Statistics software for the Mac

There a couple of things that I want to say without answering you question.

1) A couple of people have mentioned academic pricing. Make sure that you know what prices you have access to. Your school may well be able to give you access to to a package at a very low price. I just got SAS (PC only) for like $35. It's worth the time to hunt down the people at your institution who actually know about this. Of course, this holds true for all software if you are a student, but statistical packages are among the most expensive applications you might ever buy. And they have some of the deepest academic discounts (to get you hooked early).

2) Don't just choose your statistics package on price. Or even on ease of use. Being proficient with widely used packages can really help you to get a job. Talk to people in the field(s) that you are hoping to work in and find out what most people use.

At 11:36 AM -0800 11/3/04, Dave Fitch wrote:

You should understand that in SPSS the interface is really just a semi-pretty front end bolted onto a syntax interpreter [like ArcMap/ArcInfo etc.], and if you want to really take advantage of SPSS [or even if you work with files of more than 30 variables...] you're going to have to learn to program it [I use it on a PC at work]. If Tomoharu Nishino is right and you can't program the Mac version, then avoid it like the plague.

You CAN program SPSS.

Tomoharu was talking about a programming _environment_. I think that he meant that you basically just have a text window to work in when you try to work with the syntax itself.

But the programming itself can be carried across platforms.

-- =alex hoffman

A couple of other suggestions: if you're a student, you can get Matlab
for $99. (If you're not, you really don't want to know the retail price :)

There's also GNU Octave (open source, naturally), which is more or less
compatible with Matlab's programming language.

<http://www.mathworks.com/academia/student_version/index.html>

<http://www.octave.org/>

TimH

 
On Tuesday, November 02, 2004, at 11:56AM, Tomoharu Nishino <cheshirenekomac.com> wrote:
<snip>
>
>SPSS: Expensive. Mac was feature limited relative to Windows when
>both were at v.11, and is now 2 versions behind---at one point in the
>late 90s, they dropped the Mac all together. Included analysis tools
>is limited compared to STATA or JMP, and you must purchase add-on
>modules to do more sophisticated analysis. No command line interface,
>and GUI is clumsy. No built in scripting language (clumsily uses
>Applescript). No programming environment. They've gotten rid of the
>"grad student edition" which made it more affordable.
>
>JMP: Most comparable to STATA. Robust data analysis, nice scripting
>environment, good graphics. No real programming environment. Most of
>the analysis techniques missing from JMP is not likely to be missed by
>most users, nor is the programming environment---the two are probably
>comparable for most people not requiring the greater power and
>flexibility of STATA. The biggest difference between JMP and STATA is
>that the former has evolved primarily as a business tool, whereas the
>latter has evolved as an academic research tool. This means that STATA
>is supported by a large network of researchers who provide things like
>.ado files and publish web pages and papers on using and extending
>STATA. So, often times you are not dependent on "tech support" when
>you encounter an issue using STATA.

Unfortunately (and this irks me to no end), JMP allows ODBC data import only with their Windows and Linux versions. As someone who pulls large datasets out of Oracle, the ability to query the database directly is a huge plus. SPSS (and STATA, I believe) both permit this with their Mac versions.

Stefan

On 11/5/2004 8:26 AM, "Alexander Hoffman" <ahoffmanaledev.com> wrote:
> You CAN program SPSS.
>
> Tomoharu was talking about a programming _environment_. I think that he meant
> that you basically just have a text window to work in when you try to work
> with the syntax itself.
>
> But the programming itself can be carried across platforms.

Sometimes. In my experience using the latest version of SPSS, it was often
necessary to modify syntax files when crossing platforms.

On Nov 5, 2004, at 11:26 AM, Alexander Hoffman wrote:

> At 11:36 AM -0800 11/3/04, Dave Fitch wrote:
> You should understand that in SPSS the interface is really just a
> semi-pretty front end bolted onto a syntax interpreter [like
> ArcMap/ArcInfo etc.], and if you want to really take advantage of SPSS
> [or even if you work with files of more than 30 variables...] you're
> going to have to learn to program it [I use it on a PC at work]. If
> Tomoharu Nishino is right and you can't program the Mac version, then
> avoid it like the plague.
>
>
> You CAN program SPSS.
>
> Tomoharu was talking about a programming _environment_. I think that
> he meant that you basically just have a text window to work in when
> you try to work with the syntax itself.
>
> But the programming itself can be carried across platforms.

I was not very clear, so let me try to clarify. In addition to a robust
set of analytical tools, IMHO a stats program should have two
things---a scripting facility to automate your analysis tasks, and a
matrix programming environment. For the average user, the latter may
not be necessary, but the former is often very useful.

A scripting facility basically allows you to bypass the GUI and
automate the analytical task at hand using commands *already built into
the statistics program*. It allows you to automate repetitive tasks,
allows you to check and debug analyses, and it makes the task of
handling data sets with many variables easier. For academics, it also
aids in replication---just provide the script and data, and others can
see exactly what you did. This is what, I believe, Dave is referring
to.

SPSS seems to lack this facility. From their Mac version FAQ: " There
is no SPSS Scripting Facility. Instead, you can partly automate your
work by using AppleScript." Applescript is a poor substitute for a
real built in scripting facility that accesses the underlying command
set directly.

The other issue, is a matrix programming environment. Stats programs
are really just matrix algebra environments that are designed to run
specific matrix algebra computations in response to certain commands,
and pull out the appropriate numbers and display them in a reasonably
comprehensible manner. But sometimes, you will want to run an
analytical routine that is not built into the stats program---i.e.,
does not have a command that will execute the appropriate matrix
computations. When that happens, it is often useful to be able to
"roll-your-own" command by accessing the matrix algebra environment
directly. As far as I can tell, SPSS does not allow you to do that,
but STATA does---even if STATA does not have the command to do what you
need, you can program it yourself just as long as you understand the
underlying matrix algebra. In fact, a lot of people have extended
STATA by doing just that, often making their programs available to the
general public through plug-ins called .ado files.

Admittedly, this latter feature is probably not necessary for 99% of
users since the commands included in even the most "basic" stats
packages these days are very extensive. However, since learning a
stats package, its idiosyncrasies and command line syntax is pretty
cumbersome, ideally you would only have to do it once. So the one you
choose should be extensible enough to meet your future needs even if it
seems like overkill now.

Tn

Depending on the depth of analysis you require, a viable free solution
may be StatCrunch.

<http://www.statcrunch.com/>

Bill
---
http://billbauer.net

As a non-statistician I have found R very much worth investigating,
although I'm very much a beginner.

It has good graphing capabilities (under X11). Install R and run
"demo(graphics)" to check it out.

R scales well too -- and you can keep your data sets in a powerful
backend like postgres. Additionally postgres supports pl/r, which allows
postgres procedures to be written in R. See

    http://www.joeconway.com/oscon-pres-2003-1.pdf

A short, direct intro to R can be found here:

    http://www.odot.state.or.us/tddtpau/r/rbook.pdf

Cheers
Rory

--
Rory Campbell-Lange
<rorycampbell-lange.net>
<www.campbell-lange.net>

On 11/9/2004 9:23 AM, "Tomoharu Nishino" <cheshirenekomac.com> wrote:
> A scripting facility basically allows you to bypass the GUI and
> automate the analytical task at hand using commands *already built into
> the statistics program*. It allows you to automate repetitive tasks,
> allows you to check and debug analyses, and it makes the task of
> handling data sets with many variables easier. For academics, it also
> aids in replication---just provide the script and data, and others can
> see exactly what you did. This is what, I believe, Dave is referring
> to.
>
> SPSS seems to lack this facility. From their Mac version FAQ: " There
> is no SPSS Scripting Facility. Instead, you can partly automate your
> work by using AppleScript." Applescript is a poor substitute for a
> real built in scripting facility that accesses the underlying command
> set directly.

Although, as my previous posts indicated, I'm not a fan of SPSS, it does
have this functionality; SPSS calls it "syntax." SPSS syntax files work much
like Stata's "do file" language -- they're programs that perform data
manipulation and analysis using the software's extensive command set. In the
FAQ you quoted above, I believe SPSS is referring to GUI scripting: On the
Windows side, you can create scripts that perform sequences of GUI actions.

Does anybody know what happened to Joel West's excellent MacStats
site, which use to be at <http://homepage.mac.com/MacStats/">?

Arno.

At 11:35 AM -0800 11/10/04, Arno Wouters wrote:
>Does anybody know what happened to Joel West's excellent MacStats
>site, which use to be at <http://homepage.mac.com/MacStats/">?

Joel kept that site while pursuing his PhD at this institution. I'll
see if I can track him down.

--
Andrew Laurence
atlaurenuci.edu

At 10:10 AM -0800 11/11/04, Andrew Laurence wrote:
>>Does anybody know what happened to Joel West's excellent MacStats
>>site, which use to be at <http://homepage.mac.com/MacStats/">?
>
>Joel kept that site while pursuing his PhD at this institution.
>I'll see if I can track him down.

Update - Joel retired the site when [a] he no longer used the
software and [b] Apple started charging $100 for .Mac. It was mostly
OS 9 at the time, and only a smattering of OS X.

If anyone wants to host and keep it up, he has offered the contents.

--
Andrew Laurence
atlaurenuci.edu

I agree with the previous posts about the virtues of STATA and JMP.

The academic version of JMP (JMP IN) is a bargain at $79.

http://www.jmp.com/product/academic.shtml

Lee Willoughby

I have posted the MacStats web site at http://www.weborial.com/macstats/ and thank you, Andrew, for helping put us in touch. I've made some minor updates but over the next month will go through and bring everything up to date .

BTW I second the arrogance of SPSS... no online patches for Macs ("wait for your replacement disks and if you don't get them in a few months, call us"), "wait six months" for Tiger support... I wish the learning curve wasn't so high but will check out JMP (Stata's too pricey for me).

Does anyone have any experience with this web-based option for calculating statistics?

I recently asked a similar question on this list <http://emperor.tidbits.com/webx/TidBITS/Talk/449/12>. Based on the recommendations in that thread, I tried out R <http://www.r-project.org/>. (Because of it's name, information about R can be impossible to find with Google. Stick to "R project" for searching.)

Last week I finally put R through its paces on two recent experiments from our lab. It performed spectacularly. It's pretty easy to learn using online tutorials -- in particular John Verzani's tutorial which is a course in introductory statistics using R <http://wiener.math.csi.cuny.edu/verzani/tutorials/simpleR/index.html>.

The highlight of yesterday: figuring out the 15 or so commands to import, parse, slice and graph a 3-way comparison of control subjects using a scatterplot and a violin plot. Then using BBEdit to search and replace the word "control" with my two experimental conditions, pasting that back into R, and generating a report with all 6 graphs in about 3 keystrokes! Now that's how a program ought to work.

But the MAJOR advantages of R are that it is absolutely cross-platform (Linux, MacOS, Windows) and that it's open source. You've a good chance of accessing your data 10 years from now, which I wouldn't say with the commercial packages. The user base is large, active, and productive. The S language on which it's based is a well-accepted standard in statistics, and has been since its development in the early 1980s. R has stood the test of time and is likely to continue to do so.

There is one significant caveat: R is not a mouse lover's program. It is relentlessly command-line driven, and even the graphs cannot be edited with mouse clicks. It's trivial to take the PDF graphs into Illustrator, though, so this limitation hasn't been a problem for me.

--Ashish Ranpura.

DataDesk from http://datadescription.com/ is what we use in our one semester statistics class taught in the Math Department at Cornell. It was developed by Paul Velleman, Professor in Cornell's School of Industrial and Labor relations and a student of John Tukey from Bell Labs and inventor of many ideas and principles of exploratory data analysis.

DataDesk is superb for exploratory data analysis. The program emphasizes data analysis and content not chart junk (use Keynote or that other program people use for that :-). The professional version is pricy. A student version is bundled with several textbooks Velleman has been involved in writing. DataDesk was originally developed for the Mac and makes splendid use of drag and drop, clickable, live interfaces and everything a Mac User might wish for (you can lasso points in a graph and the data values from the underlying data tables will be highlighted, and vice/versa). The student version is light on multivariate statistics and enormous data sets.

The CD for student and full versions support both Mac and that OS from Redmond.

For what it's worth, I got some anomolous results with statcrunch. There are lots of free programs out there for the Mac though! I sure learned a lot from doing http://www.weborial.com/macstats ! (and in a couple of weeks when work slows down, I'll be doing a lot more updates...!)

But the real reason I'm posting is to make everyone aware that SPSS released a new update to v. 11 which really, really speeds things up and resolves a lot of issues - and lets it run under Tiger.