Email discussion lists for the UK Education and Research communities

## FSL@JISCMAIL.AC.UK

#### View:

 Message: [ First | Previous | Next | Last ] By Topic: [ First | Previous | Next | Last ] By Author: [ First | Previous | Next | Last ] Font: Proportional Font

#### Options

Subject:

Re: randomise design box setting with quadratic

From:

Date:

Tue, 4 Sep 2018 17:48:19 +0100

Content-Type:

text/plain

Parts/Attachments:

 text/plain (87 lines)
 ```Dear Andersson and other FSL experts, I was reading this post and had some questions regarding the model that is specified: EV1: intercept EV2: age EV3: age^2 EV4: sex To test the linear effect of age (positive or negative), use: C1: [0 1 0 0] C2: [0 -1 0 0] To test the quadratic effect, create a third contrast that is: C3: [0 0 1 0] And define an F-test that combines C1 and C3. This F-test will be the test for quadratic effects. a) Why is it important to model the intercept and what value should be assigned to this intercept EV? b) Why is only 1 contrast explicitly specified for the quadratic effect (C3: 0 0 1 0)? Shouldn't the following also be modelled C4: 0 0 -1 0? c) Why is it necessary to set up an F-contrast that combines C1 and C3 in order to test for quadratic effects? Wouldn't C3 and C4 probe this already? Many thanks for the assistance! Mauro Subject:Re: randomise design box setting with quadratic From:"Anderson M. Winkler" <[log in to unmask]> Reply-To:FSL - FMRIB's Software Library <[log in to unmask]> Date:Thu, 29 Dec 2016 23:44:50 -0200 Hi Yang, Create a design that is: EV1: intercept EV2: age EV3: age^2 EV4: sex To test the linear effect of age (positive or negative), use: C1: [0 1 0 0] C2: [0 -1 0 0] To test the quadratic effect, create a third contrast that is: C3: [0 0 1 0] And define an F-test that combines C1 and C3. This F-test will be the test for quadratic effects. All the best, Anderson On 28 December 2016 at 03:02, Yang, M. H. <[log in to unmask]> wrote: Hi all, I want to test linear and quadratic effects of age on FA with general linear models (GLMs) while regressing out the effects of sex. I finished TBSS and got my "mean_FA","mean_FA_skeleton" and "all_FA_skeletonized" file. Fist I tried to test linear,I set design box with GUI and put "age" in EV1 dummy variable "gender" in EV2, And I set contrast ev1 [1] and ev1 [-1] to test positive and negative effect. Am I right? How to setting design box if I want to test guadratic effects? Just put age square in EV3 and contrast [age gender agesquare]=[1 0 1] ? Best wishes, M. H. ######################################################################## To unsubscribe from the FSL list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1```