Sensitivity of the population size estimates for different - PowerPoint PPT Presentation

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS Sensitivity of the population size estimates for different imputations of a covariate B.F.M. Bakker 2 P.G.M. van der Heijden 3 S.C. Gerritse Utrecht University 2 Statistics Netherlands/VU University 3 Utrecht University/University of Southampton

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS C APTURE R ECAPTURE

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS U SUAL R ESIDENTS Usual Residents

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS U SUAL R ESIDENTS Usual Residents Residing in the Netherlands for 1 year.

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS U SUAL R ESIDENTS Usual Residents Residing in the Netherlands for 1 year. How to get an estimate of Usual Residents?

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS U SUAL R ESIDENTS Usual Residents Residing in the Netherlands for 1 year. How to get an estimate of Usual Residents? Link Population Register (PR) to Crime Suspects Register (CSR) and Employment Register (ER)

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS U SUAL R ESIDENTS Usual Residents Residing in the Netherlands for 1 year. How to get an estimate of Usual Residents? Link Population Register (PR) to Crime Suspects Register (CSR) and Employment Register (ER) CSR has no usual residence →

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS U SUAL R ESIDENTS Usual Residents Residing in the Netherlands for 1 year. How to get an estimate of Usual Residents? Link Population Register (PR) to Crime Suspects Register (CSR) and Employment Register (ER) CSR has no usual residence → EM algorithm

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS U SUAL R ESIDENTS Usual Residents Residing in the Netherlands for 1 year. How to get an estimate of Usual Residents? Link Population Register (PR) to Crime Suspects Register (CSR) and Employment Register (ER) CSR has no usual residence → EM algorithm and Predictive Mean Matching

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS U SUAL R ESIDENTS Usual Residents Residing in the Netherlands for 1 year. How to get an estimate of Usual Residents? Link Population Register (PR) to Crime Suspects Register (CSR) and Employment Register (ER) CSR has no usual residence → EM algorithm and Predictive Mean Matching Estimate usual residence with Capture-Recapture

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS U SUAL R ESIDENTS Usual Residents Residing in the Netherlands for 1 year. How to get an estimate of Usual Residents? Link Population Register (PR) to Crime Suspects Register (CSR) and Employment Register (ER) CSR has no usual residence → EM algorithm and Predictive Mean Matching Estimate usual residence with Capture-Recapture

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS R ESEARCH Q UESTION Will different methods of handling missing data affect the population size estimate under capture recapture? And if so: How?

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS S CENARIO ’ S Three scenario’s 1 Maximal model for EM algorithm

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS S CENARIO ’ S Three scenario’s 1 Maximal model for EM algorithm 2 Restrictive model for EM algorithm

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS S CENARIO ’ S Three scenario’s 1 Maximal model for EM algorithm 2 Restrictive model for EM algorithm 3. Predictive Mean Matching (PMM) imputation

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS E XPECTATIONS Based on previous research and expert knowledge within CBS:

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS E XPECTATIONS Based on previous research and expert knowledge within CBS: Usual residents between: 150.000 - 200.000

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS E XPECTATIONS Based on previous research and expert knowledge within CBS: Usual residents between: 150.000 - 200.000 Population wise:

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS E XPECTATIONS Based on previous research and expert knowledge within CBS: Usual residents between: 150.000 - 200.000 Population wise: CSR missing values very specific, will only resemble subset of the data

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS R ESULTS - T OTAL Results

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS R ESULTS - O THER W EST Results

C APTURE -R ECAPTURE U SUAL R ESIDENTS S CENARIO ’ S E XPECTATIONS R ESULTS Thank you! Email: s.c.gerritse@uu.nl, sc.gerritse@cbs.nl Website: http://scgerritse.wordpress.com/