Advanced search
1 file | 108.44 KB

Additional weighting for education affects estimates from a National Health Interview Survey

Author
Organization
Abstract
National Health Interview Surveys are used to produce country-wide results for a substantial number of health indicators. However, if some educational groups are underrepresented in the sample, estimates may be biased. This study investigated the impact of the use of post-stratification weights that adjust for the population distribution by education on estimates from the Belgian Health Interview Survey 2013. For 25 health-related indicators that match the European Core Health Indicator shortlist, estimates were computed using two different sets of post-stratification weights: one based on age group, gender and province only and the other one including also education. The Census 2011 was used as auxiliary data source. Statistical differences between the two estimates were assessed with the Delta method. If education is not included as post-stratification weighting factor, low educational groups (ISCED 0-2) represent 31.1% of the total study population aged 25 years and older. If education is taken into account this proportion rises to 40.3%. The use of post-stratification weights adjusting for the population distribution by education has an impact on several survey estimates. The most pronounced effect is an increase in the estimated proportion of people with diabetes (+0.73%; 95% CI 0.19-1.27; relative increase +11.6%), asthma (+0.52%; 95% CI, 0.06-0.98; relative increase +12.4%) and difficulties to cover their health expenses (+2.31%; 95% CI, 1.52-3.10; relative increase +9.4%). Including education in the calculation of post-stratification weights reduces bias due to educational differences in survey participation. Auxiliary information used to calculate post-stratification weights for national health surveys should include education.
Keywords
SOCIOECONOMIC DIFFERENCES, POST-STRATIFICATION, EUROPEAN-UNION, NONRESPONSE

Downloads

  • (...).pdf
    • full text
    • |
    • UGent only
    • |
    • PDF
    • |
    • 108.44 KB

Citation

Please use this url to cite or link to this publication:

Chicago
Van Der Heyden, Johan, Dirk De Bacquer, Lydia Gisle, Stefaan Demarest, Rana Charafeddine, Sabine Drieskens, Jean Tafforeau, Herman Van Oyen, and Koen Van Herck. 2017. “Additional Weighting for Education Affects Estimates from a National Health Interview Survey.” European Journal of Public Health 27 (5): 892–897.
APA
Van Der Heyden, J., De Bacquer, D., Gisle, L., Demarest, S., Charafeddine, R., Drieskens, S., Tafforeau, J., et al. (2017). Additional weighting for education affects estimates from a National Health Interview Survey. EUROPEAN JOURNAL OF PUBLIC HEALTH, 27(5), 892–897.
Vancouver
1.
Van Der Heyden J, De Bacquer D, Gisle L, Demarest S, Charafeddine R, Drieskens S, et al. Additional weighting for education affects estimates from a National Health Interview Survey. EUROPEAN JOURNAL OF PUBLIC HEALTH. 2017;27(5):892–7.
MLA
Van Der Heyden, Johan, Dirk De Bacquer, Lydia Gisle, et al. “Additional Weighting for Education Affects Estimates from a National Health Interview Survey.” EUROPEAN JOURNAL OF PUBLIC HEALTH 27.5 (2017): 892–897. Print.
@article{8533876,
  abstract     = {National Health Interview Surveys are used to produce country-wide results for a substantial number of health indicators. However, if some educational groups are underrepresented in the sample, estimates may be biased. This study investigated the impact of the use of post-stratification weights that adjust for the population distribution by education on estimates from the Belgian Health Interview Survey 2013. 
For 25 health-related indicators that match the European Core Health Indicator shortlist, estimates were computed using two different sets of post-stratification weights: one based on age group, gender and province only and the other one including also education. The Census 2011 was used as auxiliary data source. Statistical differences between the two estimates were assessed with the Delta method. 
If education is not included as post-stratification weighting factor, low educational groups (ISCED 0-2) represent 31.1\% of the total study population aged 25 years and older. If education is taken into account this proportion rises to 40.3\%. The use of post-stratification weights adjusting for the population distribution by education has an impact on several survey estimates. The most pronounced effect is an increase in the estimated proportion of people with diabetes (+0.73\%; 95\% CI 0.19-1.27; relative increase +11.6\%), asthma (+0.52\%; 95\% CI, 0.06-0.98; relative increase +12.4\%) and difficulties to cover their health expenses (+2.31\%; 95\% CI, 1.52-3.10; relative increase +9.4\%). 
Including education in the calculation of post-stratification weights reduces bias due to educational differences in survey participation. Auxiliary information used to calculate post-stratification weights for national health surveys should include education.},
  author       = {Van Der Heyden, Johan and De Bacquer, Dirk and Gisle, Lydia and Demarest, Stefaan and Charafeddine, Rana and Drieskens, Sabine and Tafforeau, Jean and Van Oyen, Herman and Van Herck, Koen},
  issn         = {1101-1262},
  journal      = {EUROPEAN JOURNAL OF PUBLIC HEALTH},
  language     = {eng},
  number       = {5},
  pages        = {892--897},
  title        = {Additional weighting for education affects estimates from a National Health Interview Survey},
  url          = {http://dx.doi.org/10.1093/eurpub/ckx005},
  volume       = {27},
  year         = {2017},
}

Altmetric
View in Altmetric
Web of Science
Times cited: