Next: Variance ratio smoothing
Up: Alternatives to the Random
Previous: Alternatives to the Random
Facing this dilemma of an invalid population analysis (fixed-effect approach) and a valid one
(random-effect) but difficult to apply usually because of overestimation of variances due to
small samples (random-effects), some alternatives have been investigated. One is the ``conjunction
analysis'' (Friston et al. [5]), which uses all the subjects' maps to localise
where all the subjects activated (at a chosen level; see single subject analysis,
it is a
thresholded map of the
map over the subjects). Then using simple probability theory, one can
relate the ``level of activation'' (p-value calculated for example from Gaussian Random Field
Theory - see Worsley(1999 submitted)) and the proportion of the population which shows this
activation at the previously defined level (in the single-subject analysis). Let
and
mean respectively, tested status of activation with the experiment and true status of
activation, while
and
is the status activated or not, simple probability calculus
gives:
Where
is the chosen single-subject level of activation (p-value level to threshold each
map),
is the power or sensitivity (
is the specificity) of the experiment which
is not known and can be set at
to provide a lower bound of the proportion
of the
population showing the effect. Setting
gives
Thus the conclusion about the population is
qualitative,
with a certainty of 0.95 (
), we can say that at least 80%
(
) of the population would activate at level 0.001 (
). The results given above
can also be used to decide on a sample size[6],
for the above conclusion one
would need at least
subjects.
Remarks:
The
could be calculated using permutation testing procedure instead of using random
field theory. The conjunction could also be defined as a given proportion of subjects; this could
cope better with problems such as poor localisation due to registration problems, as conjunction
analysis is certainly very sensitive to subject outliers in terms of the locations of the
activations. If one decides that
must activate to define a conjunction then given
:
with
as given above. One must notice that the amount of
conjunction
must be at least the expected
, otherwise the multiplicative function
introduced in the above equation is not monotonic with
. This problem is also linked with
values of
and the sample size
. Roughly speaking when
is close to
one
would need an
very close to
and so a large
to achieve some gain in performing a
conjunction of only
.
Next: Variance ratio smoothing
Up: Alternatives to the Random
Previous: Alternatives to the Random
Didier Leibovici
2001-03-01