Correlational finding on Happiness and Self employed
Subject code: E02ae03

StudyAndersson (2008): study SE 1991
TitleHappiness and Health: Well-Being among the Self-Employed.
SourceThe Journal of Socio-Economics, 2008, Vol. 37, 213 - 236
DOIDOI:10.1016/j.socec.2007.03.003
Public18+aged, working people, Sweden, followed 9 years, 1991-2000
SampleSelection from probability samples
Non-Response
Respondents N =1998

Correlate
Author's labelSelf-employment
Page in Source 221,227,233
Our classificationSelf employed, code E02ae03
Operationalization
1: self-employed 
0: wage-earners
Remarks
Assessed at T1 (1991) and T2 (2000)

Observed Relation with Happiness
Happiness
Measure
StatisticsElaboration/Remarks
M-FH-g-sq-v-3-eD%=+ ns
CURRENT employemnt by CURRENT happiness
1991 % happy
-Self-employed  64.1%
-Wage-earners   61.4%
difference      +2.7%
M-FH-g-sq-v-3-eD%=+ p < .01
2000 % happy
-Self-employed 70.5%
-Wage-earners  59.1%
difference     +11.4%
M-FH-g-sq-v-3-eLRCD=+.34 p < .01
LRCD controled for time dummy (2000 vs 1991).
M-FH-g-sq-v-3-eLRCD=+.38 p < .01
LRCD additionally controled for:
-age
-gender
-place of birth
-education
-marital status
-place of residence 
-having children at home
M-FH-g-sq-v-3-eb=+.09 p < .01
b controled for:
-age
-gender
-place of birth
-education
-marital status
-children
-place of residence
M-FH-g-sq-v-3-eD%=+ ns
T1-T2 CHANGE employment by T1-T2 CHANGE happiness
Self-employed both in T1 and T2
-T1(1991):  63.7%
-T2(2000):  69.0%
difference  +5.3%
M-FH-g-sq-v-3-eD%=+ ns
Wage-earner in T1 and self-employed in T2
-T1(1991):  67.0%
-T2(2000):  72.3%
difference  +5.3%
M-FH-g-sq-v-3-eD%=+ ns
Wage-earner both in T1 and T2
-T1(1991):  61.1%
-T2(2000):  59.2%
difference  -1.9%
M-FH-g-sq-v-3-eD%=+ ns
Self-employed in T1 and wage-earner in T2
-T1(1991):  66.7%
-T2(2000):  55.6%
difference  -11.1%
M-FH-g-sq-v-3-eb-fix=+.34 ns
b-fix not controled
M-FH-g-sq-v-3-eb-fix=+.35 ns
b-fix controled (see below)

These fixed effects are not negative, which means 
that the higher happiness of the self-employed 
does not result from self-selection of the happy 
into self-employment
M-FH-g-sq-v-3-eb=+.09 p < .01
b and b-fix controled for:
-age
-gender
-education
-marital status
-having children at homechildren
-place or residence

Negative year dummy (b = -.04) indictes that 
wage-employed became less happy from 1991 to 200


Appendix 1: Happiness measures used
CodeFull Text
M-FH-g-sq-v-3-eSelfreport on single question:

Do you generally feel that your life is a source of personal satisfaction?
1 no, never
2 yes, sometimes
3 yes, most of the time


Appendix 2: Statistics used
SymbolExplanation
bREGRESSION COEFFICIENT (non-standardized) by LEAST SQUARES (OLS)
Type: test statistic
Measurement level: Correlate: metric, Happiness: metric
Theoretical range: unlimited

Meaning:
b > 0 A higher correlate level corresponds with a higher happiness rating on average.
B < 0 A higher correlate level corresponds with a lower happiness rating on average.
B = 0 Not any correlation with the relevant correlate.
b-fixREGRESSION COEFICIENT in fixed effects analysis
Type: test statistic
WDH symbol: b-fix
Primary correlate level: metric
Secondary correlate level: nonmetric
Happiness level: metric
Theoretical range: unlimited
Meaning:
Variant of usual (non-standardized) regression coefficient (b), which controls for the secondary variables, by focusing on differences from a fixed level, such as the mean in a category. Aims to reduce the residual variance and to improve the precision of the regression coefficient.

References:
Wikipedia: http://en.wikipedia.org/wiki/Fixed_effects_model
Non-technical text: http://www.jblumenstock.com/files/courses/econ174/FEModels.pdf
D%DIFFERENCE in PERCENTAGES
Type: descriptive statistic only.
Measurement level: Correlate level: dichotomous, but nominal or ordinal theoretically possible as well. Happiness level: dichotomous
Range: [-100; +100]

Meaning: the difference of the percentages happy people at two correlate levels.
LRCDLOGIT REGRESSION COEFFICIENT DICHOTOMOUS
Statistic: logit regression coefficient
Measurement level: Correlate: metric, Happiness: dichotomous
Theoretical range: unlimited

LRCD > 0 A higher level corresponds to the probability of a higher happiness rating.
LRCD< 0 A higher level corresponds to the probability of a lower happiness rating.
LRCD = 0 Not any correlation with the relevant correlate.

Remarks:
Signs and significance of coefficients estimated by logit estimates are comparable to REGRESSION COEFFICIENTS estimated by LEAST SQUARES (OLS). Relative magnitudes of the regressor to those of other regressors are also comparable to Ordinary Least squares Estimations. Magnitudes of coefficients are roughly comparable to OLS if divided by 4 (Wooldridge, Introductory Econometrics, 561-562, 2002)
Source:
Ruut Veenhoven, World Database of Happiness, Collection of Correlational Findings, Erasmus University Rotterdam.
https://worlddatabaseofhappiness.eur.nl