The University of Auckland New Zealand Using Advanced Techniques with Existing Data The Work Programme of COMPASS Research Centre 6 th Wellington Colloquium Statistics NZ, Conference Room 3 August 2012
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
nd
Using Advanced Techniques with Existing Data
The Work Programme ofCOMPASS Research Centre
6th Wellington ColloquiumStatistics NZ, Conference Room3 August 2012
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndColloquium Outline
10.00 Introduction – Peter DavisContributing to Social ScienceContributing to Policy
10.45 Micro-simulation work programmeModelling the Early Life-CourseDr. Barry Milne, Research FellowPolicy Modelling and Demographic AgeingRoy Lay-Yee, Senior Research Fellow
12.15 LUNCH BREAK
13.00 Updating the NZSEI – Dr. Barry Milne
13.30 Health Services Research and Policy Evaluating Performance in the Public Hospital SectorProfessor Peter Davis (University of Auckland COMPASS)Dr. Barry Milne (University of Auckland, COMPASS)Dr. Jaikishan Desai (Victoria University Wellington, HSRC)
15.00 AFTERNOON TEA - CONCLUSION
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
nd
Core team members: two research fellows, two statisticians, and data managerExpertise in modelling and health services research
Survey/statistical work (NZ Election Survey)
Summer school
Data repository
COMPASS Team –Open for business!
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndPresentation Outline
Colloquium PurposeCommunicate! Collaborate! Transfer! Translate!
COMPASS Work Programme
Contributing to Social Science• Updating a socio-economic index• Managing research data• Summer school
Contributing to Policy• Lifting the performance of the health system• Promoting micro-simulation techniques
Colloquium
A N U P D A T E O F T H E N E W Z E A L A N D
S O C I O E C O N O M I C I N D E X ( N Z S E I )
B A R R Y M I L N E
B R I A N B Y U N , A L A N L E E , P E T E R D A V I S
Assessing socio-economic status through occupation
5
DISCLAIMER: Access to the data used in this study was provided by Statistics New Zealand under conditions designed to give effect to the security and confidentiality provisions of the Statistics Act 1975. The results presented in this study are the work of the author, not Statistics New Zealand.
Why measure SES?
Research Can test hypotheses about the impact of unequal distribution
of opportunities, advantages, resources and power on Health, wellbeing, life choices, use of services, crime
Moderating the impact of other risk factors
Can investigate SES stability and mobility, both within one’s life and inter-generationally
Describing populations
Funding allocation Social and health services are sometimes funded (in-part)
based on the socio-economic characteristics of the areas that they serve.
6
MANAGING RESEARCH DATAFOR PROFESSIONAL SOCIAL SCIENCE
PETER DAVISMARTIN VON RANDOWGERARD COTTERELL
RC33 CONFERENCE, THE UNIVERSITY OF SYDNEYTUESDAY, 10 JULY 2012
MANAGING RESEARCH DATA
www.compass.auckland.ac.nzwww.nzssds.org.nz
MANAGING RESEARCH DATA
Data access – preserving and making available research data sets & metadata
Research support – ‘Enhanced Publications’and related knowledge products
Teaching – hosting metadata, teaching data subsets and associated workbooks
Preserving and making available research data sets & metadata
DATA HOLDINGS & EXTERNAL USAGE
50 data sets archived including New Zealand Election Study data (1987–2008)– Professor Jack Vowles
International Social Survey Programme data for New Zealand (1991–2010) – Professor Philip Gendall
World Internet Project for New Zealand (2007 & 2009)– Professor Charles Crothers
Health data sets (adverse events, oral health care,primary care, sexual health) – Professor Peter Davis
150 registered users; 30 specific data requests Mainly re NZES, ISSP and oral health surveys
Promoting enhanced publications and related knowledge products
EXPLANATION
What are they Publications enriched with three types of information:
Research data (evidence of the research) Extra materials (to illustrate or clarify) Post‐publication data (commentaries, ranking)
Why are they useful Promote the availability of reusable scientific data Allow verification of the outcomes of the research Reduce the need to ‘re‐invent the wheel’(i.e. code and data are publicly available for re‐use)
Improving efficiency and underpinning transparency in research: enhancing research
resources via a research repository
Improving efficiency and underpinning transparency in research: enhancing research
resources via a research repository
Emma Gullery
COMPASS Research Centre and the Department of Statistics
End of Project Presentation
February 2012
Improving efficiency and underpinning transparency in researchEmma GulleryEmma Gullery
Project Objectives
Utilise the New Zealand Social Science Data Serviceby publishing online methods and data that have been analysed.
Supply well‐documented work for collaborationso other researchers do not have to replicate the same work.
Compile user‐friendly code available for download online.
Emma GulleryEmma Gullery Improving efficiency and underpinning transparency in research
Data & Filtering
National Minimum Data Set (NMDS) Records of hospital discharge data as far back as 1988. Analyses performed on the data to assess quality of healthcare,
provisions of services and government policy review. Includes both public and private hospital events. Filtered to make comparable across years.
Filtering Involves removing certain groups of observations for different reasons. MoH performs several steps to enable valid comparisons across time. Changes in code definitions and diagnosis criteria often affect subgroups.
Emma GulleryEmma Gullery Improving efficiency and underpinning transparency in research
Filtering Code
Emma GulleryEmma Gullery Improving efficiency and underpinning transparency in research
Annotations
Emma GulleryEmma Gullery Improving efficiency and underpinning transparency in research
nzssds.org.nz – Indicators
Emma GulleryEmma Gullery Improving efficiency and underpinning transparency in research
The nzssds.org website – In‐hospital mortality
Emma GulleryEmma Gullery Improving efficiency and underpinning transparency in research
Hosting metadata, teaching data subsets and associated workbooks
TEACHING
Workbooks and teaching data sets designed for self-directed learning of SPSS Three produced so far based on:
ISSP 2002 (family & gender roles); ISSP 2009(social inequality); NZES 2008 (election data)
Metadata used in teaching Illustrative study metadata used in survey class Five studies with varying survey characteristics
Survey data used in teaching Quantitative projects for research methods class ISSP used extensively for range of topics of interest
FUTURE DEVELOPMENTS
Data service functions1. Data saving and sharing
Qualitative data Links with librarians, science policy
2. Enhanced publications & knowledge products Encourage “open science” platform
3. Teaching data sets and workbooks Parallel suite for qualitative data Links with other departments, universities
New Zealand Social Statistics Network ‐ short courses in research methods
• Research methods courses offered since 2005• Courses
– Wellington February each year– Auckland – July
• Most courses are 5 days• For more information
– www.nzssn.org.nz– Contact: [email protected]
Proposed courses for Feb. 2013, Wellington
• INTRODUCTION TO STATISTICS• QUALITATIVE RESEARCH TECHNIQUES• CASE STUDY RESEARCH• APPLIED COMPUTER‐ASSISTED QUALITATIVE DATA ANALYSIS USING NVivo• INTRODUCTION TO STRUCTURAL EQUATION MODELLING USING AMOS™ or Mplus• INTRODUCTION TO SURVEY DESIGN• INTRODUCTION TO SOCIAL NETWORK RESEARCH AND ANALYSIS• FUNDAMENTALS OF SPSS• DATA ANALYSIS USING STATA• INTERMEDIATE STATISTICS• INTRODUCTION TO PROGRAM EVALUATION• APPLIED MULTIVARIATE USING STATA• MIXED METHODS IN SOCIAL RESEARCH• APPLIED STRUCTURAL EQUATION MODELING USING Mplus• ADVANCED QUALITATIVE DATA ANALYSIS USING NVivo 9
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndPresentation Outline
Colloquium PurposeCommunicate! Collaborate! Transfer! Translate!
COMPASS Work Programme
Contributing to Social Science• Updating a socio-economic index• Managing research data• Summer School
Contributing to Policy• Lifting the performance of the health system• Promoting micro-simulation techniques
Colloquium
Lifting the performance of New Zealand’s health system.
A research collaborative
26
The performance question
The performance question How do we apply evidence and science to our increasingly
expensive and strained health systems to maximise their fairness (equity), their efficiency, and their effectiveness?
Health System in Transition report on England (2011) – LSE Health/European Observatory
• Doubled nominal expenditure 1997-2008• Continuous structural innovation and transformation• 50,000 more doctors; 100,000 more nurses and midwives
Yet• Productivity did not increase• Health inequalities were not reduced
Although• Access to elective care improved, as did overall health status
27
28
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
nd
Introducing Microsimulation
WorkshopSydney, July 9, 2012
International Sociological Association – Research Committee RC33 on Logic and Methodology – 8th Conference on Social Science Methodology
Roy Lay-Yee and Barry Milne
University of Auckland, New Zealandand COMPASS Research Centrewww.compass.auckland.ac.nz
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndWhat is micro-simulation, how to do it
30
• Various features
• Computing platform
• Data integration
• Implementation
• Application
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndMicro-simulation method: Creating a virtual world
Start with a real sample of people. A cross-sectional sample of the NZ population over 65
Apply statistically-derived rules to reproduce patternsA sample of NZers over 65 with typical biographies over a year
We have created a virtual world (our simulation model)
Predict what might happen if conditions were to changeTest plausible scenarios in a society that is demographically ageing
JAVA / R tool for simulation, to assist policy decision making31
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndMicrosimulation: Relevance
to policy making
32
Provides an evidence base and a decision-support tool
Recognises importance of human agency (in social context)
Focuses on micro processes that generate macro structure→ pathways for policy intervention
Enables thought experiments, testing policy scenarios, making policy options explicit
It’s easier for policy makers to grasp and interpret (narratives, graphics)
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndInternational example 1:EUROMOD - Europe
33
• A static tax-benefit model for the European Union (2000s)
• Enables researchers and policy analysts to calculate, in a comparable manner, the effects of taxes and benefits on household incomes and work incentives for the population of each country and for the EU as a whole
Institute for Social & Economic Research University of Essex
https://www.iser.essex.ac.uk/euromod/
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndInternational example 2:LIFEPATHS - Canada
34
• Dynamic model of individuals and families from 1872 birth cohort to today• Creates synthetic life histories from birth to death that are representative of the
history of Canada’s population• Can be used to evaluate government programs, or to analyse societal issues of a
longitudinal nature, e.g. intergenerational equity
Statistics Canada
http://www.statcan.gc.ca/microsimulation/lifepaths/lifepaths-eng.htm
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndInternational example 3:
APPSIM - Australia
35
• Dynamic model simulates the life cycle of 200,000 individuals (1% sample of census) from 2001 to 2050
• Shows how the Australian population develops over time under various scenarios• Allows the social and fiscal impacts of policy changes over time to be simulated
National Centre for Social & Economic Modelling University of Canberra
http://www.natsem.canberra.edu.au/models/appsim/
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndData integration
36
• Data sources – quantitative, qualitative, findings from other studies, guesstimates
• Combining data
• Combining parameters
• Data → base file (initial conditions)
• Parameters (from statistical analyses) → simulation rules
Integration
PCASO: New Zealand and Australian data sources and model contributions
Study National Health Surveys
General Practice Survey
(Doctors)
National Health Survey
General Practice Survey
(Patient visits)
Country New Zealand New Zealand Australia New Zealand
Year 1996/7 (children)2002/3 (adults)
2001/2 1995 2001/2
Sample Children & adults Doctors (GP) Children & adults Patient visits
N 13,548 244 53,828 9,272
Model Component
Community Practitioner Morbidity; Community
Morbidity; Practitioner
37
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndApplication
38
• Test ‘what if’ policy scenarios
o Projection into the future; alternative settings
o Simulate impact of policy change (beforehand)
o Assumes everything else is the same
The
Uni
vers
ity o
f Auc
klan
dN
ew Z
eala
ndColloquium Outline
10.00 Introduction – Peter DavisContributing to Social ScienceContributing to Policy
10.45 Micro-simulation work programmeModelling the Early Life-CourseDr. Barry Milne, Research FellowPolicy Modelling and Demographic AgeingRoy Lay-Yee, Senior Research Fellow
12.15 LUNCH BREAK
13.00 Updating the NZSEI – Dr. Barry Milne
13.30 Health Services Research and Policy Evaluating Performance in the Public Hospital SectorProfessor Peter Davis (University of Auckland COMPASS)Dr. Barry Milne (University of Auckland, COMPASS)Dr. Jaikishan Desai (Victoria University Wellington, HSRC)
15.00 AFTERNOON TEA - CONCLUSION