Studying Scarring Effects Using Factorial Designs: Rating or Ranking?

Early job insecurity is a much-discussed topic across European countries. Research overwhelmingly found that being unemployed after graduation affects employment chances and also future wages negatively, other research, however, did not find such scarring effects. Some of this mixed evidence may be due to the different ways in which data were collected. Evaluating the effects of potentially stigmatizing applicant characteristics on hiring chances, such as previous unemployment spells, is known to be prone to social desirability bias. Factorial survey experiments (FSE) and forced choice experiments (FCE) have been suggested to alleviate some of these problems.

In this workshop contribution, we gauge the capability of FSE and FCE to estimate effects of early career unemployment spells on recruiters’ hiring decisions. Using data obtained from a survey with sequentially implemented FSE and FCE with 2000 recruiters in Bulgaria, Greece, Norway, and Switzerland we compare FSE and FCE using multilevel linear regression models and multilevel probit models with random effects. Our preliminary results suggest that FCE may be better suited to gather valid data with minimal social desirability bias.