I am a MINITAB user and don't claim to have used it with the sample sizes your talking about. BUT, I would venture to say that it has the best reliability analysis tools of any general statistics package out there. Only Weibull ++ and (maybe) WeibullSmith are better. By "reliability analysis" I'm referring to nonparametric and parametric analysis of censored data. If that is the type of survival analysis you are wanting to do, you might call Minitab Inc. to see how MINITAB will handle the sample sizes you have. By the way, you can download a FULL working copy of MINITAB 13 from their website and try for yourself - http://www.minitab.com .The only catch is the software stops functioning after 30 days. I think MINIAB will be more expensive than NCSS, so if NCSS has the analytical tools you need it may be a better choice. If you'll be using the tool for a long time (not just a one time analysis) I suggest you get demos of all of them and see for yourself which meets your needs. Its the only sure way to tell. Ken "Robert Ehrlich" <[EMAIL PROTECTED]> wrote in message news:[EMAIL PROTECTED]... > I vote for NCSS with hard copy manuals (they are worth it). The maximum number of > samples that we have input to NCSS is around 500,000 and it still ran albeit very > slowly (I was using the time series procedure). > > all of your potential choices are OK. My choice of NCSS revolves around price, > good documentation and friendliness of an informed bu not (stats) prfessional user. > > Good Luck > > ELANMEL wrote: > > > I would appreciate any recommendations from the group for Software for a > > research project I am begining. I am considering NCSS, Minitab and SPSS (and > > maybe Stata). I am running Windows 98, using a home computer. > > > > I need to be able to merge two fairly large datasets (totaling about 50,000 > > observations x 800 variables, but with ~3000 observations used for analysis > > once merged), and perform a variety of statistical analyses, including > > correllations, regressions, logistic regression, factor analysis and survival > > analysis. > > > > NCSS is the least expensive option ($300 academic price w/ electronic manuals), > > Stata slightly more expensive, and SPSS much more expensive. The minitab site > > doesn't give prices, but I expect it to be lower than SPSS. > > > > Any help would be appreciated. Please email responses if you can! > ================================================================= Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =================================================================