![]()
I have found these websites to be so useful/important/crucial that they just
had to be given a special place on the website! Now that many websites
invariably resemble the outcome of over-zealous design students let loose with
graphical software and an unlimited budget, it is refreshing to note that some
sites still value sheer depth of content and knowledge above "glamour", adverts,
and cookie-laden pages!
MDS-X v.5
(7th July,
2006)
OpenStat
v.4
StatCalc
v.1 and 2
Grant Devilly's Software
John Crawford's Neuropsychology software page (7th
July, 2006)
The
NewMDSX software website. This is where the "Edinburgh"
MDS-X package now resides. Many of these fine programs came from the
Borg-Lingoes series - incorporating some of Guttman's scalogram analyses along
with PINDIS and INDSCSAL. The programs are now embedded in a fully functional
Windows-hosted environment which is XP compatible; each program contains newly
designed printable graphical output. NewMDSX v.4.0 includes the excellent
WOMBATS (Work Out Measures Before Attempting To Scale!) routine for producing a
variety of dissimilarity and similarity indices. The main features of the
program suite are listed at:
http://www.newmdsx.com/features.htm. Program Anaysis modules include: CANDECOMP, HICLUS, INDSCAL-S,
MDPREF, MINIRSA, MINISSA-N, MRSCAL, PARAMAP, PINDIS, PREFMAP, PRINCOMP, PROFIT,
TRISOSCAL, CONJOINT (formerly UNICON) and WOMBATS. I fell in love with
this software back in the early 1990s. Great stuff and a great site to visit.
The software is now a chargeable item - with several pricing categories, but a
reduced-functionality free trial download is available from the website so that
users can try before they buy. The original MSDOS programs may be downloaded
entirely for free from
http://www.newmdsx.com/programs.htm.
A new improved
version of NewMDSX was released on 1st June 2006. Version 5.0 replaces all
previous versions of NewMDSX.
The MDS-X project team are:
|
Professor A.P.M. Coxon (University of Edinburgh) |
apm.coxon@ed.ac.uk |
|
Dr A. Brier (University of Southampton) |
|
|
Professor C.L. Jones (University of Toronto) |
|
|
Mr D.T. Muxworthy (University of Edinburgh) |
|
|
Mr W. Otto (University of Zurich) |
|
|
Dr S.K. Tagg (Strathclyde University) |
|
|
Dr. Wijbrandt H. van Schuur (University of Groningen) |
|
|
Roy Omond (Blue Bubble Ltd) |
|
|
Niels Veldhuijzen (National Institute for Educational Measurement, Arnhem) |
![]()
Bill
Miller's Openstat Statistical software package. A free general
purpose statistics program that Bill is writing in his "retirement time"! This
is absolutely first rate analysis software that comes complete with a series of
accompanying book chapters (pdf format) that really are "stats book" chapters in
their own right. I recommend this software to students who cannot afford to
purchase SPSS or
STATISTICA for home use, but who nevertheless need to implement some statistical
calculations. In fact, this package outstrips even these two in terms of some
specialist functionality (such as data simulation and modelling). Bill maintains a newsletter for those who are interested - and is
always available via email. Download today and be pleasantly surprised! For the
programmers amongst us - Bill even provides the Borland C++ code for the
routines. However, on top of OPENSTAT, there are a variety of other useful
routines which can be downloaded for free - such as:
The MatMan
package which provides tools for learning about vector and matrix algebra. The user can enter scalars, vectors and matrices and perform a variety of operations among and on them. Scripts of operations may be saved so that a series of operations can be performed repeatedly on different sets of values. Sample scripts are included for matrix operations such as a solution for multiple regression analysis.The GradeBook
package provides the teacher at nearly any level of teaching with a means of recording test scores, attendance, etc. for a class of students. The teacher can create specifications for multiple tests and how they are to be weighted in the total course grade. Grades can be automatically assigned by specification of "cut" scores for the letter grades to be assigned. Scores can be weighted and combined as raw scores, z-scores, t-scores, percent correct or percentile ranks. Reliability indexes can be estimated for the individual tests as well as the weighted combination of tests and other grading criteria.The ItemBank
package provides the means for entering and storing five types of test items: multiple choice, true or false, matching, sentence completion and essay. The user can create a coding structure with multiple levels for categorizing each item in the bank. Entered items are stored with the correct answer or preferred answer as well as both classical and item response theory parameter estimates. Specifications for a test can be constructed that reflect item classifications (content), as well as types of item to include. One can also randomly select items to provide a means of creating multiple tests over the same content areas and choose to print the test or administer it on the computer screen.PicView
is a small program for viewing pictures you have saved on your diskettes or hard drive. It will view .BMP and .JPG files in any sub-directory and let you print them on the printer.CRISP and CRISPACT
are two packages written for private practice personnel such as psychologists. CRISP is the Counselor Record Information System Package and CRISPACT is an accounting package for storing charges and receipts as well as producing billings and billing summaries for client accounts.Openstat is constantly being updated - so check for the latest version the website - and sign up to the user-group to keep informed about statistical and program issues.
![]()
StatCalc. The StatCalc software (K. Krishnamoorthy Dept of Mathematics
University of Louisiana at Lafayette
http://www.ucs.louisiana.edu/~kxk4695/) is like a PC calculator that computes
table values and other relevant statistics for 25 probability distributions.
These are the distributions for which probabilities, parameters and moments are
computed: Beta, Bivariate Normal, Cauchy, Chi-square, Extreme Value
Distribution, F-Distribution, Gamma, Inverse Gaussian, Laplace, Logistic,
Lognormal, Noncentral Chi-square, Noncentral F, Noncentral t, Normal, Pareto,
Rayleigh, Student’s t, Weibull, Binomial, Poisson, Hypergeometric, Logarithmic
Series, Geometric and Negative Binomial. A help-file
is provided for each distribution. It also includes some nonparametric table
values, tolerance factors, and the bivariate normal distribution. Version 1.1.
of the software is downloadable from the website - and is FREE! Version 2.0 of
the software is now chargeable, and is available from the author (via email
request). StatCalc 1.1 and 2.0 comes with a reference book (including help
files, several illustrative examples, methods used for computing table values,
algorithms for random number generations, and hundreds of formulas). However, The
new version of this software is magnificent now - as is the new v.2.0 pdf manual
of 311 pages - containing the formulae, derivations, and computational
approximations used for each of the 34 distributions, as well as some really
good "how to" examples. Note that each distribution function is linked via a
document reference to the appropriate area of this manual -a very useful
feature. Not only does the software permit the associated probabilities to be
computed for each distribution, but it also contains implementations of some of
the more popular statistical inference tests and associated confidence and power
values for each. For those wanting a serious reference application and manual of
statistical distributions and associated attributes - this is
a goldmine of a resource! Version 1.1 is good - I can assure you that version
2.0 is a magnitude improvement and a "must have" for any researcher who is
involved in work which requires key components of knowledge of certain
statistical distributions. As an example of the kinds of routines and functions
available for a distribution, look at the contents section for the Normal
Distribution:
Normal Distribution
19. Computing Probabilities, Parameters
and Moments
20. t-Test and CI for Mean
21. Sample Size for t-test
22. Sample Size for given Precision of CI
23. Test about Normal Variance
24. Confidence Interval for Normal Variance
25. Test and CI for µ1-µ2
26. Power Computation of Two-Sample t-test
27. Testing and CI for µ1-µ2 when Variances are Unknown and Arbitrary
28. Testing and Interval Estimation of the Ratio of Variances
29. Exact Tolerance Intervals for a Normal Population
30. Exact Factors for Equal-Tail Tolerance Intervals
31. Simultaneous Hypothesis Testing for Quantiles
![]()
Grant
Devilly's Software. The Swinburne University (Australia) Centre For
Neuropsychology site run by Prof. Grant Devilly and his students. Essentially a
site devoted to Clinical & Forensic Psychology, Incident Recovery & Restorative
Treatment. As well as much useful clinical information and associated links, the
site contains a download area for some excellent specialist statistical software
routines which clinicians and clinical researchers might find of great interest.
Look under the "Resources for Psychologists" Header for the item "Free Software"
...
There are free and professional versions of the main packages - and screen shots etc. are available on the site. The software is of a very high quality and very user-friendly. The programs have obviously been tested in the field!
![]()
All programs are for PCs. Once downloaded, these programs can be run by any of the normal Windows procedures i.e. by clicking on file in File Manager, by using Windows 95 start menu, or by placing a shortcut on desktop etc.
When clicking to download the programs: most web browsers are configured to recognise that the files in questions are executables. If you have any problems (i.e. the browser treats them as text files), hold down the shift key when clicking.
As an illustration of these programs, you can view screen captures of the input form and results form; the example is singlims.exe a program for testing whether an individual's score is significantly lower (or higher) than a control sample (the program also provides a point estimate of the abnormality of the individual's score and 95% confidence limits on the abnormality)."
From John's siteThis is an excellent software and intellectual resource. As with Grant Devilly's software, it is clear that these programs have obviously been tested in the field!
![]()