Presentation is loading. Please wait.

Presentation is loading. Please wait.

Synthetic estimators in Ireland Anthony Staines DCU.

Similar presentations


Presentation on theme: "Synthetic estimators in Ireland Anthony Staines DCU."— Presentation transcript:

1 Synthetic estimators in Ireland Anthony Staines DCU

2 What are synthetic estimators? Estimates of something you haven't got Estimates of something you haven't got Typically estimates for a small area of something Typically estimates for a small area of something Making maximum use of what you have Making maximum use of what you have

3 Example Lung cancer risk Lung cancer risk Smoking is a key explanation Smoking is a key explanation Suppose you want to study the geography of lung cancer Suppose you want to study the geography of lung cancer

4 What you have Smoking data from a national survey by age and sex Smoking data from a national survey by age and sex Small area level data on population and cancer incidence by age and sex Small area level data on population and cancer incidence by age and sex

5 What you can do at once Estimate prevalence for small areas included in the study Estimate prevalence for small areas included in the study Using the sample in the study Using the sample in the study

6 What's wrong with this? The areas you need may not be included The areas you need may not be included The estimates will be very imprecise The estimates will be very imprecise

7 You can do better In some obvious ways In some obvious ways And some not so obvious And some not so obvious

8 What you assume National age and sex specific rates apply in each small area National age and sex specific rates apply in each small area

9 And so From these you calculate small area specific prevalence estimates From these you calculate small area specific prevalence estimates This is indirect standardisation This is indirect standardisation Can be done smarter Can be done smarter requiring aggregation properties to hold requiring aggregation properties to hold Adding in area level covariates (urban/rural etc.) Adding in area level covariates (urban/rural etc.)

10 Can you do better? Yes Yes

11 How?

12 Model based estimators These have a long history These have a long history Many diverse applications Many diverse applications Combine survey data and some kind of 'census data' Combine survey data and some kind of 'census data' 'Census data' is that available for every area of interest 'Census data' is that available for every area of interest

13 Roughly Use the survey data to estimate relationships Use the survey data to estimate relationships at the relevant level at the relevant level between survey covariates between survey covariates and the census data and the census data

14 Then Assume the same relationship applies in the other areas Assume the same relationship applies in the other areas

15 Issues Modelling can be hard Modelling can be hard Remember these are predictive models, not explanatory models Remember these are predictive models, not explanatory models Data not easy to get at the right small area level Data not easy to get at the right small area level

16 Models models using individual level covariates only models using individual level covariates only models using area level covariates only models using area level covariates only models combining individual and area-level covariates models combining individual and area-level covariates

17 Limits Available data Available data Confidentiality Confidentiality Complexity of methods, esp. multi-level methods Complexity of methods, esp. multi-level methods Validation Validation

18 Spatial data limits Have to be able to link survey and census to the same set of small areas Have to be able to link survey and census to the same set of small areas Given the primitive systems in the UK and the nearly non-existent systems in the Republic this is a lot of work Given the primitive systems in the UK and the nearly non-existent systems in the Republic this is a lot of work Errors here will lead to biassed estimates Errors here will lead to biassed estimates

19 Confidentiality Need to respect confidentiality of survey respondents Need to respect confidentiality of survey respondents May limit the data available for these purposes May limit the data available for these purposes May need to design survey and survey consent process carefully to get good estimates May need to design survey and survey consent process carefully to get good estimates

20 Modelling Can become very complex Can become very complex Clustered survey designs Clustered survey designs Survey weights Survey weights Variable selection Variable selection Model diagnostics Model diagnostics

21 What and where to model Data may exist at many different geographies Data may exist at many different geographies Multi-level models with individual, household, local and regional effects can be considered Multi-level models with individual, household, local and regional effects can be considered GIS might be very useful here for data handling GIS might be very useful here for data handling Not advisable to aggregate covariates at different spatial levels Not advisable to aggregate covariates at different spatial levels This is just making a bad embedded synthetic estimator This is just making a bad embedded synthetic estimator

22 Validation Not easy to do, but essential Not easy to do, but essential How do you validate your synthetic estimates? How do you validate your synthetic estimates? Cross-validation? Cross-validation? Another survey? Another survey? ?

23 Options How about How about Health Atlas Ireland? Health Atlas Ireland? This is a system built for HSE, (led by Howard Johnson) to plan health services This is a system built for HSE, (led by Howard Johnson) to plan health services It already has It already has Maps Maps Census Census HIPE HIPE Mortality data Mortality data

24 Census output options Recently they have developed a very flexible census output system Recently they have developed a very flexible census output system Uses census data at ED level Uses census data at ED level Locations of houses Locations of houses Assumes that all the houses in a DED are exchangeable Assumes that all the houses in a DED are exchangeable

25 Census output options Allocates census data to any given area Allocates census data to any given area Directly weighted by using the number of households and the ED composition of the desired area Directly weighted by using the number of households and the ED composition of the desired area

26 Futures? Modern design of surveys Modern design of surveys Could readily be extended to do SA from almost any survey data where the necessary geographical data have bene collected Could readily be extended to do SA from almost any survey data where the necessary geographical data have bene collected Greatly improves value for money of large scale surveys Greatly improves value for money of large scale surveys


Download ppt "Synthetic estimators in Ireland Anthony Staines DCU."

Similar presentations


Ads by Google