Tuesday 16th July
Wednesday 17th July
Thursday 18th July
Friday 19th July
Download the conference book
Measuring Occupation Cross-nationally |
|
| Convenor | Dr Eric Harrison (City University London) |
| Coordinator 1 | Dr Kea Tijdens (University of Amsterdam) |
Analysis of peer reviewed publications based on ESS and the GSS shows that socio-demographic or 'background' variables are the most widely used. Of these, occupation of respondents and where applicable, their partners and parents, is a rich source of information but one that is problematic to measure precisely, accurately and consistently.
Much effort has been invested in producing standardised coding frames for use worldwide, the relevant one here being the new International Standard Classification of Occupations (ISCO-08), but its consistent implementation across countries will present many challenges to cross-national measurement equivalence, as countries attempt to map from a) their own classifications to ISCO and/or b) ISCO88 to the new revision.
We invite methodological and substantive papers in this session. Methodology can include evaluations of the quality of existing occupational data and/or new approaches to the collection and coding of occupational data in the field. Contributions which use data that has been coded to the new revision of ISCO will be particularly welcome.
This paper examines the influence of religion on female participation to the labor market using data relative to women aged between 18 and 60 years in 47 European countries drawn from the European Values Study (EVS). We investigate the determinants of the probability of being employed rather than jobless in a LOGIT framework. The results show that women belonging to the Orthodox and, even more, Muslim denominations present a higher risk of non-employment than the agnostics, while being a Protestant increases the probability for a woman to be employed. Although its intensity is slightly weakening, the association between religious affiliation and female labor supply is robust to different sets of controls for individual and household heterogeneity as well as for welfare regimes and country specificities. Once disentangling religiously active and non-active women, we find that there are small differences between them in the case of the Orthodox and Muslim women, while active Catholic women tend to work less and non-active Protestant women tend to work more than average.
Between 2004 and 2006 a consortium of academics and statistical offices worked to develop a prototype version of a harmonized European Socio-economic Classification (EseC) that could be used to understand differences in social structures and socio-economic inequalities across the European Union. The resulting classification - rooted in the concept of employment relations - was designed to be derivable from any dataset containing information about occupation, employment status, establishment size and supervisory responsibility (Rose and Harrison 2010). Work has begun within Eurostat to develop a simplified classification based on the latest version of ISCO, and one which would be less sensitive to the incomplete and/or imperfect labour market information found on some surveys.
One of the first surveys to code occupations to the new ISCO08 classification was the European Working Conditions Survey (EWCS) undertaken in 2010. This paper uses EWCS data to assess the operational, criterion and construct validity of a revised ESeC based on ISCO08, and compares the explanatory power of different versions of the Mark II ESeC, for instance using 3 digit occupation code, 2 digit occupation code, and with and without information about supervisory status. It makes use of the ESeC schema's flexibility to be used in a nine-, five and three-class format.
Most indicators of status or social class are based on occupation. The problem arises in case of respondents who do not work during the survey (such as unemployed, housewives, pensioners, retirees, etc.) In fact, respondents who do not work, can be divided into those who: (i) have never had a job - these cases are lost since there is no basis to classify them to any occupational category, (ii) respondents who currently do not hold any job but have worked in the past. The social position of the latter category is most commonly defined in terms of their last occupation. It is assumed that the last occupation, even held many years ago, is equivalent to its counterpart for economically active individuals. It cannot be excluded, however, that categories of active and non-active respondents differ in a substantial way with respect to key determinants of their social position such as financial situation, prestige, lifestyle, and social capital. This, in turn, may affect their values, attitudes and opinions. Using the European Social Survey data we test the hypothesis that economically active and non-active respondents significantly differ in terms of the objective and psychological variables, such as the household income, social trust, religiousness, support for egalitarianism, life satisfaction, etc. In discussion we confront the benefits of taking into account the last occupation of people who do not work with drawbacks of this approach. Generally, it results in decrease of validity of occupational measures in the whole sample.
Survey questions with many response categories (long-list variables) are typically asked in open-ended questions with field- or office-recoding, but can also be asked in closed format. While in other survey modes the number of response categories is at most 50, web-surveys allow for a choice of thousands, using search trees and/or text string matching with a lookup database. For measuring occupations, the lookup database cannot cover all possible responses, because the number of response categories is not known, as for many countries the stock of job titles may exceed 100,000 and the occupational distribution has a very long tail, thereby challenging the number of categories in the lookup database.
The paper uses the data of the 2009 representative LISS web-survey on work and wages for the Netherlands (N = 3508). For the occupation question the survey used a, obligatory 3-step search tree with a lookup database of approx. 1,700 occupational titles, followed by a text box, completed by 32% of the respondents. The paper analyses how many unique job titles remained, after cleaning for typing errors, redundant text and alike, and after identifying the uniqueness of the job title in the text box by comparing it to the ticked title in the lookup database. The objective of the paper is to identify 1) the share of respondents with job titles in the long tail of the distribution and 2) the optimal number of occupational titles in a lookup database.