A Web Survey Analysis of the Subjective Well-being of Spanish Workers Martin Guzi Masaryk University Pablo de Pedraza Universidad de Salamanca APPLIED ECONOMICS MEETING 2014
Frey and Stutzer (2010) state that the research of subjective well-being (SWB) should remain open to constructing different indicators for different aspects of life. In this paper we confirm that job characteristics and working conditions constitute significant determinants of SWB.
Empirical evidence Oswald (2002) uses 1996 Eurobarometer surveys to show that occupation, hours of work, job security, trade union affiliation and commuting time to work have impact on job satisfaction in European countries. Drobnic, Beham and Prag (2010) use 2003 European Quality of Life Survey and discover that the impact of working conditions on life satisfaction is stronger in Southern and Eastern European countries relative to Western European countries.
Research in Spain In 1999, Spanish Ministry of Labor and Social Affairs initiated the Quality of Working Life Survey (ECTV) Namkee (2007) uses 1999-2004 ECVT to show the effect of work flexibility, independence, social contacts, trust in superiors, a pleasant and low-stress work on workers satisfaction with life and job. Burón (2009) uses 2004 ECTV to identify the negative effect of long-working hours, long commutes, overqualification, and previous unemployment experience on job satisfaction.
Characteristics of online data We exploit data obtained through a Continuous Voluntary Web Survey (CVWS). Web survey is a very attractive tool to access respondents at lower cost and larger numbers. The resulting sample is not obtained randomly, the coverage is lower among groups without access to the Internet (coverage error) and the sample only includes web visitors who respond to the questionnaire (non-response error).
Wage Indicator data Web survey is posted at www.tusalario.es. Every web-visitor is invited to complete a websurvey, which takes 10 to 20 minutes. The survey has detailed questions about earnings, working conditions and employment contracts, education, occupation, and household characteristics. Most importantly, the survey includes satisfaction questions in different domains such as life, job, and the combination of family and work.
Aim of this research First, we demonstrate the role of working conditions in the different domains of SWB. Second, we contribute to the debate on web survey data quality, reliability, and validity for scientific use.
The comparison of Spanish online and traditional surveys Table 1 The number of observations in the samples Year ESS WI 2005 642 5508 2006 564 3142 2007 206 5962 2008 1038 3249 2009 933 2010 847 2011 995 454 Total 3445 20095 Source: The European Social Survey 2005-2011, Wage Indicator 2005-2011. Note: The sample is limited to employed individuals aged 18-65 years.
ESS WI WI with PSA (1) (3) (2) mean s.dev mean s.dev mean Life-satisfaction 7.36 1.76 6.79 1.80 6.77 Female 0.48 0.50 0.44 0.50 0.40 Edu: Primary 0.51 0.50 0.21 0.40 0.43 Edu: Secondary 0.26 0.44 0.25 0.43 0.24 Edu: Tertiary 0.23 0.42 0.54 0.50 0.33 Single 0.31 0.46 0.53 0.50 0.38 Married 0.61 0.49 0.42 0.49 0.55 Divorced 0.07 0.25 0.04 0.20 0.06 Widowed 0.02 0.13 0.00 0.06 0.01 Age 18-24 0.10 0.30 0.08 0.27 0.11 Age 25-34 0.25 0.43 0.53 0.50 0.25 Age 35-44 0.28 0.45 0.27 0.45 0.28 Age 45-54 0.23 0.42 0.10 0.31 0.25 Age 55-64 0.15 0.35 0.02 0.13 0.11 Health: excellent 0.20 0.40 0.28 0.45 0.31 Health: good 0.51 0.50 0.33 0.47 0.29 Health: poor 0.24 0.43 0.24 0.42 0.23 Health: very poor 0.06 0.23 0.15 0.35 0.16 Foreign-born 0.07 0.25 0.07 0.26 0.06 Self-employed 0.15 0.36 0.01 0.07 0.00 Log income 7.45 0.58 7.37 0.58 7.32
The quality of online data We test the data quality of the WageIndicator sample against European Social Survey (ESS). Respondents in the WI sample report a substantially lower satisfaction with life. Comparison reveals that highly educated and younger participants are overrepresented in the WI sample. The shares of females and foreignborn are equal in both samples. We adjust WI sample with weights based on gender, age, education and Spanish regions.
ESS WI (1) (3) (2) WI with PSA Female 0.045 0.056 ** 0.067 Edu: Secondary -0.111 0.077 ** -0.022 Edu: Tertiary 0.005 0.163 *** 0.077 Married 0.569 *** 0.476 *** 0.473 *** Divorced -0.235 * 0.105 * 0.173 Widowed -0.329-0.069 0.155 Age 15-24 ref. ref. ref. Age 25-34 -0.178-0.152 *** -0.206 ** Age 35-44 -0.351 *** -0.428 *** -0.394 *** Age 45-54 -0.433 *** -0.388 *** -0.414 *** Age 55-64 -0.222-0.164-0.158 Health excellent ref. ref. ref. Health good -0.361 *** -0.507 *** -0.491 *** Health poor -0.758 *** -1.11 *** -1.127 *** Health very poor -1.338 *** -1.789 *** -1.769 *** Foreign-born -0.073 0.042 0.046 Self-employed -0.034 0.114-0.069 Log income 0.391 *** 0.315 *** 0.265 *** r2 0.094 0.149 0.137 N 3445 20095 20095 Note: Dependent variable is satisfaction with life variable. Presented are OLS estimates with standard errors in parenthesis. * / ** / *** indicate significance at the 10% / 5% / 1% level.
Comparison in the regression While the representativeness of the WI sample can be a concern, it is shown that estimates from the WI sample exhibit very similar patterns to the estimates from the ESS sample. Estimates from the WI sample with weights are closer to ESS estimates so we proceed with the WI sample with weights.
Satisfaction with life Satisfaction with job Satisfaction with work and family (1) (3) (5) Female 0.018 0.038-0.167 *** Edu: Primary ref. ref. ref. Edu: Secondary -0.052-0.014 0.025 Edu: Tertiary 0.005 0.024 0.057 Single ref. ref. ref. Married 0.531 *** 0 0.084 Living with partner 0.347 *** -0.057-0.021 Divorced 0.358 *** 0.062 0.019 Widowed 0.289 0.177 0.326 * Age 15-24 ref. ref. ref. Age 25-34 -0.267 *** -0.127 ** -0.096 Age 35-44 -0.449 *** -0.12 * -0.029 Age 45-54 -0.44 *** -0.007 0.222 *** Age 55-64 -0.213-0.065 0.334 *** Health excellent ref. ref. ref. Health good -0.498 *** -0.091 ** -0.402 *** Health poor -1.118 *** -0.386 *** -0.72 *** Health very poor -1.745 *** -0.816 *** -1.235 *** Foreign-born 0.147 * 0.044-0.032 Self-employed -0.133 0.594 *** -0.122 Occupation prestige 0.006 *** 0.008 *** 0.001 Log personal income 0.248 *** 0.264 *** -0.035 Main household earner -0.12 ** 0.023-0.087 ** Lives with child aged 0-5y 0.071 0.025-0.089 ** Lives with child aged 6-17y -0.107 0.013-0.037 House owner 0.319 *** 0.079 * 0.18 *** Lives with parents -0.116 0.008-0.085 r2 0.149 0.102 0.148 N 20095 20095 20095
The baseline models Health status is being identified as a very strong predictor in all domains of SWB. The U-shape pattern with age is confirmed. Compare to men, female are less satisfied with family and work balance. Personal income is not associated with satisfaction with the family and work balance. The house ownership exhibits a positive correlation with SWB in all domains.
Estimates from augmented models Satisfaction with life Satisfaction with job Satisfaction with work and family (1) (2) (3) Log personal income 0.127 ** 0.085 ** -0.068 * Occupation prestige 0.003 0.004 *** 0 Permanent contract -0.041-0.159 *** -0.054 Supervisory position 0.011 0.084 ** -0.087 ** Over-qualified for job -0.247 *** -0.263 *** -0.033 Member of trade union 0.114 ** -0.061 * -0.029 Working hours >40-0.246 *** -0.095 ** -0.264 *** Works in night -0.1 ** -0.053-0.321 *** Works on weekend -0.08-0.077 ** -0.3 *** Work: commutes 15-45min -0.091 * -0.007-0.174 *** Work: commutes >45-0.156 * 0.012-0.364 *** Looking for another job -0.375 *** -0.723 *** -0.279 *** Good career opportunities 0.626 *** 0.744 *** 0.288 *** Job is insecure -0.475 *** -0.347 *** -0.102 * Past unempl. experience -0.059-0.069-0.111 ** r2 0.197 0.261 0.215 N 20095 20095 20095
Job characteristics and well-being 1. Jobs with high pay are characterized by better working conditions, the positive impact of personal income is diminished when job characteristics are controlled. 2. Health status substantially influences SWB, this result is robust to the inclusion of working conditions. 3. We find a strong link between job insecurity and low well-being in all domains. 4. Conversely good career opportunities and job stability are positively correlated with satisfaction scores.
Job characteristics and well-being 5. Certain job characteristics such as long working hours, irregular working schedules and long work commutes have strong negative effects on overall worker life quality. Workers largely underestimate the negative effects of commuting on the quality of their life. 6. Supervisory position exhibit a positive effect on job satisfaction, but negative effect on family. 7. Past unemployment experience has a negative effect on the satisfaction with the family and work balance.
Conclusion We identify several aspects of work characteristics on different domains of personal well-being. Conclusions from this paper show that data from web-surveys can be used to study the well-being of nations.
Comments and feedback: martin.guzi@econ.muni.cz pablodepedraza@usal.es
Variable name Definition Mean Satisfaction with life Satisfaction with life as-a-whole is measured on an ordinal 10-point scale. A higher value means that a person currently feels more 6.73 satisfied. Satisfaction with job Measured on an ordinal 5-point scale from highly dissatisfied (1) to highly satisfied (5). A higher value means that a person currently feels more satisfied. 2.99 Measured on an ordinal 5-point scale from highly dissatisfied (1) to Satisfaction with the highly satisfied (5). A higher value means that a person currently combination of family and work feels more satisfied. 3.19 Female Female =1, male=0 0.41 Edu: primary (International Standard Classification of Education 0-2)=1, otherwise=0 (reference) 0.39 Edu: secondary (ISCED 3-4)=1, otherwise=0 0.23 Edu: tertiary (ISCED 5-6)=1, otherwise=0 0.38 Single Never married and not living with a partner=1, otherwise=0 (reference) 0.41 Married Married=1, otherwise=0 0.53 Living with partner Never married and living with a partner =1, otherwise=0 0.13 Divorced Divorced =1, otherwise=0 0.05 Widowed Widowed =1, otherwise=0 0.01 Age 15-24 Age of respondent 15-24=1, otherwise=0 (reference) 0.12 Age 25-34 Age of respondent 25-34=1, otherwise=0 0.29 Age 35-44 Age of respondent 35-44=1, otherwise=0 0.30 Age 45-54 Age of respondent 45-54=1, otherwise=0 0.22 Age 55-64 Age of respondent 55-64=1, otherwise=0 0.08 Health status Satisfaction with health is measured on an ordinal 4-point scale from highly dissatisfied (1) to highly satisfied (4). A higher value means that a person currently feels more satisfied. 2.77 Foreign-born Respondent was not born in Spain=1, otherwise=0 0.06 Self-employed Respondent is self-employed=1, otherwise=0 0.01 Occupation prestige Occupational prestige is measured by index with 0 being the lowest possible score to 100 being the highest. The index conversion into ISCO categories is created by Ganzeboom and Treiman (2006) 43.08 Personal income Logarithm of gross monthly income in EUR (it include bonuses, if these were received in the last wage) 7.31 Child 0-5 years Child below age 6 lives in the household=1, otherwise=0 0.14 Child 6-17 years Child older than 5 lives in the household=1, otherwise=0 0.24 House owner House is owned=1, otherwise=0 0.78 Main household earner Contributes most to household income and not single=1, otherwise=0 0.50 Lives with parents Lives with parents=1, otherwise=0 0.21 Permanent Contract Respondent has permanent employment contract=1, otherwise=0 0.78 Supervisor = Respondent has supervisory position=1, otherwise=0 0.39 Over-qualified for job Respondent is overqualified for the job=1, otherwise=0 0.32 Member of trade union Member of a trade union=1, otherwise=0 0.37 Hours of Work The contractual hours for a worker in dependent employment. Works more than 40 hours per week=1, otherwise=0. 0.15 Work in the evening Respondent works regularly in the evenings=1, otherwise=0 0.60 Work on weekends Respondent works regularly on Saturdays or Sundays=1, otherwise=0 0.26 Work commutes 0-15 min Commuting 0-15min one way=1, otherwise=0 0.43 Work commutes 15-45 min Commuting 15-45min one way=1, otherwise=0 0.47 Work commutes >45 min Commuting 45+ min one way=1, otherwise=0 0.10 Looking for another job Respondent has been looking for another job in past 4 weeks=1, otherwise=0 0.24 Good career opportunities Respondent has good career opportunities in organization=1, otherwise=0 0.21 Job is insecure Job will become redundant next year=1, otherwise=0 0.12 Past unemployment experience Respondent was looking for the first job longer than one year=1, otherwise=0 0.13