Correcting for non-response bias using socio-economic register data Liisa Larja & Riku Salonen LFS Workshop, 17.-18. May, 2018, Reykjavik
The problem of the growing non-response 18 November 2018 Liisa Larja & Riku Salonen
Non-response is often correlated with the labour market status 18 November 2018 Liisa Larja & Riku Salonen
Effect of the bias: tertiary attainment in 30-34-years population 18 November 2018 Liisa Larja & Riku Salonen
Methodology: Seven different weighting schemes were constructed as follows: GREG_base: sex, 5-year age group and region (20 areas based on NUTS3) GREG_current: base + status in the unemployment register GREG1: base + level of education GREG2: base + origin GREG3: base + urban/rural GREG7: current + level of education GREG8: current + level of education + origins + urban/rural 18 November 2018 Liisa Larja & Riku Salonen
Calibraton process Population frame sample Age 15-24: 15 % 25-54: 50 % 55-74: 34 % Status in the job-seeker register Job-seeker: 7 % ILOSTAT: n/a Response set calibration to match the population frame Age 15-24: 13 % 25-54: 49 % 55-74: 37 % Status in the job-seeker register Job-seeker: 6 % ILOSTAT: Employed: 59% Unemployed 5 % Not in the labour force 36 % Estimates: Age 15-24: 15 % 25-54: 50 % 55-74: 34 % Status in the job-seeker register Job-seeker: 7 % ILOSTAT: Employed: 60 % Unemployed 6 % Not in the labour force 34 % 18 November 2018 Liisa Larja & Riku Salonen
18 November 2018 Liisa Larja & Riku Salonen
18 November 2018 Liisa Larja & Riku Salonen
18 November 2018 Liisa Larja & Riku Salonen
Discussion Using appropriate socio-economic auxiliary data in the estimation process may significantly improve estimation by correcting the bias caused by non-response and by improving the precision of the estimates. As compared to the models using only demographic auxiliary data, our results on estimates using auxiliary socio-economic data show bias as large as 1,5 percentage points in the employment rate. Further tests remain to be done income, student status, age of the youngest child? other indicators: NEET, working time, etc. subpopulations: men/women, youth, elderly, foreign-born, highly/least educated, etc What kinds of experiences do you have on using other than demographic auxiliary data? Ruotsalaisilta kannattaa pyytää ensin kommenttia, miksi he eivät ole ottaneet mukaan koulutusta kalibrointimuuttujaksi, vaikka ovat asiaa testanneet? tanskalaisilta, millaisia kokemuksia heillä on sen käytöstä? 18 November 2018 Liisa Larja & Riku Salonen