Experimental Design
Participants first received general information about the study, including its stated purpose, procedure, expected duration, compensation, voluntary nature of participation, confidentiality, and data protection in accordance with applicable regulations. To reduce social desirability bias and avoid demand effects, the study was framed as an investigation of how the composition and wording of short job postings influence employer attractiveness. Participants then provided informed consent to participate and to the anonymous processing of their data.
Subsequently, the first part of the questionnaire collected demographic information, including gender, age, highest educational attainment, current employment status, years of professional experience, and country of residence. Participants who indicated that they were retired were excluded to ensure that the sample reflects individuals who are actively participating in the labour market.
Following the demographic questions, an attention check was administered. Participants were presented with the instruction: “To demonstrate that you are reading the questions carefully, please select ‘Agree’ for this item.” Four response options were provided. This attention check was implemented to identify and exclude inattentive respondents, thereby enhancing the quality and reliability of the collected data. Ensuring that participants carefully read and understood the questions helps to reduce noise and potential bias in the experimental results.
In the second part of the questionnaire, participants provided job- and salary-related reference information. Specifically, they reported their interest in changing jobs within the next 12 months, the type of position they would seek (entry-level, experienced professional, managerial, or other), their current job title, their minimum acceptable annual gross base salary excluding bonuses and variable compensation components, and the working hours associated with this salary expectation. These responses served as individualized benchmarks for generating the salary offers in the subsequent forced-choice tasks.
This section was followed by an attention check regarding the reported salary. Participants were asked to indicate how confident they were in the accuracy and realism of the amount they had provided. The purpose of this check was to ensure that respondents did not enter arbitrary values, but instead carefully considered their actual minimum salary expectations. By confirming the validity of these self-reported benchmarks, we ensured that all salary information was perceived as meaningful and that subsequent decisions in the forced-choice tasks were not artificially influenced by offers falling below participants’ expected minimum salaries.
The experimental section consisted of two sequential parts. First, all participants completed the control condition forced-choice task, followed by the treatment condition forced-choice task. Each condition included three forced-choice tasks, which were presented in a randomized order to minimize order effects.
In each task, participants were simultaneously presented with two job postings and were required to select one of the two. The positioning of the advertisements (first or second) was randomized across tasks to control for potential positional bias. Each job posting included a brief narrative company description, comprising firm size, years of market presence, customer base, and general management orientation. Each element was described in a single sentence to ensure comparability across profiles while maintaining realism. For the experiment, three distinct company profiles were designed (A, B, and C), which generated three pairwise comparisons within the job postings: A & B, A & C, and B & C, which are illustrated within the appendices. Below the company description, participants were shown their previously indicated job title along with an annual gross base salary offer. The critical difference between conditions was that, in the treatment condition, each job posting additionally displayed a sustainability score below the salary, accompanied by a brief explanation of the score.
For each pair of job postings, salary levels were generated algorithmically relative to each participant’s stated minimum salary expectation. One offer was randomly set between 0% and 20% above the participant’s minimum salary, while the alternative offer was also within the 0% to 20% range but strictly lower. Thus, both offers exceeded the participant’s reservation wage, yet one was always financially superior. The assignment of which job posting received the higher or lower salary was randomized across tasks to control for potential positional biases.
To illustrate, consider a participant whose minimum acceptable salary was 50.000,00 €. In one task, job posting a offered 58.000 € (+ 16,00 %) with a sustainability score of 70, whereas job posting b offered €54,000 (+ 8,00 %) with a higher sustainability score of 90.
Following the experimental section, participants completed a structured sustainability preference elicitation. First, they allocated 100 points across the three broad sustainability dimensions, environmental, social, and governance, reflecting the relative importance of each dimension in their employer selection decisions. Second, they allocated 100 points within each dimension across six specific subdimensions. For the environmental dimension, these included climate change mitigation, climate change adaptation, sustainable use of water resources, transition to a circular economy, pollution prevention, and protection of biodiversity and ecosystems. For the social dimension, the subdimensions comprised working conditions, equality and diversity, health and safety, human rights in the supply chain, training and development, and impact on local communities. For governance, the subdimensions included board independence, executive compensation policy, anti-corruption measures, transparency and disclosure, risk management, and compliance systems. The constant-sum format forced respondents to make explicit trade-offs, thereby reducing the likelihood of uniformly high importance ratings.
Following the treatment condition, an attention check was administered to ensure that participants had noticed and cognitively processed the sustainability information presented in the job postings. Respondents were asked to indicate whether they had perceived and considered the sustainability score when making their choices. This check served to confirm engagement with the treatment and to enhance the internal validity of the experimental results by identifying participants who may have ignored the sustainability information.
At the conclusion of the survey, all participants received a debriefing, which provided detailed information about the study’s purpose, the experimental manipulations, and the use of sustainability scores in the job postings. The debriefing ensured transparency, clarified any potential misconceptions, and reinforced ethical standards by informing participants about the nature of the trade-offs they had encountered during the experiment.