JISCMail - SPM Archives

------------------ 原始邮件 ------------------

发件人: "MCLAREN, Donald";<[log in to unmask]>;

发送时间: 2016年4月8日(星期五) 晚上11:10

主题: Re: [SPM] comparing parametric modulations in different orders

For (1): You could do this. You will have 1 PM regressor for cond1 and 2 PM regressors for cond2 in the design matrix.

For (2): No. It doesn't make sense to compare the quadratic term to the linear term. At the second level, you'd want to compare the 2 linear terms and the 2 quadratic terms as you want to compare between conditions. This also means you need the quadratic term for cond1 as you want to show that the quadratic terms are significantly different.

Best,

Donald

Best Regards,

Donald McLaren, PhD

On Thu, Apr 7, 2016 at 8:51 PM, 春亮 <[log in to unmask]> wrote:

Hi all,

      I tried to set up parametric modulations (PMs) for cond1 and cond2 in my experiment. There is priori hypothesis that responses of relelvant brain regions would show linear relationships with PMs in cond1, and quadric relationships with PMs in cond2. My questions are:

        1). Can I only model the linear effects of PMs for cond1, and only the quadric effects of PMs for cond2 in the 1st level analysis?

         2). Afterwards, in the second-level analysis, to compare these PMs in different orders between cond1 and cond2 with a paired t-test in the 2nd level analysis?

All my best,

Chunliang