when it comes to corrolating BS with some external variables. Forexample, I got a significant LV showing condition A vs. condition B. when I correlate the condition A Brain scores with age, there is significant positive correlation. However, when I compute the correlation between absolute BS of condition A and age, the correlation starts to be significantly negative. So, my question is whether I need to correlate BS or absolute BS? On the other word, what is a difference between, forexample 50 and -50 as a brain score for condition A?
I do know that probabely the best way of doing so is to include age as a behavior in a behavioral type of PLS analysis, but for now, I do need to know whether I should consider BS or abs(BS) to be correlated with some external measures(e.g. age).
Thanks in advance
Kami
I'm Online
Posted on 05/15/12 09:32:10
Number of posts: 229
those subjects with negative scores will tend to show the opposite pattern (generally) across the brain compared to those with positive brain scores-
this is potentially very important information for you to consider - and the degree to which it is an 'issue' depends on how large the differences are. (If the spread of brainscores is very large, the LV is less likely to be significant)
so the easy answer, is yes, keep the sign, use that when correlating with some external measure - you may find that the people whose brain scores are on the negative side of the scale show you something quite interesting with respect to your external measure
cheers,
Nancy
Posted on 05/15/12 09:52:56
Number of posts: 3
Login to reply to this topic.
