Self-identification of occupation in web surveys: requirements for search trees and look-up tables

Abstract

Can self-identification of occupation be applied in web surveys by using a look-up table with coded occupational titles, in contrast to other survey modes where an open format question with office-coding has to be applied? This article is among the first to explore this approach, using a random sampled web survey (N=3,224) with a three-level search tree with 1,603 occupations and offering a text box at the bottom of each 3rd level list. 67% of respondents ticked in total 585 occupations, of which 349 by at least two respondents and 236 by only one, pointing to a long tail in the distribution. The text box was used by 32% of respondents, adding 207 occupational titles. Multivariate analysis shows that text box use was related to poor search paths and absent occupations. Search paths for five of the 23 first-level entries should be improved and the look-up table should be extended to 3,000 occupations. In this way, text box use and, thus, expensive manual coding could be reduced substantially. For such large look-up tables semantic matching tools are preferred over search trees to ease respondent’s self-identification and thus self-coding.