University Dropouts and Firm Hiring

No intervention planned.

2024-02-27

2024-09-30

The primary dependent variable is the choice variable of an alternative (the “candidate”) being chosen in the discrete choice experiment among two alternatives with four attributes each. The variable takes the value 1 for the chosen alternative and 0 for the alternative that was not chosen.

Our mixed multinominal logit model allows for the estimation of heterogeneous preferences among respondents. Therefore, preference heterogeneity will be investigated depending on company characteristics, such as company size, industry, language region and further available company characteristics.

We examine a discrete choice experiment that is embedded into an online survey of roughly 5000 firms in Switzerland. The respondents are asked about their preferences in regards to filling a job position.

The survey will be carried out by a professional survey institute among firms with at least 2 employees who participate in the 'Nahtstellenbarometer' survey which is conducted from February to April, and late summer on behalf of the Swiss State Secretariat for Education, Research and Innovation.
Participating firms are presented three choice sets consisting of two hypothetical candidates for a position as assistant to the management. Choice sets always consist of two candidates, one of them always a university dropout and the other always being a high school graduate who has not spent any time in university education. Dropout candidates vary in 4 attributes: years since graduation (1 or 2 years), what they studied (business administration, law, philosophy, German studies), gender (female, male), and reservation wage (gross annual wage of CHF 60,000, 70,000, 80,000). High school graduates also vary in four attributes: years since graduation (1 or 2 years), how they've spent their time since graduation (work and travel, traineeship), gender (female, male), and reservation wage (gross annual wage of CHF 60,000, 70,000, 80,000).
The primary outcomes of interest are the coefficients corresponding to our four attributes and their levels, and in the case of years since high school graduation and activities since graduation, their two-way interaction. These coefficients are estimated using mixed multinominal logit models (MXL) with the likelihood of an alternative being chosen based on given attribute levels as the dependent variable.

Randomization of the block of choice sets assignment to firms, the order of presented choice sets, and the order of alternatives within choice sets was done by computer using Stata.

Firm.

No

No clustering.

We expect roughly 5000 participating firms drawn by the Swiss Federal Statistics Office.

No clustering or different treatments. However, in order to optimize our mixed logit DCE model and to keep the workload for respondents low, a total of 21 choice sets are subdivided into 7 blocks of 3 choice sets each are randomly assigned to firms, with expected 5000 firms participating, 714 firms per block.