> Could you please let me know whether my scripts are correct to
> fulfill my purposes?

The script itself seems to be fine. One concern is that you have only two subjects. A more statistically valid analysis would usually require at least 10 subjects.

> If these 5 runs are scanned consecutively in the same session,
> coupld you please give me some scripts about how to analyze
> them according to my purpose?

Basically you can concatenate those 5 runs after timing correctoin and volume registration, and then run deconvolution/regression analysis.

You can see more details including an example script in HowTo5 ([afni.nimh.nih.gov]). Check specifically section "G. Concatenate runs with AFNI '3dTcat'".

> could you please give me some explanation about the difference
> of between 'the estimation of difference between two factor level
> means ( option -adiff or -bdiff )' and 'the estimation of a contrast
> ( option -acontr or -bcontr)' ? I do not completely understand
> their difference.

As you can see from the 3dANOVA manual, -adiff/bdiff tests whether two levels of a factor are equal. Two subbriks will be generated with one being intensity of the difference and the other its t statistic (double-tailed).

-acontr/bcontr is used to test any linear combinations among the levels of a factor, including difference between two factor levels. In other words, -acontr/bcontr is more general than -adiff/bdiff, and in fact you can run whatever contrast tests without resorting to -adiff/bdiff and -amean/bmean. In this sense -adiff/bdiff and -amean/bmean superfulous.

Gang