Powerpoint Slides from Most Recent RDC Information Session for

Report
Statistics Canada
Research Data
Centre Program*
Facilities across Canada housing detailed
confidential microdata and documentation
files from Statistics Canada
Statistics Canada released data that would
otherwise not be available into “secure” sites.
1
About statistics Canada data:
•
•
•
Canadian Community Health Survey (CCHS)
Ethnic Diversity Survey (EDS)
General Social Survey (GSS selected cycles)
–
–
–
–
–
–
–
–
•
•
•
•
•
•
Access to and Use of Information Communication Technology
Education, Work and Retirement
Family
Health
Social Engagement
Social Support and Aging
Time Use
Victimization
Longitudinal Survey of Immigrants to Canada (LSIC)
National Longitudinal Survey of Children and Youth (NLSCY)
National Population Health Survey (NPHS)
Survey of Labour and Income Dynamics (SLID)
Workplace and Employee Survey (WES)
Youth in Transition Survey and the Programme for International Student
Assessments (YITS-PISA)
2
Two ways researchers can get access to
Statistics Canada Survey data:
Research Data Centre
• Access restricted
• Must work with data
within centre
• Results can only be
released from centre
after scrutiny
• Must submit proposal,
be “approved,” get
security clearance,
sign contract
Data Liberation Initative
• Actual data released to
researchers with
minimal restrictions
• Data can be
downloaded via library
website
3
Two sources of StatCan data:
Data liberation initiative:
- At just about every university and many colleges
across Canada
Research Data Centres
- full-time centres at larger Canadian universities
(UBC, Alberta, Calgary, Toronto, Western,
Waterloo, Montreal, etc.)
- Part-time centres at other universities, including
UVic, SFU, Queen’s, Saskatchewan
- No centre at smaller institutions (e.g., Vancouver
4
Island University, UNBC)
DLI Restrictions
• No longitudinal data (in some cases,
cross-sectional waves, not linked and with
unique identifiers stripped, are available,
but in other cases survey not available at
all)
• Many variables treated as “confidential”
and deleted from dataset or coarsely
categorized
5
Files
available
RDC
DLI
-Those listed on RDC
site
-Other files if
arragements can be
made
- Those listed on DLI site –
see UVic library web page
under Data Acquisition
Files not
available
Information
on files not
avail.
Any linked longitudinal
dataset
Recent waves of NCLSY
Some newly available
surveys
Cluster numbers
Cluster numbers,
Geographic detail
Demographic detail
Other
6
RDC
DLI
Who may
access
- Faculty with approved
projects
-Graduate students with
approved projects
(+faculty co-investigator)
- Any member of UVic
community with NetLink ID
Where files
may be
worked on
In Data Centre only
May be downloaded to be
used anywhere, with
agreement not to
redistributed
Initial contact Doug Baer or Lee
Kathleen Matthews, Data
Grenon (StatCan Analyst Librarian, UVic
located in Vancouver at
UBC)
7
censored
variables
Full versions of
datasets with
censored
variables +
datasets not
otherwise
available can be
worked on in a
“Research Data
8
Centre”
Full versions of datasets with censored variables +
datasets not otherwise available can be worked on
in a “Research Data Centre”
As part of the application process, you will be asked to indicate
why the DLI datasets are not sufficient.
- Dataset not available at all via DLI (some longitudinal
datasets)
- Linked longitudinal file for the dataset not available via
DLI
- Variables needed for the research are either (a) not
available or (b) too coarsely categorized [not so with
originally collected data] to be usable for research
project
Examples: exact income, exact age, city size,
community in which respondent lives (esp. for
multi-level analysis), religion, church attendance
(some surveys),
9
Statistics Canada Research
Data Centre Program
There are RDCs across Canada at most major
universities with doctoral programs:
• New Brunswick (branch at Moncton), Dalhousie
• Toronto , York
•Waterloo branch at Guelph ( WLU participates)
•McMaster (Brock participates)
•Western (branch at Windsor)
•Queen’s (part-time site)
•U of Ottawa (Carleton and UQ-Gatineau participate)
•Manitoba
•U of Saskatchewan (part-time site)
• 2 Alberta sites: U of Alberta; Calgary (branch at Lethbridge)
• Manitoba
•Consortium (U de Montreal) with branches at UQAM,
Sherbrooke, Laval
•McGill
•BC universities consortium
BC consortium: UBC, SFU, UVic, Vancouver
Island Univ., UNBC
10
The UVic branch works within the British Columbia
Interuniversity Research Data Centre network
• “main” site is at UBC; open 9-5 M-F
• UVic site has more restrictive hours (arranged termby-term in consultation with researchers).
– Currently 15.5 hours/week (sometimes a bit less in summer)
– Exact hours worked out in consultation with users
11
Data used most frequently at RDCs
survey of labor income dynamics (SLID)
Victimization GSS cycle 3, 8, 13, 18
Participation and Activity Limitation
Family history GSS cycle 5, 10, 15 and 20
homicide data-pilot
national graduate survey / Follow-up Survey of 2000 Graduates
Canadian Health survey 1.2
Canadian Health survey 2.2
Canadian Health survey .1
12
Major Statcan surveys:
(ALL VERY WELL SUPPORTED AT RDCs)
• Workplace and Employment Survey
• Canadian Community Health Survey
• Health Services Access Survey
• General Social Survey
Longitudinal:
• National Population Health Survey
• Survey of Labour & Income Dynamics
• National Longitudinal Survey of Children and Youth
• Longitudinal Survey of Immigrants to Canada
• Youth in Transition Survey
• Workplace & Employment Survey
Census (presently 1991,1996,2001,2006)
13
& other data can be arranged
• There is presently a project involving BC
Administrative Health data (to be linked to
Stats Can survey data)
• For a very large list of StatCan Surveys,
see the DLI website (UVic library)
http://gateway.uvic.ca/data/default.html
 click on “DLI collection”
future plans: see below
14
What is the process for gaining access?
http://www.statcan.ca/english/rdc/application.htm
15
Application process works through SSHRC
Graduate students must have faculty member as coinvestigator
16
Project proposal
• Proposal evaluation by SSHRC peer review and
Statistics Canada
• Very few are turned down… though must
establish that confidential data are required to
complete project
– Does project have scientific merit? is access to
confidential microdata necessary? Does researcher
have expertise to conduct research?
– Takes 5-8 weeks
• Proposals that are part of SSHRC or CIHR
grants forgo the SSHRC peer review process
– Approvals typically 3-4 weeks
17
Process:
•
•
•
•
Submit proposal
Proposal approved
Security check on applicant
oath, investigator becomes “deemed
employee” of statistics canada
• Orientation session at UVic
• Issued access card for card reader
18
UVic facilities:
6 workstation lab
with room for
expansion to up to
10 workstations
workstations now have
widescreen monitors or
dual screen
configuration
Server for data
Most commonly
used statistical
software packages
Some highly
specialized software
packages
Hours are worked out to suit the needs of
active researchers. Currently: 16
hrs/week (slightly reduced Feb 2011 due
to staff transition)
We try to be open at least 4 hours on 3
different days of the week.
19
Software
Standard stats packages: SPSS (19), SAS (9.2) STATA
(11) [Stata/SE on 2 machines)
Open-source stats: R
Multilevel models: HLM, LISREL, MPlus
SEM models: LISREL, MPlus
Specialized (Bayesian, MCMC etc.): WinBugs
Other software can
be obtained if
demand exists.
20
Security process
• No output or notes can be taken out of the room
• Users have file drawers and access to printer
inside the centre
• Output listings and notes (if typed into a
computer file) can be released after they are
“vetted” by a Statistics Canada Analyst at the
main BC site
•Files are sent via high
security network to
Vancouver
•Files that are approved
for release are emailed
back to researcher
•Pass card works only during
centre hours (swipe in, swipe
out protocol)
21
Can I work at other RDCs too?
Can I work with other researchers?
What about other researchers at other universities?
• Access is “network wide”
• Files are stored on a “project” basis
(researchers, RAs, etc. have own account but
access to shared files)
• UVic researchers are part of the BC consortium
and could go to the UBC site if more intense
periods of research are required (35 hrs/week
vs. 15); project files can be sent to and from the
branch via the security intranet
22
Preparation:
• Check to see if dataset is one of standard RDC datasets:
check the national RDC website (see handout)
– Extensive data documentation provided for listed datasets
– If what you are interested is not on the list, check with Doug Baer
• Is a public use file available? Check with Kathleen
Matthews [email protected] or on library web site.
• Verify that variables needed for research are not on public use file.
If possible, use public use file to explore data, etc.
• If further dataset documentation required, ask Doug Baer
• Go to SSHRC web page to put together application. Don’t hesitate
to consult Doug Baer for help. Be prepared to specify variables to
be used. Where a public use version of the dataset is available, be
prepared to make clear why RDC access is needed (e.g., “a needed
variable is suppressed on the public use file”).
23
Statistics Training
• Summer Institutes:
– SPIDA (York University)
– ICPSR (U Michigan)
– University of Western Ontario
– Population Health BC
• Likely workshop on Structural Equation Models
June 2011
– Seminar at the Congress for the Humanities &
Social Sciences (none this year)
• Special workshops and seminars (Baer):
– Possible: Statistics Canada bootstrap weights24
Contact information:
More numbers on
handout
Doug Baer, Academic Director
(Sociology)
[email protected]
(721) – 7581
Cornett, A365
RDC (853) 3196 ([email protected])
RDC Analyst at UBC: Geoff 604-822-0263
([email protected])
Centre web site (shows hours): web.uvic.ca/rdc
25
Future:
• Plans are in development to add the following to
RDC dataset collection:
–
–
–
–
Cancer Registry (pilot project in progress at BCIRDC)
HRSDC administrative data
CPP-disability data
Homicide data (Cdn. Centre for Justice Statistics)
[under review: pilots only]
– Census
– Business data: (selected datasets from Small
Business & Special Surveys Division)
26

similar documents