SOEP Brown Bag Seminar

This presentation investigates the response patterns of survey respondents who are asked for the name and address of their employer. Results are of interest for survey methodologists as information on employers is a sensitive piece of information which is difficult to ask for and which might endanger participation to surveys, especially for panel studies like the Socio-economic Panel (SOEP). Results are also of interest for potential users of linked-employer-employee data as these response patterns may lead to selection issues. The SOEP Linked-Employer-Employee-Study (SOEP-LEE) is used as an example and a short introduction to this study will be given. The SOEP-LEE study consist of survey of German employers which was conducted in fall and winter 2012/2013 (N=1708). An "employee first" approach was used by which establishments were interviewed based on address information given by employed participants in the SOEP. The collected establishment data can be linked back to the SOEP to investigate the influence of establishment characteristics on individual level outcomes. Despite its advantages, this approach is used rarely in establishment surveys; hence factors and mechanisms which influence the naming of respondents' employers are not researched sufficiently. More research into respondents' provision of an employer's name and address is however necessary for two reasons: First, item nonresponse to the employer question decreases the sample size of the linked employer-employee sample. Second, selectivity bias might occur when systematic differences exist between respondents who are willing to give their employers name and address and those respondents who are not. This is especially true if the probability of naming the employer, and hence the probability of an establishment being included in the study, depends on characteristic which are of substantial interest for data analysts, such as income, work satisfaction or the health effects of work environments. Systematic differences in respondents' tendency to name their employer are investigated using logistic regression models and the effects of different groups of factors are estimated: (1) personal characteristics of the survey respondent such as demographics, health and personality variables, (2) survey and interview specific factors, (3) organizational characteristics such as size and branch of the company, (4) job characteristics such as income, supervision and time arrangements, and (5) attitudes towards the job such as job satisfaction and worries of job loss. Insights into the response process may guide the design of future questions on the topic and the decision as to when and where to ask this question in a survey context.