assess 2006 - ASSESS SPSS User group

| | |
|ASSESS 2006 |[pic] |
| | |
| |(photo JFH 25 Dec 2005 |
|Old Dog, Old Tricks |) |
| | |
|Using SPSS syntax to avoid the mouse trap | |
| | |
| | |
|John F Hall[1] | |
1: Introduction
I was (and at heart still am) a survey researcher. From 1970 until 1992
(when I took early retirement)
this was at senior level. I specialised in advisory, design and
collaborative survey work, getting value for money, and in research rescue
jobs. According to the late Dr Mark Abrams, I "moved easily and widely in
the world of social research", developing subjective social indicators
("Quality of Life") and conducting surveys (for clients or on research
grants). From 1973 until 1992, and even after retirement, I provided
expert advice on, and training in, the management and analysis of social
survey data using SPSS. From 1972 to 1993 I used SPSS on an almost daily basis (exclusively in
syntax mode and on a range of computers (CDC 2000, ICL1900 and 2900 series,
Dec10, Dec20 and finally a Vax cluster) to process and analyse dozens of
surveys, either for clients or as part of professional and academic
research programmes. In 1974 I organised the international conference at
LSE, Social Science Data and the New SPSS (the final design exercise
preparing for a new release) and in 1978 set up and chaired the UK SPSS
Users' Group (UKSUG, a predecessor of ASSESS) and edited its newsletter. I
have trained or advised hundreds of researchers and students to use SPSS. This presentation draws on my experiences of using SPSS, not only on
surveys I have worked on, but also from the hands-on postgraduate course,
Survey Analysis Workshop, (part-time, evening)
which I developed and taught at the then Polytechnic of North London (PNL)
from 1976 to 1992. Most students, at least in the early days, came with
little or no previous experience of statistics or computers: many of them
could not even type!. I have lived in France since 1994, but maintain an
active interest in the development of, and training in, social research
methods in the UK, and in the Mark Abrams Prize[2], awarded annually by the
Social Research Association for the best piece of work linking survey
research, social theory and/or social policy. In 2001 the Social Research Association, in return for a 300 word review,
offered a copy of Julie Pallant's SPSS Survival Manual[3] which, motivated
in equal measure by curiosity, vanity and greed, I duly requested. Apart
from a short consultancy in 2000 for the Institute of Employment
Studies[4], I had been out of serious SPSS action since 1993. All my
previous experience of SPSS had been on mainframes in batch mode (with some
interactive on the Vax) using VMS and EDT: my PC experience was limited to
DOS, VPPlanner and WordStar4 (keyboard only, no mouse). I had barely got
the hang of using a mouse for email with Outlook Express and had never used
Word, Windows or SPSS for Windows. Imagine my absolute horror (and panic!)
when I opened the Pallant book and found it was about SPSS for Windows
using nothing but drop-down menus and point-and-click. (It also had
practically nothing on data checking, data entry, manipulation or
tabulation, but that's another story.) My first ever search of the web was
for SPSS for Windows tutorials!.
Through the web, I located a number of university staff teaching SPSS as
well as dozens of course outlines and several down-loadable tutorials, the
most useful of which was from computer services at Bogazici University in
Turkey[5]. To help me get started, Jane Fielding (University of Surrey)
kindly sent me her entire course notes (and had to explain how to open a
blank syntax file!). To help with the review, SPSS France provided an
evaluation version of SPSS11 for Windows. After much frustration (and a
very steep and rapid learning curve) my first version of the Pallant review
ran to 3,500 words, more than 10 times the required size, but was
eventually whittled down to around 1,700 and was published in SRA News[6]
in November 2002. On the strength of this review (plus the extensive additional comments[7]
trimmed from the original) and with an undertaking not to use SPSS for
personal gain, I was awarded a 5-year licence to the full version. This
has allowed me not only to restore several of my early surveys as SPSS
portable files, but also to convert and update extensive training materials
from my PNL courses for SPSS for Windows. In return, all such training materials have been, or will be, made
available to SPSS Inc.: they will also be made freely available to members
of ASSESS or any other bona fide colleagues, teachers and students using
SPSS. Some are already posted on the Market Research Portal site[8], but
this site has problems with tabular formats, colour text and graphics and
an alternative site[9] is currently under development for the more complex
offerings. 1.1: Before SPSS Between 1965 and 1968, when I was a young interviewer/researcher at Salford
University, I contributed to a series of one-off computer programs[10] for
data entry and tabulation, some of which were later generalised[11] for
multiple sets of tables. They were written in Algol for an English
Electric KDF9[12] computer (all of 64K!! RAM) which occupied a room the
size of a small terraced house and took anything up to 40 minutes c.p.u.
time to produce a single table: all data and programs had to be punched on,
and all output produced on, and read from, 8-hole paper tape. You couldn't
see what you were typing (but you got very good at reading the hole-
patterns!. To get printout you had to put both program and output tapes
through a special tape-reader. 8-hole paper tape Paper tape punch
Basically these programs were for tabulation only, but with some extra
programs for specific tasks (e.g. error checking, data transformation,
percentages, chi-square calculation at both table and cell level), but
little else apart from table titles and no row or column labelling at all.
At the University of Birmingham[13] I modified them for the departmental
PDP11 (everything still on paper tape). At Salford, they survived as the
Salford Survey Suite until 1979 when I advised maintenance be discontinued
in view of the widespread availability of SPSS. The experience of writing
this suite gave me a unique programming insight into the handling of arrays
and other computing processes which was later invaluable in assessing the
capabilities and facilities (or lack thereof) in other software. It also
gave me a lifelong belief in the value of cultivating good working
relations with computer staff. In 1966, the only other piece of technology available was a manual
typewriter with a limited character set, but at least it was portable and
had a wide carriage to take foolscap paper sideways. In those pre-Tippex
days, all corrections and amendments to questionnaires and reports had to
be retyped and copies made using carbon paper. This machine still exists,
but was last used in February 1973 (by me) to type the questionnaire for a
"quickie" survey of attitudes of senior pupils[14] in a girls' public
school. | | |
|Manual (portable) typewriter |80-column Hollerith card |
| | | Elsewhere, data from questionnaire surveys were typically transferred to 80-
column Hollerith cards
and then counted on card-sorters or tabulators or processed by computer
using proprietary software (eg Donovan Data Systems) or by software
packages for statistical analysis (most of which were written in Fortran on
80-column cards) which were complex and difficult to use, especially by
social scientists, even if there was a manual. Boxes of cards were very
heavy and could easily be damaged by repeated processing or by being
dropped. | | |
|IBM 026 Card punch |Card-sorter |
| | |
| | |
You also needed some kind of printed guide to data layout, such as the
blank sheets used by the SSRC Survey Unit, indicating which data were
stored in each of the 80 columns. | | |
|[pic] |[pic] |
| | |
|SSRC Survey Unit 1971 (2 x 40) |SSRC Survey Unit 1975 (4 x 20) | This early technology placed severe limits on the design and conduct of
survey research, although it made for high-powered thinking and very
careful work! That's why, in card-based systems, much early statistics was
developed on 2 x 2 tables. It also affected what was possible, in the
time available, for the provision of practical training and courses. 1.2: The origins of SPSS SPSS first appeared in 1968, and was written in Fortran IV for an IBM 360
by three postgraduate students[15]. It came from Chicago to Edinburgh via
Tony Coxon and was implemented in 1970[16] at Edinburgh Region