Dear Agnieszka

As you observed, because you have one value per subject, you cannot enter this at the 1st level. Note that theoretically, it makes sense since you don't want to 'remove' an effect from scan to scan. What you want is to look at the effects of interest having these 3 nuisance similar across subjects - this can be done at the 2nd level - in the ANOVA model enter the gps and conditions and add the 3 covariates.

Cyril

> Hello,
>
> I have data from an event related reading experiment, where items from 3
> conditions (words, pseudowords and a baseline – looking at a cross) are
> randomly intermixed. I have two groups of participants: patients and
> controls which I would like to compare using the 2nd level analysis using
> SPM5. I would like to include 3 variables as covariates (defined as
> nuisance variables), in my SPM5 analysis: ADHD Index (a score on the
> standardised questionnaire), Dyspraxia Index (a score on a non-standardised
> questionnaire, number of ‘yes’ responses) and % correct on the Post test.
> The only experience with covariates and SPM analysis which I have to date
> is with movement parameters (obtained in the Realignment stage of data
> pre-processing) which I entered as multiple regressors when setting up a
> Design Matrix. I presume that the ADHD, Dyspraxia and Post-test covariates
> can be entered at this stage. Is this correct? However, what bothers me
> here is the fact that in contrast to the motion parameters where, I think,
> there was a covariate for each scan (or more precisely, a set of
> covariates), here I have only one score (per covariate) per participant,
> e.g. participant1 would have the following scores: 68, 17, 70.
> I would be grateful for suggestions on how to enter such covariates into my
> 1st level analysis. Also I presume that one only does it in the 1st level
> analysis and not in the 2nd level analysis?