Workplace Amenities, Flexibility and Team Composition among Older Workers in Italy: A Conjoint Survey Experiment

Respondents complete a conjoint experiment concerning hypothetical job offers. They are asked to imagine that they are looking for a new job and have received two offers from firms and positions that are equivalent in all respects except for the characteristics shown in the experiment. In each task, respondents compare Job A and Job B, select the job they prefer overall, and assign a 0–10 rating to each option.

The job profiles vary along four dimensions:

- Net monthly salary: EUR 1,700; EUR 1,900; EUR 2,100; EUR 2,300.
- Employer-provided Long-Term Care insurance: an individual policy paying EUR 500, EUR 1,000, or EUR 1,500 per month if the employee becomes unable to perform at least two of six Activities of Daily Living because of illness, injury, or cognitive deterioration.
- Work-location flexibility: always on-site; one day of smart working per week; two days of smart working per week.
- Office team composition: one of three pre-specified visual team profiles. The three images are fixed in the survey instrument and vary the apparent age and gender composition of a four-person office/operational team.
The scenario is explicitly hypothetical. Respondents are instructed to compare the two offers with one another rather than with their current job. Each respondent answer to 5 different job offer choices.

Attribute levels are randomly assigned to the hypothetical profiles within each task according to the conjoint design. The analysis assumes random assignment of attribute levels across profiles and tasks. The 4 attributes are independently randomized, with no additional restrictions on (unreasonable) combinations of outcomes. The conjoint does not randomize the visualization order of the attributes, which is fixed: salary, LTC insurance; job flexibility and team composition.

Participants are not randomly selected: they are over-45 workers from 24 mid or big Italian firms. All eligible employees received and email of the survey company through their employer. The email asked employees whether they are willing to participate in the survey and, if yes, asked them to register into a dedicate web-form. Thus, participation to the study is voluntary.

2026-09-30

2026-11-30

For profile (j) in task (t) evaluated by respondent (i), the main outcome is Y_ijt, a dummy variable that takes value 1 if the specific job offer is selected as the preferred job in that pair, and (0) otherwise. The primary estimands are the Average Marginal Component Effects (AMCEs) of each randomised attribute level on the probability that a profile is selected.

The forced-choice outcome provides a direct measure of the relative attractiveness of hypothetical jobs and attractiveness of their characteristics. Because attribute levels are randomly assigned, differences in the probability that profiles are selected identify the causal effect of the assigned attribute levels within the stated-choice environment,

Profile rating: rating of each job from 0 (“I do not like it at all”) to 10 (“I like it very much”).
Monetary valuation / wage-equivalent valuation: the amount of monthly net salary associated with the estimated preference for a non-wage attribute, using experimentally varied salary as the monetary numeraire.
Choice–rating consistency: whether the profile chosen in the forced-choice question also receives the higher 0–10 rating.
Heterogeneous treatment effects across pre-specified respondent groups.

Ratings provide a cardinal complement to the forced-choice outcome. Marginal means are useful for describing absolute profile attractiveness, while AMCEs remain the primary causal estimands. Monetary valuations translate non-wage amenities into an economically interpretable monthly-salary metric, conditional on the salary coefficient being positive and sufficiently linear over the experimental range.

Heterogeneity groups: age, gender, extra-work commitments, (extra-work) caregiving responsibilities, firm sector, and subjective perceived age.

Respondents complete a conjoint experiment concerning hypothetical job offers. They are asked to imagine that they are looking for a new job and have received two offers from firms and positions that are equivalent in all respects except for the characteristics shown in the experiment. In each task, respondents compare Job A and Job B, select the job they prefer overall, and assign a 0–10 rating to each option.

The job profiles vary along four dimensions:

- Net monthly salary: EUR 1,700; EUR 1,900; EUR 2,100; EUR 2,300.
- Employer-provided Long-Term Care insurance: an individual policy paying EUR 500, EUR 1,000, or EUR 1,500 per month if the employee becomes unable to perform at least two of six Activities of Daily Living because of illness, injury, or cognitive deterioration.
- Work-location flexibility: always on-site; one day of smart working per week; two days of smart working per week.
- Office team composition: one of three pre-specified visual team profiles. The three images are fixed in the survey instrument and vary the apparent age and gender composition of a four-person office/operational team. (i) 1 Young; 2 Mid-Age; 1 Old; (2) 1 Young; 1 Mid-Age; 2 Old; (3) 1 Young; 3 Old.
The scenario is explicitly hypothetical. Respondents are instructed to compare the two offers with one another rather than with their current job.
Each respondent repeat the stated-choice five times.

Not available

Attribute levels are independently randomly assigned to the hypothetical profiles within each task according to the final programmed conjoint design. The analysis assumes random or design-balanced assignment of attribute levels across profiles and tasks.

Profile within respondent-by-task.

No

Not applicable for treatment assignment. The survey includes workers from 24 firms, but firms are not randomisation units.

We expect between 5000 and 10000 participants.

There are no between-respondent treatment arms.

Analysis Plan