Time
and place:
3:30-4:30,
Presentation in Room 2030, Psychology-Anthropology-Sociology
Building (PAS), University of Waterloo campus
4:30-5:00,
Open House at the SWORDC site (PAS 2228)
Presenters:
John Goyder and Keith Warriner (SWORDC
Co-Directors)
Who Should Attend? This event is oriented to
graduate students in any of the social sciences, whether it
be the more traditional disciplines such as economics, psychology,
demography or sociology or more applied units such as health
studies, consumer studies, leisure studies, child and adolescent
studies, family relations etc. Students in statistics, business
administration, accountancy and management sciences may also
have research interests relevant to the SWORDC. Students from
all the SWORDC partner universities (University of Waterloo,
Wilfrid Laurier University, University of Guelph, Brock University,
University of Windsor and the University of Western Ontario)
are cordially invited. Also, faculty members from the SWORDC
group are most welcome to attend.
A
SWORDC Primer
The Southwestern Ontario Research Data Centre, or SWORDC,
is a computing lab which houses Statistics Canada data sets.
It is one of nine such Research Data Centres recently established
across Canada, from Vancouver to Halifax. While university
library facilities, such as the Tri-Universities Data Resource
which serves three of the partner universities involved in
SWORDC, make Statistics Canada data widely and quickly available
to all comers, but in abbreviated form, the copies of surveys
in an RDC are master files. A master file has no deleted variables
or collapsed codes, as is necessary for the protection of
confidentiality in a public use version from a library. Confidentiality
is protected within an RDC such as the SWORDC by restricting
access and by vetting output taken out of the lab. The lab
is a locked facility staffed full-time by a Statistics Canada
Analyst. To use the centre, a project has to be proposed in
writing and approved. The researcher becomes, temporarily
(and unpaid!), a "deemed employee" of STC.
Virtually
any survey collected by Statistics Canada can be used in master
file format within a Research Data Centre, but there is particular
focus (and priority with respect to data documentation) on
a series of longitudinal surveys the agency has been compiling
for about a decade. In a longitudinal survey, respondents
are interviewed several times. In the Survey of Labour and
Income Dynamics, for example, a first set of respondents were
interviewed yearly between 1993 and 1998. In 1996, the sample
was "refreshed" with a new panel that was followed
until 2001. Thus for 1996, 1997 and 1998 the first and second
panels were deliberately overlapped. The sample continues
to rotate in this fashion, with the third panel entering the
sample in 1999 and continuing until 2004. The National Population
Health Survey, the National Longitudinal Survey of Children
and Youth, the Workplace and Employee Survey and the Youth
in Transition Surveys have similar sampling schemes.
These
longitudinal designs point the way to the social science of
the future, in terms of survey analysis. Cause and effect
issues that can only be guessed about with a cross-sectional
(i.e. one-shot) design begin to be resolved when the same
variable is tracked for the same respondent for two or more
occasions. With Research Data Centres, graduate students have
access to the same data that heretofore were only available
to seasoned faculty members enjoying a special relationship
with Statistics Canada.
The
presentation on October 3rd will describe all aspects of the
application process and details of how to do a project using
SWORDC facilities. With the conceptual advances due to longitudinal
data design comes the need for increasingly sophisticated
statistical techniques, and these shall be touched on briefly.
We shall also explore interest in a follow-up session covering
the analytical possibilities opened up by longitudinal surveys.
