Software

AIAS. Measuring occupations in web-surveys. Amsterdam Institute for Advanced labour Studies. the WISCO database of occupations. Working Paper PDF

Categories
Published
of 29
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Description
AIAS Amsterdam Institute for Advanced labour Studies Measuring occupations in web-surveys the WISCO database of occupations Kea Tijdens Working Paper January 2010 AIAS University of Amsterdam Acknowledgement
Transcript
AIAS Amsterdam Institute for Advanced labour Studies Measuring occupations in web-surveys the WISCO database of occupations Kea Tijdens Working Paper January 2010 AIAS University of Amsterdam Acknowledgement This paper is based on the EurOccupations project (May April 2009), which was funded from the European Community (FP ). For more information, see The author thanks all EurOccupations team members for their feedback and Joyce Jcobs for her contribution to the drafting of the database. January 2010 Kea Tijdens, Amsterdam Information may be quoted provided the source is stated accurately and clearly. Reproduction for own/internal use is permitted. This paper can be downloaded from our website under the section Publications/Working papers. Measuring occupations in web-surveys the WISCO database of occupations Kea Tijdens AIAS WP 10/86 Kea Tijdens Page 4 Measuring occupations in web-surveys Abstract Occupation is a key variable in socio-economic research. In surveys, occupation is predominantly asked using an open response format, followed by field- or office-coding. Web-surveys can apply a closed response format, allowing for respondent s self-identification when using a detailed list of distinct occupational titles and a search tree for navigating. This article summarizes the principles underlying the WISCO Database of Occupations for web-surveys, which includes a source list of 1,594 occupational titles in English, countryspecific translations of these occupational titles, and a 3-tier search tree. The occupational titles are coded according to ILO s international classification of occupations, ISCO-08. The database has been designed as part of the EU-FP6 EurOccupations project, measuring occupations with a greater precision than ISCO 4-digit by adding further digits, and is freely available from The WISCO Database is currently used in the WageIndicator web-survey on work and wages in approximately 50 countries, see Keywords: occupations; measurement; search tree; web surveys; database Page 5 Kea Tijdens Page 6 Measuring occupations in web-surveys Table of contents ABSTRACT INTRODUCTION OCCUPATIONAL CLASSIFICATIONS AND DATA-COLLECTION Occupational classifications and their characteristics The occupation variable in major datasets Data-collection and data-processing THE PROCEDURE OF DRAFTING AND TESTING THE WISCO DATABASE OF OCCUPATIONS Defining occupations A stepwise procedure to draft the source list Usability test THE SEARCH TREE AND READABILITY ISSUES THE SOURCE LIST AND OCCUPATIONAL HIERARCHIES Introduction CEO s, directors, managers (OCC+4 and OCC+3) Heads of departments or branches (OCC+2) First line supervisors (OCC+1) Helpers (OCC-1) THE SOURCE LIST: MISCELLANEOUS No additional information needed on industry, firm size or employment status Avoiding gender bias Coping with composite occupations Handicraft workers and machine-operators Broad titles for unskilled occupations Subsistence farmers, fishers, hunters and gatherers Obsolete occupations New and emerging occupations...25 Page 7 Kea Tijdens 7. CONCLUSION REFERENCES...29 AIAS WORKING PAPERS...31 INFORMATION ABOUT AIAS...37 Page 8 Measuring occupations in web-surveys 1. Introduction Occupation is a key variable in socio-economic research, used in a wide variety of studies, among others school-to-work transitions, manpower forecasting, the gender pay gap, skill obsolescence, occupational health and safety, processes of professionalization, and social stratification (e.g.(warren, Sheridan, & Hauser, 1998). In paper-based, telephone or face-to-face surveys, it is mostly asked in an open response format. In contrast, web-surveys offer a unique possibility for a closed response format, using a search tree. As part of the EU-FP6 funded EurOccupations project ( , no , a free downloadable database of occupations has been designed for eight EU-member states. The project aimed to provide a tool for self-identification in surveys and for measuring occupations with a greater precision than the 4-digit units of the International Standard Classification of Occupations ISCO by adding further digits. It did not aim for revising ISCO or any other occupational classification. The project included a test to what extent skill levels and job content were comparable across these countries for 150 occupations. The database is used in the continuous, worldwide WageIndicator web-survey, For this purpose, WageIndicator has added translations for another twenty languages to the database. Thanks to these two projects, the World database of ISCO occupations WISCO could be drafted, including: A source list of 1,594 distinct occupational titles in English with ISCO-08 codes plus further digits Country-specific translations of these occupational titles A 3-tier search tree that allows respondents to navigate through the database In many ways, the social sciences may profit from the WISCO Database of Occupations. First, when used in multi-country web-surveys, it will increase comparability of the occupation variable across countries. Second, when used in web-surveys with large sample sizes, the detailed occupational titles allow for analyses of sub-samples previously not possible. Third, the database can be used in computer-assisted face-to-face surveys for the occupation question, when the interviewer turns the screen to the respondent. This article details the design principles used in WISCO Database of Occupations. In section 2, the context of the project is briefly discussed, notably the occupational classifications, data-collection and the occupation variable in major datasets. Section 3 outlines the methodology used in the drafting of the search tree and the source list of occupations. Section 4 details the principles underlying the search tree, such as the search paths and the alphabetic sorting, and readability issues, such as the wording of occupations and Page 9 Kea Tijdens the translations. Section 5 details decisions underlying the source list with regard to occupational hierarchies, such as skill levels and corporate hierarchies. Section 6 details miscellaneous decisions underlying the source list. Section 7 draws the main conclusions learned from the project. Page 10 Measuring occupations in web-surveys 2. Occupational classifications and datacollection 2.1. Occupational classifications and their characteristics In 1958, the International Labour Office (ILO) of the United Nations had developed the International Standard of Occupational Classification (ISCO) to harmonize the measurement of occupations, with revisions in 1968, 1988, and 2008 (Budlender 2003). In the 1990 s, the ILO has undertaken efforts to implement ISCO-88 in many countries (Hoffmann et al 1995). Today, ISCO has become the standard classification in many countries for their Labour Force Surveys or Censuses (Greenwood 2004). However, countries such as the United States, Austria, United Kingdom, Germany, France, and the Netherlands, continue using their own National Occupational Classifications (NOC). These classifications tend to differ cross-nationally with respect to the level of detail, to specific occupational titles included in the classifications, and to their logic (Ganzeboom and Treiman, 1996). Attempts to harmonize NOCs were, among others, hampered by the fact that ISCO does not allow skill levels of occupations to vary across different national contexts (Elias 1997). ISCO-08, as was the case for its predecessors, defines a job as a set of work tasks and duties performed by one person. Jobs with the same set of main tasks and duties are aggregated into the so-called 4-digit occupation units. On the basis of similarity in the tasks and duties performed, the units are grouped into 3- and 2-digit groups, which in turn on the basis of the skill level are grouped into 1-digit groups (Greenwood, 2004). ISCO distinguishes four skill levels, notably unskilled, semi-skilled, skilled and highly skilled, which are related to ISCED, the International Standard Classification of Education (UNESCO 2006, reedition). During the preparation of ISCO-08, the similarity of occupations raised few discussions, but the major discussions concerned the skill levels assumed with the ISCO codes (Elias and Birch 2006). National statistical agencies have been asked to check the assigned skill levels for a number of occupations, though there is no evidence whether and how the agencies have undertaken an empirical investigation to answer this question. From a comparison of two UK occupational classifications Elias and McKnight (2001) conclude that at the aggregate level occupational classifications appear to provide a robust method for the measurement and analysis of skill. Dumont (2006) assumes that for wage analyses the ISCO skill levels are to be preferred Page 11 Kea Tijdens instead of the proxy data for skills that are more often used, but based on simple wage regressions for four EU member states he concludes that the ISCO skill levels are not very reliable. He suggests that skill levels of occupations will vary across countries. In eight EU member states EurOccupations has undertaken a systematic empirical investigation regarding the skill levels of 150 occupations, selected from the initial source list. The results show that the vast majority of these occupations did vary largely with regard to skill levels, not to job content (see EurOccupations teams 2009). A worldwide empirical underpinning of the skill levels of ISCO occupations might strengthen the value of the classification, and thus of the data collected with the classification. Such an underpinning would require cross-over tables from national educational categories into ISCED, currently lacking for most countries outside the OECD. Furthermore, it would require clear definitions of the unit of analysis, notably vacancies, jobs or jobholders. An analysis of vacancies require cross-over tables from the vacancies job titles into the ISCO occupational units, and it needs to address the issue of average versus minimum required skill levels as well as a wide variety of methodological issues related to the measurement of vacancies. An analysis of jobs requires experts, judging the jobs skill levels. Among others, the O*Net database of Occupations in the USA applies this method. EurOccupations has applied this method for its study of the skill levels and job content of 150 occupations in eight EU member states. Yet, it seems beyond reasonable budgets to do so worldwide. Analyses of jobholders can rely on survey data of individuals, self-assessing their current skill level in relation to the job level. Finally, such an empirical underpinning has to solve the different approaches to required skill levels in countries with elaborate vocational training systems, such as Germany, versus countries with an emphasis on on-the-job training, such as the USA The occupation variable in major datasets Multi-country datasets are typically surveyed by national survey agencies with the data merged afterwards (Hoffmann 2000). In these cases, the survey operations, the question formulations or the coding procedures are mostly not fully harmonized, affecting the comparability of the resulting statistics. Most international datasets include occupation data, aggregated at ISCO 1-digit, 2-digit or at best 3-digit. For example, for most but not all EU member states, Eurostat has 3-digit ISCO information in the European Labour Force Survey (ELFS) data. The European Community Household Panel holds ISCO 2-digit occupation data. The 1990 and the 1995 European working conditions surveys (EWCS) have 1-digit ISCO and the 2000 and 2005 surveys have 2-digits. The World Values Survey has occupation data on ISCO 1-, 2-, or Page 12 Measuring occupations in web-surveys 3-digits, varying across countries. Elias and McKnight (2001), although noticing improvements in the comparability rating of national occupational distributions over the 1990 s, conclude that analyses of detailed occupational categories should be undertaken with care. They call for harmonization of survey questions, for the adoption of common coding procedures and for a common understanding of the conceptual basis of ISCO, in particular its skill concept. Regardless improvements of comparability, the national occupational data-collection may again become a problem in the years to come because statistical offices increasingly rely on administrative sources such as personnel records for their socio-economic data collections. Among others, this is an attempt to get rid of expensive survey data. Yet, most administrative data collections do not register occupation data (Budlender 2003). Personnel records may register job titles, but countries hardly have cross-over tables to ISCO occupations (Budlender 2003). This stresses the need for an occupations database for surveys Data-collection and data-processing Many socio-economic surveys include a question What is your occupation?, What kind of work do you do? or similar, using an open or a closed response format. In the open response format, respondents report their job titles as they like. This format can be used in all survey modes. It requires a recoding effort of the data-collector by means of field- or office-coding. To facilitate coding, an additional tasks and duties question may be asked. Field-coding assumes computer-assisted interviews and it is advantageous because it allows the interviewer to ask additional information if needed. Office-coding is recoding at a later point in time and is disadvantageous in budget terms. Recoding requires a coding index and many national statistical offices have developed one. Data collectors in the UK and other countries use software programs for automatic recoding, such as CASCOT or its update CASCOT2000. Advanced automatic coding tools are currently developed (Michiels and Hacking 2004). As part of EurOccupations, a multi-language coding tool has become available (Elias, Ellison and Jones 2009). Nevertheless, occupational coding is an inexact process (Elias 1997). Based on a meta-analysis of the results from occupational recoding studies, the author concludes that agreement rates increase with higher levels of aggregation, thus at 1- or 2-digits. At 3-digits, agreement rates in excess of 75 per cent are hard to obtain and 3-digit ISCO comparisons between countries will be exposed to the low level of reliability associated with occupational classification. Page 13 Kea Tijdens In the open response format survey questions, respondents tend to report a detailed job title, as they know it from their employment contract, job classification scheme, collective bargaining agreement, job advertisement, or just from a common understanding in the workplace. They may report highly disaggregated occupations, such as Lithographic stone grinder, or very firm-specific job titles, e.g. Appls Prog I or highly aggregated categories, e.g. Clerical worker or Teacher, or they may be not specific at all, e.g. Employee of department X, Senior supervisor, or Dogsbody. In case of field-coding questions for clarification can be asked. In case of office-coding these reported job titles have to be classified either in highly aggregated categories or as unidentified data. Thus, the open response format question may lead to aggregation differences and to unidentifiable occupation data. In the closed response format, a tick list offers a choice of occupational titles or categories. This selfidentification method can be used in all survey modes, but the choice-set varies across the modes. Telephone-surveys allow for asking at most 5 highly aggregated occupational categories. Paper-based or faceto-face surveys allow for a choice of at most 50 categories when using show-cards, mostly a mixture of aggregated and disaggregated occupations. A limited choice-set may result in lower data quality, because it is difficult to assure consistency in how respondents fit their own job titles into the highly aggregated categories, introducing aggregation bias (De Vries and Ganzeboom 2006). Based on a comparison of three datasets, the authors show that socio-economic status indicators derived from self-identification on a 9-category list are slightly better than those from recoded open response format questions, but that the joint data leads to the best results. Web-surveys allow for a choice-set of a thousand or more occupational titles, when using a search tree to navigate through the choice-set. For four reasons, this is method advantageous. First, respondents are offered a choice from a list of occupations, all at the same level of aggregation. Second, unidentifiable occupational titles are absent. Third, field- or office-coding is not needed. Finally, in case of cross-country data-collections, survey operations and the choice-set will be comparable across countries. Page 14 Measuring occupations in web-surveys 3. The procedure of drafting and testing the WISCO Database of Occupations 3.1. Defining occupations The primary aim of WISCO Database of Occupations is its use for valid self-identification of occupation in web-surveys. Given that respondents prefer to indicate their job titles rather than aggregated categories, the source list of occupational titles had to be close to the wording used in job titles, thus requiring a long list of occupational titles, though these occupations have to be distinct from each other as synonyms or overlapping occupational titles may confuse respondents. Yet, the longer the list, the higher the average respondents reading-time and the higher the likelihood of dropout during survey completion. The source list therefore has to optimise between the demand to include as many distinct occupational titles as possible to facilitate valid self-identification and the demand to be as brief as possible to reduce reading time. The length of the source list is further determined by the search tree, consisting of a 2- or 3-tier tick list, detailing broad categories in the 1st tier to detailed items in the 2nd and 3rd tier. To prevent visitors from scrolling, a standard search tree on a computer screen can cope with some 20 items in the 1st tier and up to 20*20=400 items in the 2nd tier and 20*20*20=8,000 items in the 3rd tier. As 400 occupational titles definitely are too few, the search tree had to consist of 3-tiers with a limit of 8,000 occupational titles. A third argument for an efficient source list is the number of jobholders. Occupations with few jobholders are preferably not included, whereas occupations with large numbers of jobholders are preferably broken down in two or more occupational titles. As a rule of thumb, a 0.01% limit of the labour force was used. Thus, the Secretary of state is not included, even though it is distinct from all other occupational titles. By contrast, the Clerk has been broken down into several distinct occupational titles. Similar breakdowns have been made for teachers, nurses, attendants, marketing staff, IT staff, social workers, sales assistants, sales representatives and a few other large occupations. Two sources have been used for an impression of the size of an occupation, notably the distributions over 3-digit ISCO-88 occupational groups for 6 EU member states, using ELFS, and the distributions over 5-digit ISCO-88 occupations for 10 European countries and 10 countries outside Europe, using WageIndicator data. Page 15 Kea Tijdens The WISCO Database of Occupations employs the following definition: An occupation is a bundle of job titles, clustered in such a way that survey respondents in a valid way will recognize it as at their job title; an occupation identifies a set of tasks distinct from another occupation; an occupation should have at least a not-negligible number of jobholders and it should not have an extremely large share in the labour force A stepwise procedure to draft the source list Since 2001, when the WageIn
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks