Corrected p-value during correlation analysis

Hi everyone,

I am interested in doing multiple corrections while doing some correlation analysis.

I have brain thickness data of 50 subject's 14 brain networks: 7 networks on left hemisphere and 7 networks on right hemisphere i.e. X = 50x7x2. I would like to see whether thickness of each network of these 50 subjects is correlated with age i.e. my age variable is a vector of size 50 values i.e. Y = 50x1.

Could you please tell me how and which MATLAB program can I use to estimate 14 correlation coefficient values and corrected p-value of each of these correlations?

Thanks a lot.

Answer 1

[Walter Roberson]

T1 = reshape(thicknesses, [], 14);
T2  = reshape(ages, [], 1);
T3 = [T1, T2];
c = cov(T3);
correlations_you_care_about = reshape( c(end,1:end-1), 7, 2);

[Sahil Bajaj]

Hi Walter, Thanks for your reply. In the last line, correlations_you_care_about = reshape( c(end,1:end-1), 7, 2); I am not sure how reshape( c(end,1:end-1), 7, 2); will calculate correlations and p-values. Are there typos in this line?

Answer 2

[Star Strider]

I would do a regression. See the ‘Statistical Analysis’ section in ‘Associations between age and gray matter volume in anatomical brain networks in middle-aged to older adults.’ Aging Cell. 2014 Dec;13(6):1068-74.

If you have the Statistics and Machine Learning Toolbox, you can do the sort of categorical regression they discuss, with multiple comparisons. The most obvious reason to adapt their statistical analysis procedure is that these authors have published in a refereed journal, and you would be using an accepted technique.

There may be other papers with statistical analysis techniques better suited to your experimental design. Search PubMed with the search string brain & volume & age for all 12276 similar articles.

I wish you success in your research.

[Sahil Bajaj]

Thanks Star for suggesting this publication. This looks a useful paper to me.

[Star Strider]

My pleasure.