On Wed, 2 Oct 2002, David B. Klein wrote:
> As a footnote, ... if you are not writing raw code
> in the stats package of your choice, you are a wimp.
As that wording might offend some readers, let me rephrase it to suggest:
* GUI and WYSIWYG interfaces are inherently restrictive. They are useful
for novices and in prescribed tasks, but "what you see is ALL you get".
* David's original query nicely encapsulated the problem that to
understand what you are doing, you have to articulate it.
* Users who dislike commands because malformed commands may lead to error
messages are deluding themselves that just because the software has not
reported an error that it has done what you wanted and intended, or that
the computer has done something useful or correct.
* Using syntax (command language) in SPSS or other packages enables you to
keep a record (log, audit trail) of data manipulations and analysis, and
to adjust or repeat the analysis.
Specifically in respect of SPSS:
* Most menu operations provide a "paste" button that makes the syntax
visible in a notepad (text) window. This is a painless way to get a
skeleton command with its punctuation correct.
* Many operations are easier and quicker in syntax than from menus. One
area in particular is defining VAR and VALUE LABELS.
* Summary diagrams of each command are available in the on-line help, so
the full range of parameters can be seen. Some parameters are not
available in menus (eg, to include missing values row/column in
CROSSTABS).
* David could also have posted to the SPSS discussion list,
<[log in to unmask]>.
Finally, a plug that the annual meeting of SPSS users in York will be on 8
November - programme and booking details were on ALLSTAT 5 Sept.
R. Allan Reese Email: [log in to unmask]
|