Statistical Navigator Professional: advice on choosing statistical procedures
| Published date | 01 March 1992 |
| Pages | 165-166 |
| Date | 01 March 1992 |
| DOI | https://doi.org/10.1108/eb045148 |
| Author | John Richardson |
| Subject Matter | Information & knowledge management,Library & information science |
Software Reviews
Statistical Navigator
Professional: advice on
choosing statistical
procedures
John Richardson Jr.
Statistical Navigator Professional advises users on which
statistical procedures may be most appropriate for their
applications. It runs on PCs and compatible computers. The
vendor is Idea Works 1991, 607 Jackson
Street,
Columbia,
MO 65203, USA. Tel:
+1
(314) 875
5827.
As librarianship adopts the social science model of research,
the statistician plays an increasingly important role in the de-
sign and dissemination of results. In the past, researchers have
been accustomed to taking an appropriate statistics course,
reading the statistical literature, and more than occasionally
consulting with a departmental colleague or campus consult-
ant. Now, however, it is possible to load software which will
advise on appropriate statistical procedure into your personal
computer.
Statistical Navigator Professional is an expert system cum
hypertext. Using KnowledgePro as the development tool, the
authors, Brent and Mirielli, created an expert system which
queries the user and makes recommendations; the hypertext
portion of the package links highlighted keywords with
brief,
textual explanations and more than 200 references to statisti-
cal sources.
Not really for beginners
Whatever the intended audience, this
5-disk
package is not
really for beginners, although it can narrow the hunt for pro-
ductive procedures. Looking over my shoulder, one of my
colleagues said, 'If you know the answers to these questions,
you don't need this package.' Despite an assertion on page
1-7 of the manual that users need no more than a basic statis-
tical background, I have to agree with my colleague.
One way for the package to address the needs of
a
broader
group of users would be to take fuller advantage of hyperlink-
ing capabilities (especially in the objectives and assumptions
section of the software). In addition, the package could in-
clude material which discusses how users can think statisti-
cally about their research problem. Such an approach would
make this software more useful for novice researchers.
Browse or consult mode can be used
The decision-making process begins with either the browse
mode or the consult mode. In the browse mode, hypertext
links are used to examine an alphabetic listing of
15
pages of
terms,
statistical analysis strategies, and a classified bibliog-
raphy of statistical sources. (Incidentally, the on-disk bibliog-
raphy, unlike the printed manual, contains quite a few
typographic errors.)
In the consult
mode,
a menu allows the user
to
select a new
consultation, a prior consultation, or to quit. If a new consult-
ation
is
selected, the broad category of analysis screening mo-
dule appears, asking for the researcher's name and project.
Proceeding further, users are asked to consider the general
types of analysis which can be done and what their research
questions are. In addition to menu selection, users can enter
descriptive phrases into the package's modest natural lan-
guage parser.
The first selection reveals eight common types of statisti-
cal analysis, from exploratory data analysis to process and
control analyses.
The next step is to examine twelve research questions.
Statistical Navigator Professional expects the user to type a
phrase beginning with 'to' (which, of course,
is
not
a
question,
but
a
way of expressing one's
purpose).
Selecting one of these
options results in a 'WARNING' message (which appears
overly threatening, as though the user did something wrong;
however, the system designer's intent seems merely advi-
sory).
Choosing to 'continue' means that users will be asked
to rate their objectives and assumptions on a 10-point Likert
scale.
Guessing is encouraged
Unfortunately, the novice user may not know how to answer
a
question.
To
meet
this
possibility the package could have con-
tinued with its hypertext approach, allowing users to select a
highlighted word and get added information in order to
answer the question accurately. However, with the current
arrangement, novice users are encouraged merely to guess at
answers.
The package offers the option to change answers before
the final rank ordering of recommended statistical procedures
The Electronic Library, Vol. 10, No. 3, June 1992 165
To continue reading
Request your trial