The word "power" can be misleading, and it would be good if you can define
your criterion specifically. Statistical power is defined as:
Power = 1 - P(type2)
One minus the probability of a type two error. Numerical recipes states
that you have to use a Monte Carlo simulation to estimate the hidden
distribution. Matlab and SAS do not support the computation of power or
the non-centrality parameter. Power does, however, increase with the
number of scans (volumes). So as long as you have enough scans, the power
will be adequate. This latter point is discussed by Henson on his website.
Linda Seltzer
UC Davis
> Dear Experts,
>
> I am planning an experimental design aiming at the investigation of bold
> response evoked by identification of odor stimuli. Recent literature
> pertaining to this topic suggests to use an event related design with
> long
> SOA. According to this literature, if an odor stimulus is delivered for 3
> sec. under two different experimental conditions, A and B, and I am
> interest
> in the activity elicited by the odor for A than B condition, then I have
> to
> implement the following experimental design.
>
> Odor stimulus condition A, duration = 3 sec.
> SOA = 10 sec.
> Odor stimulus condition B, duration = 3 sec.
> SOA = 10 sec.
> ...
> Odor stimulus condition A, duration = 3 sec.
> SOA = 10 sec.
> Odor stimulus condition B, duration = 3 sec.
>
> After I will perform the t-contrast: odor stimulus condition A > odor
> stimulus condition B.
>
> For reasons that concern the human olfactory system and my experimental
> task
> (odor identification), I cannot reduce the SOA.
>
> Now the questions are:
> 1)Does this experimental design have sufficient power to detect the
> effect
> of my interest (i.e., A > B)?
> 2)If I increase stimulus duration from 3 sec. to 6 sec., does the design
> power increase?
> 3)Should I use a variable (e.g., 10+/- 5 sec) than a fixed (e.g., 10
> sec.)
> SOA duration so as increase the design power?
>
> Many thanks in advance.
>
> Federico.
>
> Federico Tubaldi, Ph.D Student
> Dept. of General Psychology
> University of Padova, Via Venezia 8, 35131, Padova, Italy
> Present working position:
> Visiting student at the Dept. of Neuroradiology
> University of Tuebingen, Hoppe-Seyler Strasse 3, D-72076, Tuebingen,
> Germany
> email: [log in to unmask]
> mobile: 0049 0177 7466815
> mobile: 0039 335 1400451
>
|