Andrew M. Jones and Stefanie Schurer, "How Does Heterogeneity Shape the Socioeconomic Gradient in Health Satisfaction," Journal of Applied Econometrics, Vol. 26, No. 4, 2011, pp. 549-579. The data used in this paper were taken from the German Socio-Economic Panel (GSOEP), a national, representative longitudinal survey managed by the German Economic Research Institute (DIW Berlin). The version of this data-set was obtained with a license provided by the RWI Essen, where I was a graduate student at the Ruhr Graduate School in Economics. Enquiries concerning access and use of the GSOEP should be addressed directly to the DIW (http://www.diw.de/english/soep/26636.html) to the attention of Michaela Engelmann (soepmail@diw.de) or to Melody Reinicke at Cornell University, if a researcher works outside of Germany (CNEF@cornell.edu). The dataset has been prepared with the help of Panel Whiz. Access to Panel Whiz requires a license as well, which can be directly obtained from Prof. John Haisken-DeNew in exchange for a small donation to UNICEF (http://www.panelwhiz.eu/). His email is: john [AT] panelwhiz.eu. At the time of writing this article, the GSOEP had 22 waves available, running from 1984 to 2005, however, data is currently available until year 2007. In our sample we have 149,030 women and 134,626 men. The dataset in STATA format comprised 120MB. The main variable of interest is satisfaction with health (`swhealth') coded from 0 to 10 and recoded to a five point scale (1 to 5) in a variable called `swh'. This variable is called `p80x' in Panel Whiz. The analysis is conducted seperately for men (`male' == 1) and women (`female' == 1). Gender variables are constructed from a variable called `sex' in Panel Whiz. The regressors in the model are: (Dummy variables are indicated as DV that take the values 0,1: - East Germany: east (DV) - age groups: a1630, a3140, a4150, a5160, a6170, a71 (DV) (In Panel Whiz: `age') - Net annual household income (equivalised by OECD scale): income (In Panel Whiz: `h4612x'), equivalence scale is created from the three variables: `p3468x', `p3469x', `p3521x', `p3520x, which indicate: head of household, adults in household, children' which is interacted with all age-groups: ya1630, ya3140, ya4150, ya5160, ya6170, ya71 - Total years of education: yrsedu (In Panel Whiz: `p2292x') - Being unemployed: unemp (DV) (In Panel Whiz: `p171x') - Hours worked overtime: overtime (In Panel Whiz: `p2296x') - Being registered as disabled: wdisable (DV) (In Panel Whiz: `p499x') - Marital status: married, separate, single (DV) (In Panel Whiz: `p2291x') - Number of individuals in household: hhsize (In Panel Whiz: `p3468x') - Time dummy variables: y1984 y1985 y1986 y1987 y1988 y1989 y1990 y1991 y1992 y1993 y1994 y1995 y1996 y1997 y1998 y1999 y2000 y2001 y2002 y2003 y2004 y2005 (in Panel Whiz `year') All Stata .do files are ASCII files in DOS format. They are zipped in the file js-progs.zip. Unix/Linux users should use "unzip -a". To run the .do files, the following order should be considered: 0. create-GSOEP.do (Here you must use Panel Whiz first to create the dataset used in the analysis). 1. summary-stats.do 2. regression-OLOGIT.do 3. regression-pooled-logit.do 4. regression-RELOGIT.do 5. regression-CFEL.do 6. test-linearity.do 7. graphs-marginal-effects.do 8. graph-density-fixed-effect.do 9. graphs-marginal-effects-robustness.do Stefanie Schurer sschurer [AT] unimelb.edu.au