[iDoctor]

Hi,

I need a good medical statistics software for Mac OS X?
What is the best out there?

Thanks
W.

[philm]

You have a few options:

JMP
SPSS
Staview (if you can get it)

My fave which I still use is the 1991 version of SuperANOVA (runs fine under 10.4.4), but that has not been available for a few years.

You can do a limited amount of statistical analysis with Excel's Data Analysis Add-In also.

[ghporter]

I have to ask what makes a package a "medical" statistics package? Ain't stats stats?

[philm]

Well, that's largely true, but my assumption is that the OP requires statistical analysis for standard clinical datasets which require certain analyses. Engineering or other disciplines are likely to require some different forms of analysis, although the packages I mentioned above could well suit a range of subjects.

[barney ntd]

R
(though it does have a bit of a learning curve)

[kman42]

Any sort of open source packages out there? Perhaps running under x11?

[barney ntd]

R is open source. It runs multiple graphics drivers including X11 and quartz. It is probably the main stats package used in academia, so has an enormous collection of specialist modules available. Its only disadvantage is that it's mostly used by statisticians, so expects you to know what you're doing.

[philm]

Originally Posted by barney ntd

It is probably the main stats package used in academia....

My experience (as an acadmic) does not concur with this statement, although it probably varies on the sort of work you are doing. R is undoubtedly very powerful, and I think there is some analogy along the lines of R is to LaTex what JMP (and the others I mentioned above) are to Word. R is pretty geeky in my experience. Just being called 'R' suggests that for starters. As does this line from the 'What is R?' part of its web page:

'R can be considered as a different implementation of S.'

Maybe R is used by academic statisticians and the other programs are for the rest of us.

[kman42]

JMP looks interesting, but it is hard to get past the marketing mumbo-jumbo on their website (as with the others you mentioned). It looks like it has a tutorial feature that uses the program to teach the user different stats methods; is this correct? I took stats a long time ago and use what I know, but I'm always wondering if there are different/better ways of looking at things. Would it be worth while to throw down the $30 for a six month license and check it out?

kman

[philm]

Yes, JMP does kick off Help Viewer with a bunch of tutorial-type explanations - not just how to use the software, but different methods of analysing paired data, descriptions of model fitting etc etc. I would say that the tutorials are....pretty good.