Skip to main content


Log in

I feel good! Gender differences and reporting heterogeneity in self-assessed health

  • Original Paper
  • Published:
The European Journal of Health Economics Aims and scope Submit manuscript


For empirical analysis and policy-oriented recommendations, the precise measurement of individual health or well-being is essential. The difficulty is that the answer may depend on individual reporting behaviour. Moreover, if an individual’s health perception varies with certain attitudes of the respondent, reporting heterogeneity may lead to index or cut-point shifts of the health distribution, causing estimation problems. An index shift is a parallel shift in the thresholds of the underlying distribution of health categories. In contrast, a cut-point shift means that the relative position of the thresholds changes, implying different response behaviour. Our paper aims to detect how socioeconomic determinants and health experiences influence the individual valuation of health. We analyse the reporting behaviour of individuals on their self-assessed health status, a five-point categorical variable. Using German panel data, we control for observed heterogeneity in the categorical health variable as well as unobserved individual heterogeneity in the panel estimation. In the empirical analysis, we find strong evidence for cut-point shifts. Our estimation results show different impacts of socioeconomic and health-related variables on the five categories of self-assessed health. Moreover, the answering behaviour varies between female and male respondents, pointing to gender-specific perception and assessment of health. Hence, in case of reporting heterogeneity, using self-assessed measures in empirical studies may be misleading and the information needs to be handled with care.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2

Similar content being viewed by others


  1. Here, respondents are asked to rate hypothetical descriptions of a fixed level of a latent construct, such as responsiveness [12].

  2. Only variables on age and gender contain no missing values.

  3. Because foreigners are under-represented in the dataset, we only concentrate on German citizens.

  4. The wording of the corresponding question is “Do you have any prolonged health problems, diseases or disabilities?”.

  5. In their analysis, Lindeboom and van Doorslaer [9] use the health utility index (HUI3) in a quadratic specification as a proxy for the latent true health status.

  6. One assumption for the threshold parameters is that κ j  ≤ κ j+1, ∀j and that κ5 = ∞ and κ0 = −∞.

  7. Alternatively, κ1 could be set to zero. This would also identify the model.

  8. Greene and Henscher [22] discuss aspects of heterogeneity in ordered choices and present a detailed description of the generalized ordered probit model.

  9. The order condition in the generalized ordered probit model requires that the predicted probabilities are in the (0; 1) interval.

  10. The generalized ordered probit model nests the standard ordered probit model with the restriction that β1 = ⋯ = β J−1.

  11. A complete description of the procedure can be found in [25]. The related user-written Stata program regoprob2 is available at the SSC archive.

  12. The estimation with different imputations requires some caution with respect to the ‘averaging’ of the results (see [26]). For the total results, it follows that the coefficient vector of the multiple imputation analysis is given by the mean of the single estimations while for the variance–covariance estimate one has to distinguish between the within- and the between-imputation variance–covariance matrix.

  13. While our analysis helps to identify possible cut-point shifts, it remains unclear whether there is also an index shift. Following the approach by Hernández-Quevedo et al. [15], we introduced time dummies for the observation years. Only for men, the year 2008 has a strong significant negative effect.

  14. Alternatively, one can use the anchoring vignettes approach. In the SAVE dataset, no such information is included.


  1. Gerdtham, U., Johannesson, M.: New Estimates of the Demand for Health: Results Based on a Categorical Health Measure and Swedish Micro Data. Soc. Sci. Med. 49, 1325–1332 (1999)

    Article  PubMed  CAS  Google Scholar 

  2. Kenkel, D.S.: Should You Eat Breakfast? Estimates from Health Productions Functions. Health Econ. 4, 15–29 (1995)

    Article  PubMed  CAS  Google Scholar 

  3. Butler, J.S., Burkhauser, R.V., Mitchell, J.M., Pincus, T.: Measurement Error in Self-Reported Health Variables. Rev. Econ. Statist. 69, 644–650 (1987)

    Article  Google Scholar 

  4. Bound, J.: Self-Reported versus Objective Measures of Health in Retirement Models. J. Human Res. 26, 106–138 (1991)

    Article  Google Scholar 

  5. Disney, R., Wakefield, M.: Ill Health and Retirement in Britain: A Panel Data-based Analysis. J. Health Econ 25, 621–649 (2006)

    Article  PubMed  Google Scholar 

  6. Kerkhofs, M., Lindeboom, M.: Subjective Health Measures and State Dependent Reporting Errors. Health Econ. 4, 221–235 (1995)

    Article  PubMed  CAS  Google Scholar 

  7. Lindeboom, M., Kerkhofs, M.: Health and Work of the Elderly: Subjective Health Measures, Reporting Errors and the Endogenous Relationship Between Health and Work. J. Appl. Econometrics 24, 1024–1046 (2009)

    Article  Google Scholar 

  8. Ziebarth, N.R.: Measurement of Health, the Sensitivity of the Concentration Index, and Reporting Heterogeneity. Soc. Sci. Med. 71, 116–124 (2010)

    Article  PubMed  Google Scholar 

  9. Lindeboom, M., van Doorslaer, E.: Cut-Point Shift and Index Shift in Self-Reported Health. J. Health Econ 23, 1083–1099 (2004)

    Article  PubMed  Google Scholar 

  10. van Doorslaer, E., Jones, A.M.: Inequalities in Self-Reported Health: Validation of a New Approach to Measurement. J. Health Econ 22, 61–87 (2003)

    Article  PubMed  Google Scholar 

  11. Bago d’ Uva, T., Van Doorslaer, E., Lindeboom, M., O’Donnell, O.: Does Reporting Heterogeneity Bias the Measurement of Health Disparities? Health Econ. 17, 351–375 (2008)

    Article  Google Scholar 

  12. King, G., Murray, C.J.L., Salomon, J., Tandon, A.: Enhancing the validity and cross-cultural comparability of measurement in survey research. Amer. Polit. Sci. Rev. 98, 184–191 (2004)

    Google Scholar 

  13. Jürges, H.: True Health vs Response Styles: Exploring Cross-Country Differences in Self-Reported Health. Health Econ. 16, 163–178 (2007)

    Article  PubMed  Google Scholar 

  14. Etilé, F., Milcent, C.: Income-Related Reporting Heterogeneity in Self-Assessed Health: Evidence from France. Health Econ. 15, 965–981 (2006)

    Article  PubMed  Google Scholar 

  15. Hernández-Quevedo, C., Jones, A.M., Rice, N.: Reporting Bias and Heterogeneity in Self-Assessed Health: Evidence from the British Household Panel Survey. HEDG Working Paper 05/04, University of York (2005)

  16. Börsch-Supan, A., Coppola, M., Essig, L., Eymann, A., Schunk, D.: The German SAVE Study: Design and Results. MEA studies, Mannheim Research Institute for the Economics of Aging (2008)

    Google Scholar 

  17. Statistisches Bundesamt; Gesellschaft Sozialwissenschaftlicher Infrastruktureinrichtungen; Wissenschaftszentrum Berlin für Sozialforschung: Datenreport 2008 - ein Sozialbericht für die Bundesrepublik Deutschland. Berlin (2008)

  18. Little, R.J.A., Rubin, D.B.: Statistical analysis with missing data. Wiley, Hoboken, NJ (2002)

    Google Scholar 

  19. Murray, C.J., Tandon, A., Salomon, J.A., Mathers, C.D., Sadana, R.: Cross-Population Comparability of Evidence for Health Policy. In: Murray, C.J.L., Evans, D.B. (eds.) Health systems performance assessment: debates, methods and empiricism, pp. 705–713. World Health Organization, Geneva (2003)

    Google Scholar 

  20. Boes, S., Winkelmann, R.: Ordered Response Models. Allgemeines statistisches Arch./J. German Statist. Society 90, 167–181 (2006)

    Google Scholar 

  21. Long, J.S., et al.: Regression Models for Categorical and Limited Dependent Variables. Sage, Thousand Oaks (1997)

    Google Scholar 

  22. Greene, W.H., Hensher, D.A.: Modeling ordered choices. Cambridge University Press, Cambridge (2010)

    Google Scholar 

  23. Williams, R.: Generalized Ordered Logit/Partial Proportional Odds Models for Ordinal Dependent Variables. Stata J 6, 58–62 (2006)

    Google Scholar 

  24. Boes, S.: Three Essays on the Econometric Analysis of Discrete Dependent Variables. Dissertation, University of Zurich (2007)

  25. Pfarr, C., Schmid, A., Schneider, U.: Estimating ordered categorical variables using panel data: a generalized ordered probit model with an autofit procedure. Discussion Papers in Economics 02/10, Department of Law and Economcis, University of Bayreuth (2010)

  26. StataCorp.: Stata: Release 11. Statistical Software—Multiple Imputation. StataCorp LP, College Station (2009)

    Google Scholar 

Download references


The authors gratefully acknowledge useful comments from two anonymous referees, Andreas Schmid, Martin Schellhorn, Stefan Boes and the participants of the 2010 annual conference of the German Association of Health Economics (dggoe) in Berlin.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Udo Schneider.



See Tables 5, 6, 7.

Table 5 Random effects ordered probit estimates
Table 6 Random effects generalized ordered probit; men
Table 7 Random effects generalized ordered probit; women

Rights and permissions

Reprints and permissions

About this article

Cite this article

Schneider, U., Pfarr, C., Schneider, B.S. et al. I feel good! Gender differences and reporting heterogeneity in self-assessed health. Eur J Health Econ 13, 251–265 (2012).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


JEL Classification