Further calculations on proxies and tracing rules 7-8 November 2018 Working group meeting "Statistics on Living Conditions"
Working group meeting "Statistics on Living Conditions" Introduction Should proxy interviews be allowed for subjective variables such as ‘well-being’ or personal variables such as ‘health’ or 'deprivation'? Regressions in order to evaluate to what degree proxies can affect the results for subjective variables 7-8 November 2018 Working group meeting "Statistics on Living Conditions"
Proxy: logistic regression Proxy as dependant variable Main variables: income, gender, crime, urbanisation, education, age Countries variable did not work out Results 7-8 November 2018 Working group meeting "Statistics on Living Conditions"
Proxy: logistic regression Dep variable: severe material deprivation Small basic model (gender, age, urbanization and proxy) Proxy is not significant in the model Results 7-8 November 2018 Working group meeting "Statistics on Living Conditions"
Proxy: logistic regression Last regression: another small model Dep variable: ph030 (health) (slightly recoded, (1),(2) -> (0) and (3) ->(1) Results: proxy has an effet 7-8 November 2018 Working group meeting "Statistics on Living Conditions"
Proxy: Severe material deprivation 6 logistic Regressions with PD20, PD30, PD50, PD60, PD70 and PD80 as dependent variables Proxy variable is significant in most models PD70: significant at 1,4% (highest p value, other pvalues below 0,001) Coefficients always slightly positive: it means that higher use of proxy will lead to more material deprivation 7-8 November 2018 Working group meeting "Statistics on Living Conditions"
Proxy material deprivation More details results 7-8 November 2018 Working group meeting "Statistics on Living Conditions"
Working group meeting "Statistics on Living Conditions" Conclusions We can conclude from this analysis that proxy interview influences the results of PH030 and personal deprivation variables 7-8 November 2018 Working group meeting "Statistics on Living Conditions"