From your description, I believe that 9.6 and 8.5 are not standard errors (numbers used to evaluate the uncertainty associated with the sample mean) but are standard deviations (numbers used to evaluate the spread of a population). Therefore to make the kind of comparison you're trying to do you would have to divide 9.6 and 8.5 by the square roots of the respective sample sizes. The confusion comes about because, for some reason, it has become a convention to report "mean +/- sd" in the medical literature where the "+/-" which doesn't make a lot of sense. Reporting "mean +/- std. error" would at least be a 68% confidence interval. I think that it would be better if people simply reported (mean, sd, n), that way people would have all of the relevant info.
It's great to see fMRI getting some press, but
the article fails to mention some of the important
limitations of the technology. The magnitude
of the signal is only 1-5% over the noise and
comparisons need to be made at thousands of
locations. Also only very
simple tasks can reasonably be studied.
Regardless, the technology has great promise in
medical applications. I am currently invovled in a a study where fMRI is accurately distinguishing between patients who are at high risk for AD and controls.
As an additional plug, I think quantitative neurology is great area for CS, Math etc types to
get involved in.
Bioinformatics is a great field for CIS,
mathematicians, statisticians and quantitative
types to get involved in if they're doing it
for the right reasons. That is, if they like the
science and want to make a contribution.
Otherwise the field seems to be becoming
saturated with quantitative types who are
unwilling to make a real commitment to learning
the technology.
Slashdot Mirror
From your description, I believe that 9.6 and 8.5 are not standard errors (numbers used to evaluate the uncertainty associated with the sample mean) but are standard deviations (numbers used to evaluate the spread of a population). Therefore to make the kind of comparison you're trying to do you would have to divide 9.6 and 8.5 by the square roots of the respective sample sizes. The confusion comes about because, for some reason, it has become a convention to report "mean +/- sd" in the medical literature where the "+/-" which doesn't make a lot of sense. Reporting "mean +/- std. error" would at least be a 68% confidence interval. I think that it would be better if people simply reported (mean, sd, n), that way people would have all of the relevant info.
you need to go to the apostrophe protection society
It's great to see fMRI getting some press, but the article fails to mention some of the important limitations of the technology. The magnitude of the signal is only 1-5% over the noise and comparisons need to be made at thousands of locations. Also only very simple tasks can reasonably be studied. Regardless, the technology has great promise in medical applications. I am currently invovled in a a study where fMRI is accurately distinguishing between patients who are at high risk for AD and controls. As an additional plug, I think quantitative neurology is great area for CS, Math etc types to get involved in.
Bioinformatics is a great field for CIS, mathematicians, statisticians and quantitative types to get involved in if they're doing it for the right reasons. That is, if they like the science and want to make a contribution. Otherwise the field seems to be becoming saturated with quantitative types who are unwilling to make a real commitment to learning the technology.