Do Peers Shape Working Hours Preferences? Evidence from an Information Provision Experiment

The experiment consists of three randomised treatment groups:
1) The first treatment group receives accurate information about the average weekly working hours of sector peers (sector categories follow SBI2008 classification).
2) The second treatment group receives accurate information about the average weekly working hours of neighbourhood peers (neighbourhood levels are 4-digit zipcode level in the Netherlands).
3) The third group is a control group that receives no information.

Peers are defined as individuals similar to the respondent among four dimensions: gender (man/woman), age (above/below 45), presence of at least 1 child under 12 (yes/no), and educational attainment (tertiary or below).

We want to study whether providing information about peer's labour supply impacts the respondents' preferred weekly working hours (elicited in the survey experiment). Secondary outcomes are satisfaction with own working hours (elicited in the survey experiment), the most important factors when deciding on working hours (elicited in the survey experiment), preference changes over time (elicited in previous and follow-up waves of LISS), and observed weekly working hours (administrative records and follow-up waves of LISS).
The data for the information treatment is derived from Statistics Netherlands.

2026-07-06

2026-08-25

Preferences for weekly working hours (elicited in the survey experiment)
Preference convergence

Preference convergence is defined as the absolute difference between preference and treatment the individual received: Ci = |Pi − Ti|

Satisfaction with own working hours (elicited in the survey experiment)
Most important factors when deciding on working hours (elicited in the survey experiment)
Preference for weekly working hours (follow-up waves of LISS)
Short-term change in preferences (described below)
Medium-term change in preferences (described below)
Observed weekly working hours (administrative records and follow-up waves of LISS)

Short-term change in preferences: difference between preferences measured 2 months before the survey and preferences elicited in the survey experiment
Medium-term change in preferences over time: difference between preferences measured 1 year after and preferences measured 2 months before the survey experiment

We conduct an individualised information provision experiment within the Dutch LISS panel. Respondents are randomly assigned to one of three possible groups: sector peers treatment group (S), neighbourhood peers treatment group (N), or control group (C).

Part 1:
Respondents report their sector of employment and their own current contracted weekly working hours. Respondents who are currently unemployed but actively seeking work (a small share of the sample) are instead asked to report information about their most recent job and sector. People already in retirement, still in schooling, self employed, or who are unable to work were not invited to the survey experiment.
Respondents in the treatment groups (S and N) are first asked to report their beliefs about the average contracted weekly working hours of their assigned peer group, as well as their expectations about that group's average satisfaction with contracted working hours. They also indicate how confident they are in their estimate of peers' working hours.
To avoid priming, respondents in the control group are initially asked only about their own contracted weekly working hours and sector of employment. They receive the questions on their beliefs about the average contracted weekly working hours of one randomly assigned peer group (half are asked about neighbour half are asked about sector), as well as their expectations about that group's average satisfaction with contracted working hours, only at the end of the questionnaire.

Treatment:
Following Part 1, respondents in S and N are informed about the accurate average weekly working hours of either, their sector or neighbourhood peers. The true answers are shown in comparison to their own guesses and the difference is highlighted.
Respondents in the Control Group receive no information.

Part 2:
Following the treatment, respondents are asked a set of questions that are our main outcome variables. All questions are asked to everyone in the same manner. First we ask about the satisfaction with own working hours. This is followed by information that under the Dutch Flexible Working Act (Wet Flexibel Werken, 2016), employees may request adjustments to their working hours to their employers. After receiving this information we ask about respondents' preferences for weekly working hours, the likelihood that their employer would grant a request to both increase and decrease the working hours, and the most important factors when deciding on working hours.

We invite around 3,000 individuals who are part of the LISS sample and expect a response rate of roughly 70%. Approximately 2,100 (~700 per treatment arm) panel members between the ages 18-67 are invited to respond to our experiment. We exclude individuals in early retirement, individuals who are still in school, individuals who are self-employed, and individuals who are disabled or otherwise not fit for work.

Randomisation is done in office by a computer

Randomisation is done at the individual level (every respondent is randomly assigned to one of the three groups)

No

We invite around 3,000 individuals who are part of the LISS sample.
We expect a response rate of 70% which would leave us with 2,100 individuals.

2,100 individuals

Treatment Group 1 (Sector Peers): 700 individuals
Treatment Group 2 (Neighbourhood Peers): 700 individuals
Control Group (no treatment): 700 individuals