--------------2586F275B26CD6A7B663CE05
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

One of these may be better than the other to learn the math
underpinnings depending on what kind of statistics you are planning to
do.

Overall it is more likely that a statistical package is the best
software as the primary tool for most statistical purposes.
There will of course be differences in different fields, but in my
experience over 30 years using many packages I've come to the following
position. Using this set of desiderata [human factors of the user
interface, clarity and transparency and consistency of language, wide
availability, user documentation,  documentation of what was done (to
facilitate review), point-and-click aid in reading output, readability
of output, ease of data transformation especially the most common tasks,
ease of transfer of data to/from other packages, numerical accuracy, a
wide variety of non-esoteric stats ],  SPSS has no peer.

To further put this in context::
In common with many other applied statisticians I feel that at least 80%
of the time, and as much as 95% of the time in a analysis is quality
assurance: cleaning, prepping, doing descriptives,  finding miscodes and
other outliers, and getting a basic understanding of the data.  I place
a premium on the effectiveness and efficiency of this portion of the
effort.

Only a small percentage of researchers in fields such as psych, ed,
nursing, sociology, political science, public administration, business,
market research, etc. need stat methods that are not in SPSS.  The
situation is not as clear in other fields, but SPSS has done the job for
me for on specific jobs in  ops research, inventory evaluation,
astronomy, biology, and ecology.

However, a statistician/researcher sometimes needs to do less common
tasks.  It is easy to ouput cleaned and prepared data to many other
programs.  Many (e.g. SUDAAN, (I think) WESVAR, and some SEM programs)
can read an SPSS system file since SPSS makes its interface available.
These and Stata, S, EQS, Statistica, LISREL, LIMDEP, SAS, etc. and
stand-alone single purpose programs can be useful parts of one's
statistical s/w repertoire depending on the application area. On rare
occasions it is necessary to program in FORTRAN or other languages.

"Andreas K." wrote:

> Maple or Mathematica? Which of these CAS programs is the best one to
> use for a statistician?

--------------2586F275B26CD6A7B663CE05
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit

<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
One of these may be <b>better</b> than the other to learn the math underpinnings
depending on what kind of statistics you are planning to do.
<p>Overall it is more likely that a statistical package is the best software
as the <b>primary</b> tool for most statistical purposes.
<br>There will of course be differences in different fields, but in my
experience over 30 years using many packages I've come to the following
position. Using <b>this set</b> of desiderata [human factors of the user
interface, clarity and transparency and consistency of language, wide availability,
user documentation,&nbsp; documentation of what was done (to facilitate
review), point-and-click aid in reading output, readability of output,
ease of data transformation especially the most common tasks, ease of transfer
of data to/from other packages, numerical accuracy, a wide variety of non-esoteric
stats ],&nbsp; SPSS has no peer.
<p>To further put this in context::
<br>In common with many other applied statisticians I feel that at least
80% of the time, and as much as 95% of the time in a analysis is quality
assurance: cleaning, prepping, doing descriptives,&nbsp; finding miscodes
and other outliers, and getting a basic understanding of the data.&nbsp;
I place a premium on the effectiveness and efficiency of this portion of
the effort.
<p>Only a small percentage of researchers in fields such as psych, ed,
nursing, sociology, political science, public administration, business,
market research, etc. need stat methods that are not in SPSS.&nbsp; The
situation is not as clear in other fields, but SPSS has done the job for
me for on specific jobs in&nbsp; ops research, inventory evaluation, astronomy,
biology, and ecology.
<p>However, a statistician/researcher sometimes needs to do less common
tasks.&nbsp; It is easy to ouput cleaned and prepared data to many other
programs.&nbsp; Many (e.g. SUDAAN, (I think) WESVAR, and some SEM programs)
can read an SPSS system file since SPSS makes its interface available.&nbsp;
These and Stata, S, EQS, Statistica, LISREL, LIMDEP, SAS, etc. and stand-alone
single purpose programs can be useful parts of one's statistical s/w repertoire
depending on the application area. On rare occasions it is necessary to
program in FORTRAN or other languages.
<p>"Andreas K." wrote:
<blockquote TYPE=CITE>Maple or Mathematica? Which of these CAS programs
is the best one to
<br>use for a statistician?</blockquote>
</html>

--------------2586F275B26CD6A7B663CE05--

.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to