Population and
Sustainable Development
newzealand.govt.nz
www.stats.govt.nz
Phase 2 - defining the problem
Home
Site objective
Introduction to population
Information by subject
Tools and resources
Eight-step checklist
Policy development cycle
Interpreting data
Designing statistical tables
Public policy can change demographics
Framework for analysing population structure
New trends in topical population issues
Getting the bigger picture - three common misunderstandings
Glossary
Myth busters
Population faq
Contact us
About us
Ask a question
Help
Search criteria
Search
Feedback
What do you think?
If you've got feedback we'd love to hear it
Send Feedback
Statistics New Zealand
Department of Labour
Ministry of Economic Development
Ministry of Social Development
Go to the newzealand.govt.nz website
Tools and resources
>
Policy development cycle
>
Phase 1 - strategic phase
Phase 2 - defining the problem
Phase 3 - developing and analysing alternatives
Phase 4 - presenting recommendations to decision-makers
Phase 5 - service delivery
Phase 6 - evaluation
Can official statistics and administrative data from different government agencies help with problem definition?
What are the relative strengths and weaknesses of this data?
What are the features and limitations of the statistics we are working with, in the context of the specific application they are being used for?
Do we need to consult a statistics expert?
Are we combining statistics from different sources?
If so, ensure the definitions are consistent, or that differences have been understood and allowed for. For instance, the definition and quality of data for some variables change over time within particular collections, while coverage, sampling errors and definitions vary between various administrative statistical series.
If the demographic statistics currently available are deficient in some respect, do we need to consider possible means of accessing additional statistics, or recommending system changes to collect or derive better statistics in the future?
Have we developed a profile of the characteristics of different sub-groups to assist with targeting?
If there is a serious lack of population data, what qualitative information is available that throws some light on the characteristics of policy groups or service recipients?
What is the ‘fit’ between what we really want to measure, and what available statistics allow us to measure? For example, only incidence or prevalence rates may be available but what we really want are statistics on severity.
Do we need to analyse trend data carefully to ensure the trends are real and not statistical artefacts?
For instance, has the definition of the statistical series or the way statistics are collected changed?
How sensitive is the trend to the choice of base year or period, or to the definition of the cohort?
Are there possible physical limits to the continuation of recent trends (eg of population growth in a region)?
Can we identify possible constraints on the continuation of recent trends for a sub-group derived from the experience of similar groups where trends have reached a plateau?
Do we need to allow for greater uncertainty when projecting small-area populations, or small populations with high growth rates?
Have we standardised the key measures being used or the underlying base populations appropriately?
When comparing statistics for different groups, are we using specified cases only?
For instance, a common error is to compute statistics for the Māori population, and then to label the total population minus the Māori population as non-Māori, when a significant proportion of that population is actually not specified, some of whom will be of Māori ethnicity.
What do we know about the characteristics of the non-respondents? Compare subgroups in a population with the whole population who specified the appropriate characteristic rather than using the non-subgroup as a contrast.
Are appropriate broad outcome targets set for key policy and service populations in the context of sector goals?
Do we need to analyse the data from different perspectives, for instance, using both cross-section and cohort analysis?
Have we exploited the full range of supporting contextual information in interpreting demographic statistics?
Do we need to develop alternative scenarios for different population/sub-group dynamics? Are alternative demographic scenarios logically and internally consistent?
Do we need to check the output of any modelling of demographic variables against independent data?