The WageIndicator web survey for worldwide social science research on wages

1. The WageIndicatorweb survey for worldwide social science research on wages Paulien Osse, WageIndicator Foundation Kea Tijdens, University of Amsterdam 29 March 2007, ILO

2. National WageIndicator websites with up-to-date work-related information answering visitor’s emails Salary Check providing free occupation-specific wage information controlling for age, gender, education and region Web survey asking the visitors a favor in return completing a web survey on work and wages (prize incentive) the data are used for research and as input for the Salary Check Large numbers of visitors worldwide, the public shows a desire for wage information and is willing to complete the web survey The WageIndicator concept

3. A brief history (Netherlands) • 1999 desire for wage information on Internet detailed occupation wage data needed for research • 2000 survey about work and wages in women’s magazines • 2001 launch women’s WageIndicator website with web surveyand Salary Check for 45 occupations • 2002 launch websites for men, 40+, youth • 2004Salary Check for 400 occupations • 2006 400,000 web visitors per month in NL

4. To other countries • 2004 Belgium, Denmark, Germany, Spain, Finland, Italy, Poland, United Kingdom (EU funding 6th Framework Program) Hungary (EU funding EQUAL fund) • 2005 Argentina, Brazil, Mexico India, S-Korea, S-Africa, (funding NL development aid) • 2006 USA (funding Harvard Law School) • 2007 China, Russia, Sweden (contracts about to sign)

5. The WageIndicator Foundation owns the WageIndicator concept is a not-for-profit organization Its mission statement “Share and compare wage information.Contribute to a transparent labor market.Provide free, accurate wage data through salary checks on national websites.Collect wage data through web surveys.” Founded in 2003 under Dutch law by University of Amsterdam NL branch of the international career website Monster NL Dutch Confederation of Trade Unions (FNV) WageIndictator Foundation

6. 2007 35 websites in 17 countries, most of them managed by web journalists extra websites for multilingual countries, for women, elderly workers, IT staff (India) thousands of links in other websites Web visitors must trust the information provided in a Salary Check (thus it must offer high quality information) volunteering their data in the survey receiving a response to visitor’s email Web-marketing is critical cooperation with media groups, career sites, trade unions, all with a strong Internet presence WageIndicatorwebsites

7. Media parters and web-marketing • Worldwide partners • Career site Monster • MSN • Dutch World Service • Partners with established reputations • University of Amsterdam • Erasmus of Rotterdam • Harvard Law School • Leading national newspapers and portals • Gazeta Wyborcza (PL) • El Pais (ES) • La Nacion (AR) • UOL (BR) • Sueddeutsche (DE) • Mail & Guardian (ZA) • Etc.

8. The survey • Target population: labor force • wage-earners in formal and informal economy • self-employed, free lancers, home workers (with SEWA in India) • Questions on • Occupation (4 dgt ISCO), industry (4 dgt NACE), education, work history, wages, benefits, hours, personal questions • Questionnaire • completion takes approximately 20 minutes • survey has parallel questions addressing rare groups in the labor force to prevent break-off • optimization as for the number of characters, clicks and pages

9. The technique • Questionnaire Management System QMS • developed for WageIndicator, using Open Source • manages a multi-country, multi-lingual survey • facilitates complicated routing, downloading codebooks and uploading languages • includes a search tree application for questions on occupation, industry, region • Data storage • the data is securely stored on servers in USA, NL and India • quarterly data releases

10. Web traffic • Visits totals • 2005: 4.5 million • 2006: 7.8 million • January 2007: > 800,000 • Prognosis 2007: 10 million • Variation per country, some examples • NL: > 400,000/month (since 2001, household name) • DE: > 100,000/month (since 2004, large population) • BR/AR: 25,000/month (since 2006, well linked) • ZA/IT/KR: < 1,000/month (weaker teams)

11. The response(1) • 2006 • Total visits 7.8 million • Fully completed questionnaires 158,000 • Data for research and salary check 309,000 (2004-2006) • Response rate overall • 3.85 % • Large variation across countries • Cross-country analysis of response rates

12. The response(2) • Sample size (fully completed) • <2004 53,000 in NL • 2004 43,000 in 5 countries • 2005 135,000 in 11 countries • 2006 158,000 in 17 countries • 2007 250,000 in 19 countries (expected) • Data quality is good • hardly any ‘click the first item only’ respondents • item non-response usually < 5% • very few multiple responding • in 2007 a study on break-off respondents

13. Respondent-side feedback • Feedback on the websites • visitors use the website for their decisions about schooling, occupational choice, wage negotiations, and job mobility • we know from visitor’s email • Feedback on the survey • open-ended question “If you have any comments on the questionnaire, please do so here” • passive feedback through break-off • we want visitors to have fun in completing the questionnaire (and they report back that they do)

14. Volunteer web surveys • Selection bias and Internet access • worldwide Internet access rates are increasing fast • this population is becoming more and more representative of the population at large • it will boom with wireless access • Selection bias in choice of website? • web visitors can choose out of millions of websites • only a minor part visits a WageIndicator website • web traffic can be directed in number and in target group by means of web marketing • Selection bias in completing the survey? • 1–10 % of the web visitors completes the questionnaire (f.e. Finland 10 %)

15. Findings on selection bias • In all countries • the small groups in the labor force are under- represented, f.e. workers in small part-time jobs • low educated are increasingly not underrepresented • elderly workers 55+ are underrepresented • gender representation varies across countries • In Netherlands 2002-2006 • the underrepresentation of these socio-demographic groups has declined in the past years

16. Coping with selection bias • Web marketing • addressing the target population at large • sites for sub-populations otherwise not fully reached • Routing through the questionnaire • to prevent rare groups from break-off • Weighting with aggregate data • aggregate socio economic LFS data is used for weighting national WageIndicator data • Weighting with micro data • micro-data from representative surveys will be used to develop weights, using similar questions in WageIndicator, currently explored in the US • Weighting with a reference survey • using a small reference survey for weighting, currently explored in the Netherlands

17. Volunteer web surveys are advantageous • … because they can be held continuously • costs are not linear related to sample size • investment costs are relatively large-> conducting a continuous survey is profitable • for WageIndicator continuity is a prerequisite because of web marketing investments • continuous surveys allow for temporary plug-in questions • … because they can lead to large sample sizes • allows for analyses of sub-sets • allows for presenting randomly items from a pool • allows for questions addressing relatively small groups, thus acting as a screening device

18. The research • Research community • increasing numbers of researchers use the data • On wages and working hours • cross-country wage differentials for occupations • gender pay gap and the motherhood penalty • modeling preferences for a change in working hours • On work place relations • attitudes towards collective bargaining coverage • effect of dismissals on self-perceived job insecurity • On labor markets • the multi-dimensionality of the informal labor market within and across countries • spill over effects of MNE’s in local employment

19. Is this new? • Yes, it is because … • worldwide, neither high quality aggregate data nor micro-data about wages, bonuses, and working hours are available • worldwide, WageIndicator is the first survey gathering wage data in so many countries • worldwide, it is one of the first surveys using web marketing for scientific data collection • … and because • the exchange of information from research to the public and from the public to research is not often seen

20. A GlobalWageIndicator • The plan • a GlobalWageIndicator plan to enlarge the web operation to 75 countries in 5 continents • inspired by the globalizing economy and the need for worldwide data on wages, currently not available • jointly with International Labor Organization of the United Nations, Harvard Law School, University of Belgrano (AR), and Indian Institute of Management/Ahmedabad (India) • Its aims • contributing to a transparent labor market by provi- ding reliable data about wages to a worldwide public • collecting data for worldwide wage trend reports and for researching the impact of globalization • submitting plans to funding agencies in 2007

21. The role of ILO • Support for the 75-countries plan • Using the dataset for wage data or other data • Input for funding options • post Soviet area • Arab speaking world • Sub-Saharan Africa

22.  • Thank you for your attention • www.wageindicator.org