I am sorry that I still have problem with this rule: the sum of them should be zero. In another email, I saw: >I have a basic question, let say I have tow conditions: activation and >rest. (100 scans). >Vector onsets for the activation condition is: 10 30 50 70 90. >Duration: 10 Vector onsets for the rest condition is: 20 40 60 80. >Duration: 10 > >Is it the same to specify only the activation condition (contrast >[1]) than specifing both conditions (contrast [1 - 1])? >No, it's not the same: >- [1 -1] tests for positive differences between Activation and Rest; >- [1] tests for positive differences between Activation and "Anything that is not declared", i.e., your scans 0 to 9 that are not declared in your vectors. In the second case, the sum is also not zero. Should it be invalid? Robin _____ From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of y8003012003 Sent: Wednesday, September 28, 2005 8:33 PM To: [log in to unmask] Subject: Re: [SPM] are the two designs same? it is invalid because the sum of them was not 0. but I also want to know if the following are right: [-1 -1 -1 -1 4] and what it mean. Thank you very much -----原始邮件----- 发件人:"Robin CHARK" 发送时间:2005-09-28 19:53:52 收件人:[log in to unmask] 抄送:(无) 主题:Re: are the two designs same? One follow-up question. If I want to test the significancy of the first 4 conditions only, can I define it as [1 1 1 1 0]? Thanks. Robin > hi > >> Dear all: >> >> If I had a five conditions fMRI experiment, and made the following two >> t-contrasts using spm2 : >> >> [1 1 1 1 -1] and [1 0 0 0 -1; >> >> 0 1 0 0 -1; >> >> 0 0 1 0 -1; >> >> 0 0 0 1 -1] >> >> Are they all valid? And are they the same? If not, what are they >> meaning separately? >> >> Tank you very much! >> > well, they are not all valid and they are not the same .. > 1. When you enter a contrast, the sum of coef must be 0 and thus [1 1 1 > 1 -1] is invalid (you can try but SPM contrast manager will refuse to do > that) .. if you want to oppose one condition against the others the > correct contrast would be [1 1 1 1 -4] > 2. in the second case you test if any of these contrasts (or > combinations) is significant .. but those ones are correct as they all > sum up to 0 > > best > cyril <http://www.126.com> 需 要 一 个 2000 兆 的 免 费 邮 箱 吗 ? 网易免费邮箱是中国最多人使用的电子邮箱。