On Friday 04 June 2004 04:45, Gary wrote:

Hi BandiPat,

On Thu, 3 Jun 2004 22:26:00 -0400 UTC (6/3/2004, 9:26 PM -0500 UTC my time),

BandiPat wrote:

...

...

Does anyone know of a statistical program that will run in Linux? I am looking for one that will run the Heckman's Phase 2 model, as SPSS will not run it. The only one I know of is SAS, but it is for windows only.. (costs a lot of money too, and is prohibitive) <g>

B> Just a thought, but have you ran a search for one while in Yast2 > B> Install/Remove software module? I thought there was something of this B> nature included, not sure.

I forgot about that ability... thanks, Lee, will give it a shot.

And here are a few links that might interest you: The table of equivalents / replacements / analogs of Windows software in Linux: http://linuxshop.ru/linuxbegin/win-lin-soft-en/table.shtml The Linux Software Map (LSM): http://lsm.execpc.com/lsm/ SAL (Scientific Applications on Linux): http://ftp.llp.fu-berlin.de/lsoft/index.shtml Cheers, Leen

Gentlemen, On 04 Jun 2004 10:10:48 +0200 UTC (6/4/2004, 3:10 AM -0500 UTC my time), Ulrich and others as above, wrote in part:

...

And here are a few links that might interest you:

The table of equivalents / replacements / analogs of Windows software in Linux: http://linuxshop.ru/linuxbegin/win-lin-soft-en/table.shtml

The Linux Software Map (LSM): http://lsm.execpc.com/lsm/

SAL (Scientific Applications on Linux): http://ftp.llp.fu-berlin.de/lsoft/index.shtml

U> A really good statitical software i R (www.r-project.org). It runs from U> the command line and is one of the best and most flexible open source U> statistical packages. Many thanks for your information in your replies. This is most helpful for the task at hand... Your input is most appreciated. -- Gary The beatings will continue until morale improves.

Hi Anders, On Fri, 4 Jun 2004 19:20:31 +0200 UTC (6/4/2004, 12:20 PM -0500 UTC my time), Anders Johansson wrote:

...

Does anyone know of a statistical program that will run in Linux? I am looking for one that will run the Heckman's Phase 2 model, as SPSS will not run it. The only one I know of is SAS, but it is for windows only.. (costs a lot of money too, and is prohibitive) <g>

thanks for any input.. I know it is a long-shot.

A> There's a program called GAUSS (www.aptech.com) that has a module for A> something called Heckman 2 step estimation. Not sure if that's related to A> what you're looking for, but... Anders, you have done it again... That is *exactly* what I am looking for. That is precisely it... He won the Nobel Prize (Economics) for this in 2000, for his development of theory and methods for analyzing selective samples - empirical microeconomics I can't believe you found this... Many, many thanks echoed in an endless loop! -- Gary

Hi Nick, On Fri, 04 Jun 2004 23:14:52 +0100 UTC (6/4/2004, 5:14 PM -0500 UTC my time), nick murphy wrote: n> I use JMP-in 5.1 from SAS. It is a general purpose statistical n> package that can do just about the same as SPSS. It can be bought on an n> academic license for less than $100, this is very favorable compared to n> the for commercial license cost. Very good to know. We have SPSS, but it will not support the Heckman's 2 step model, where SAS does support this via their own macro language, as developed by a third party.. price is right... now I have to compare the cost of JMP-in vs Gauss, and if JMP-in will support the macros.. thanks for your input, and now can make a decision based on all this great information.. much appreciated. -- Gary TEAMWORK...means never having to take all the blame yourself.

On 07-Jun-04 David Johanson wrote:

Gary wrote:

...

Hi Guys, Does anyone know of a statistical program that will run in Linux? I am looking for one that will run the Heckman's Phase 2 model, as SPSS will not run it. The only one I know of is SAS, but it is for windows only.. (costs a lot of money too, and is prohibitive) <g>

thanks for any input.. I know it is a long-shot.

S-Plus, a much more powerful program than SPSS, written for both win and lin, but costs around $1000. A FULL scripting stats package that easily rivals SAS for capabilities.

By and large, if thinking of S-Plus then, instead, use R (as already suggested). R is essentially the same as S-plus as far as the scripting language is concerned, and its repertoire of standard statistical models and procedures is also equivalent. R also has hundreds of additional packages (called "libraries") for additional more specialised purposes (S-Plus of course also has extra libraries, for similar reasons). Thee are, however, some differences between them with respect to repertoire of graphical capabilities (e.g. the "spin-plot" capability of S-Plus, for visualising multidimensional data, can be achieved in R by means of the rgobi library which requires xgobi on the host machine together with a compatible tcl/tk, so it's more of a hassle to set up). In principle, anything which can be done with S-plus can be done with R, and in essentially the same way. However, there may be resources which you can get for S-Plus as extras (at extra cost ... ) which are not available, or available in a different form or with a different interface, for R. So you might have to "roll your own", which could involve substantial effort in some cases. R is of course free (also Free). Likewise the superb r-help mailing list which includes some of the world's best statisticians in its contributors, and experts in R and S-Plus. Responses can vary from gently (or not so gently) correcting some misunderstanding which you have mailed to the list, to coming back in short order with a substantial block of code which respresents a new solution to a tough problem. To be frank, I don't know myself whether there is, somewhere, an R implementation of the Heckman 2-step procedure, though a quick search through the r-help archives threw up some pointers to queries about it a couple of years ago. I'm not an econometrician myself. The way to find out is to sign up to r-help and post your query. All relevant pointers can be found at http://www.r-project.org Someone has suggested matlab. I used to use matlab (and then for some years octave) since their early days (around 1990) for all statistical analysis. Their statistical resources ("toolkits") were then practically non-existent, so I had to roll my own. Since matlab and octave are array-oriented languages for numerical computation, this was relatively easy provided one had access to the mathematical formulation of the method to be implemented, compared with languages not primarily array-oriented; but it could still be a major effort. These programs do implement the best in numerical analysis algorithms, and are still amongst the leaders for signal analysis, solving systems of differential equations, and the like. Nowadays, I understand that matlab has come a long way in providing statistical functionality, and certainly octave has come some way though not a very long way. It was around the time that I was rolling up my sleeves to try to implement algorithms in octave for Generalised Linear Models that I came upon R (I knew about S, but didn't have access to it). Since then I have increasingly used R, now almost exclusively. I thoroughly recommend R for statistics. It's what it was born to do (like S), and now that it is mature it is superbly capable. Best wishes, Ted. -------------------------------------------------------------------- E-Mail: (Ted Harding) Fax-to-email: +44 (0)870 167 1972 Date: 07-Jun-04 Time: 09:07:24 ------------------------------ XFMail ------------------------------