help > One sample t-test and surrogate data
Showing 1-2 of 2 posts
Feb 26, 2019 10:02 AM | Nick Heinrich
One sample t-test and surrogate data
Dear Andrew,
First of all, thank you for this very nice tool! I would like to ask something about the underlying process of the one sample t-test. What is written in the manual is: "The one sample test randomly flips the sign of each data point for each permutation".
Does that mean that the data are tested not solely against zero, but basically against a random network (a network that includes random values within each cell of the adjacency matrix) generated by this permutation method?
If that is the case, would the values have a mean of 0 and SD based on the data of the "original" network (the one tested against 0)?
Would that actually qualify as a process involving surrogate data?
Thank you, in advance, for your time,
Sincerely,
Nick Heinrich
First of all, thank you for this very nice tool! I would like to ask something about the underlying process of the one sample t-test. What is written in the manual is: "The one sample test randomly flips the sign of each data point for each permutation".
Does that mean that the data are tested not solely against zero, but basically against a random network (a network that includes random values within each cell of the adjacency matrix) generated by this permutation method?
If that is the case, would the values have a mean of 0 and SD based on the data of the "original" network (the one tested against 0)?
Would that actually qualify as a process involving surrogate data?
Thank you, in advance, for your time,
Sincerely,
Nick Heinrich
Feb 26, 2019 12:02 PM | Andrew Zalesky
RE: One sample t-test and surrogate data
Hi Nick,
The one-sample t-test is a very simple test to asses the null hypothesis of having an average connectivity strength of zero across subjects.
If the one-sample t-test yields a significant results, this means that the average connectivity strength is greater than (or less than) zero, depending on the contrast.
The data are never tested against a random network or surrogate data. The BCT toolbox has functions to test against random networks.
Andrew
Originally posted by Nick Heinrich:
The one-sample t-test is a very simple test to asses the null hypothesis of having an average connectivity strength of zero across subjects.
If the one-sample t-test yields a significant results, this means that the average connectivity strength is greater than (or less than) zero, depending on the contrast.
The data are never tested against a random network or surrogate data. The BCT toolbox has functions to test against random networks.
Andrew
Originally posted by Nick Heinrich:
Dear Andrew,
First of all, thank you for this very nice tool! I would like to ask something about the underlying process of the one sample t-test. What is written in the manual is: "The one sample test randomly flips the sign of each data point for each permutation".
Does that mean that the data are tested not solely against zero, but basically against a random network (a network that includes random values within each cell of the adjacency matrix) generated by this permutation method?
If that is the case, would the values have a mean of 0 and SD based on the data of the "original" network (the one tested against 0)?
Would that actually qualify as a process involving surrogate data?
Thank you, in advance, for your time,
Sincerely,
Nick Heinrich
First of all, thank you for this very nice tool! I would like to ask something about the underlying process of the one sample t-test. What is written in the manual is: "The one sample test randomly flips the sign of each data point for each permutation".
Does that mean that the data are tested not solely against zero, but basically against a random network (a network that includes random values within each cell of the adjacency matrix) generated by this permutation method?
If that is the case, would the values have a mean of 0 and SD based on the data of the "original" network (the one tested against 0)?
Would that actually qualify as a process involving surrogate data?
Thank you, in advance, for your time,
Sincerely,
Nick Heinrich