help
81 Subscribers 
help > Correlation with age
Showing 1-4 of 4 posts
Mar 7, 2018 07:03 PM | Sofia Abrevaya
Correlation with age
Hello! I have a large sample of controls, and I want to see whether
there is a correlation between age and connectivity.
How do I set up the design matrix and statistical test?
How do I set up the design matrix and statistical test?
Mar 8, 2018 12:03 AM | Andrew Zalesky
RE: Correlation with age
Hi Sofia,
The manual has an example of this design. This requires a simple design matrix that would comprise a columns of 1's followed by a column with the age of each subject.
The contrast would be [0 1] or [0, -1]
And the test would be t-test.
Andrew
Originally posted by Sofia Abrevaya:
The manual has an example of this design. This requires a simple design matrix that would comprise a columns of 1's followed by a column with the age of each subject.
The contrast would be [0 1] or [0, -1]
And the test would be t-test.
Andrew
Originally posted by Sofia Abrevaya:
Hello! I have a large sample of controls, and I
want to see whether there is a correlation between age and
connectivity.
How do I set up the design matrix and statistical test?
How do I set up the design matrix and statistical test?
Jan 22, 2019 05:01 PM | Daniel Brennan - Icahn School of Medicine at Mount Sinai; New York University
RE: Correlation with age
Hi Andrew,
Piggybacking this thread; are the resulting networks using this design "correlated" with the predictor variable (here, age)? Or is the result a beta weight of some sort?
Piggybacking this thread; are the resulting networks using this design "correlated" with the predictor variable (here, age)? Or is the result a beta weight of some sort?
Jan 23, 2019 04:01 AM | Andrew Zalesky
RE: Correlation with age
Hi Daniel, if the predictor and explanatory variables are both
demeaned and normalized by the standard deviation, the beta value
is indeed the Pearson correlation coefficient.
Andrew
Originally posted by Daniel Brennan:
Andrew
Originally posted by Daniel Brennan:
Hi Andrew,
Piggybacking this thread; are the resulting networks using this design "correlated" with the predictor variable (here, age)? Or is the result a beta weight of some sort?
Piggybacking this thread; are the resulting networks using this design "correlated" with the predictor variable (here, age)? Or is the result a beta weight of some sort?