help

help >

**Using NBS to compare matrices between groups in cases where only 1 matrix exists per group**Feb 12, 2019 10:02 PM | Eric Plitman

Using NBS to compare matrices between groups in cases where only 1 matrix exists per group

Hi,

I wanted to start by thanking everyone involved for the creation of NBS.

I was hoping to ask whether NBS can be utilized to compare structural covariance matrices between groups, in scenarios where only 1 matrix exists per group.

I have detailed my current approach below, which does lead to the cycling of Permutations up to 5000, for example; however, the "Max Size Random", "Max Size Actual", and "Lowest p-value" columns stay stagnant at 0.0, 0.0, and 1.0, respectively, despite any manipulation of the "Threshold" and "Significance" parameters.

Trying to compare two matrices (e.g. patients vs controls), I set:

Design Matrix to a path to a .txt file that contains:

0 1

1 0

Statistical Test to "t-test"

Contrast to [-1,1] or [1,-1]

Connectivity Matrices to a path to the first of two .txt files containing 20x20 covariance matrices (simply Pearson correlations)

Method to "Network-Based Statistic"

Component Size to "Extent"

Any potential advice that might be offered in this regard would be extremely appreciated.

Thank you very much for your time and assistance,

Eric

I wanted to start by thanking everyone involved for the creation of NBS.

I was hoping to ask whether NBS can be utilized to compare structural covariance matrices between groups, in scenarios where only 1 matrix exists per group.

I have detailed my current approach below, which does lead to the cycling of Permutations up to 5000, for example; however, the "Max Size Random", "Max Size Actual", and "Lowest p-value" columns stay stagnant at 0.0, 0.0, and 1.0, respectively, despite any manipulation of the "Threshold" and "Significance" parameters.

Trying to compare two matrices (e.g. patients vs controls), I set:

Design Matrix to a path to a .txt file that contains:

0 1

1 0

Statistical Test to "t-test"

Contrast to [-1,1] or [1,-1]

Connectivity Matrices to a path to the first of two .txt files containing 20x20 covariance matrices (simply Pearson correlations)

Method to "Network-Based Statistic"

Component Size to "Extent"

Any potential advice that might be offered in this regard would be extremely appreciated.

Thank you very much for your time and assistance,

Eric

## Threaded View

Title | Author | Date |
---|---|---|

Eric Plitman |
Feb 12, 2019 | |

Andrew Zalesky |
Feb 13, 2019 | |

Klaas Bahnsen |
Dec 3, 2019 | |

Andrew Zalesky |
Dec 3, 2019 | |

Yi-Chia Kung |
Jul 15, 2021 | |

Andrew Zalesky |
Jul 15, 2021 | |