On Jun 24, 3:30 pm, Bill <connelly.b... from gmail.com> wrote:
> On Jun 25, 5:42 am, jonesmat <jones... from physiology.wisc.edu> wrote:
>> > But even with bugs I think it's better than most other packages. I am
> > not sure it will do the stats you need (but maybe). We always use
> > Prism for KS tests and such, after doing detection & measurement in
> > Axograph.
>> > Matt
>> Wait, hold the phone. Prism can do KS tests? I know it uses them to
> test for normality, but I didn't think you could use them to test
> between two distributions of your choice. Is that a standard feature?
> What version?
>> Axograph is good (indeed the best I have seen), and I was at a
> neuroscience course in Australia (ACANhttp://acan.jcs.anu.edu.au/acan.html)
> with John Clements, the guy who makes Axograph. We were running the PC
> version and it still wasn't 100% stable, certainly not stable enough
> for me to use it for acquisition of data (though John was bug fixing
> throughout the course, so maybe its better now).
>> Indeed, if I could get it to work with our CED data loggers I would
> probably be trying to convince the boss the use it.
Sorry, you're right. Prism uses KS to test for normaility but doesn't
have it built in as a standard nonpar test. It does, however, have the
Mann-Whitney, which is the one I always use. It compares median ranks
and is more powerful than KS, except in the case where there are a lot
of identical values between groups (i.e., "ties" - this will
presumably be rare in mini analysis, as long as your gain is set
reasonably high), or when the two groups differ in shape, rather than
in means (i.e., if both have a similar mean, but one has larger tails,
so that the cuulative distributions cross near the middle - also rare
in mini analysis). I'm using Prism 4.0a (2003), so maybe a newer
version would also have KS. Anyway, I think this program rocks, and
the help documentation is actually better than any statistics textbook
I've ever seen. Plus the author (Harvey Motulsky) has been known to
respond helpfully to emails about stats questions (my former postdoc
once got a lot of advicefrom him).
Axograph X v.1.2 is the latest one, and this is a "synchronized update
of both Mac OS X and Windows versions" - according to the recent
announcement. The new website is http://www.axograph.com/.