[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Accuracy tests for PSPP
From: |
John Darrington |
Subject: |
Re: Accuracy tests for PSPP |
Date: |
Sun, 18 Aug 2013 07:02:02 +0200 |
User-agent: |
Mutt/1.5.21 (2010-09-15) |
If you download the source, build PSPP yourself, you can run the self tests
to see for yourself. The tests are located in the "tests" directory of the
source and are available for everyone to inspect. As we add new features,
or become aware of a bug, we add a test for it. There are currently just
over 1000 of them. These are run typically everyday. You can see today's
run at http://benpfaff.org/~blp/pspp-master/latest/log.html
How do they work? Typically, they compare the results that PSPP produces
to particular "known" results. From where do the known results come? Most
come from running the same tests on other software, usually SPSS. However
a few results have been taken from published databases such as
http://www.itl.nist.gov/div898/strd/
Much of the low level mathematical routines in PSPP come from the GNU
Scientific Library which has its own test suite.
I hope this information helps.
On Sun, Aug 18, 2013 at 12:07:14AM -0400, Hugo Alejandro wrote:
Hello. I wonder what the PSPP method used to test the accuracy of the
statistical results supported by the program and if they can be shared.
The information will use it to attach a report and slideshow that will
perform before the academic council of the university to which I am
affiliated.
The purpose is to present a report at the request of the academic council
of
the university, so it can be approved the official use of PSPP, as an
alternative to other similar software by teachers and students in
introductory courses statistics, biostatistics and research methodology in
the faculty of medicine and social sciences.
The idea is to set a precedent and that students have the right to choose
and
forcing teachers to be trained and accept the diversity of tools available
to
accomplish the same purposes.(and not force students to buy expensive
software license or use pirated softare)
For now is focused on the use of PSPP (SPSS is used very extensively in the
University), but this will also help to rely on other software like
OpenStat
or R.
The method I use now is to use modified databases, which conducted the
analysis using PSPP and compared with SPSS and Stata.
Any help is welcome. Thanks.
--
PGP Public key ID: 1024D/2DE827B3
fingerprint = 8797 A26D 0854 2EAB 0285 A290 8A67 719C 2DE8 27B3
See http://keys.gnupg.net or any PGP keyserver for public key.
signature.asc
Description: Digital signature