Web Survey Bibliography

Title Measuring the very long, fuzzy tail in the occupational distribution in web-surveys
Year 2014
Access date 11.06.2014

pdf (476 KB)


Relevance & Research Question: Most surveys have one or more questions with thousands of response categories, the so-called long-list variables, such as occupation, industry, car brand, and alike. Typically an open-ended format with field- or office-recoding is used for these questions. In PAPI, CAPI and CATI, a closed format question restricts the number of responses to at most 50, being the maximum of a show card. In web-surveys closed format questions offer new opportunities, because the number of responses is not limited. For the survey design this requires a lookup database with all responses, a search tree to allow respondents to navigate through the database, and text string matching to ease the search.
This paper focusses on the requirements for a lookup database in web-surveys. The database can unfortunately not cover all possible responses, because no country has a full registration of job titles; in many countries the stock of job titles may well exceed 100,000. Typically, the occupational distribution has a very long tail, challenging the number of job titles in the database. The research aims to identify the share of respondents with job titles in the long tail of the distribution.
Methods & Data: The paper uses the data of the 2009 representative LISS web-survey on work and wages for the Dutch labour force (N = 3,508). For the occupation question this survey used a compulsory 3-step search tree with a lookup database of 1,614 occupational titles. The search tree included an option ‘other’ with a text box. The responses were coded.
Results: Of the 1,614 titles, 600 were ticked by 68% of respondents and 32% used the text box. 12% could have ticked an occupation from the lookup database. Of the remaining 16%, approximately one in ten had a job title that was mentioned at least twice and the rest had a unique job title. The long tail of the occupational distribution covered one in seven respondents.
Added Value: This study contributes to the quest to prevent manual coding of occupations in CAWI and CAPI and to the optimal number of occupational titles in a lookup database.

Year of publication2014
Bibliographic typeConferences, workshops, tutorials, presentations

Web survey bibliography - Tijdens, K. G. (27)