The Role of Earning Profiles in Educational Preferences

We conduct our experiment in the context of a large survey on education in Germany. Respondents are randomly assigned to one of two groups: a treatment group that sees a labeled graph and a control group that sees an unlabeled graph of earnings trajectories. They are then asked which path they would recommend for a hypothetical child deciding between the two career paths. The sample includes an oversample of parents with children aged 0–6, allowing for separate analyses by parental status. In the control group, respondents are provided with information on annual lifetime earnings for each career path, which are presented as “Option 1” and “Option 2”. In the treatment group, respondents receive the same earnings information, as well as additional information on the associated educational pathway, distinguishing between “Option 1: Vocational education (apprenticeship)” and “Option 2: University education”. This research design allows us to examine to what extent education pathways, beyond differences in earnings profiles, shape preferences over career and education choices of the next generation.

2026-05-11

2026-07-07

We are mainly interested in the choice between career paths 1 and 2, elicited in the questionnaire.

Respondents will be presented with a graph showing two hypothetical income paths over the life cycle from age 18 to 55. One path reflects an earnings profile similar to that of the median German individual with a university education, while the other reflects an earnings profile similar to that of the median German individual with a vocational qualification.
Respondents are then asked to indicate, assuming they have a child who must choose between these two career paths, which path they would recommend.

Question: “Imagine you have a child in their teenage years who can choose between two career paths. The graph shows how much they would earn annually over their lifetime in each of the two cases. In Option 2, annual income is initially lower and later higher than in Option 1.

What would you advise your child to do? [Treatment screen]:”
o Option 1: Vocational education (apprenticeship)
o Option 2: University education

We are interested in the heterogeneity of preferences by respondents’ own educational background to study intergenerational persistence in education. For this, we use questions from a sociodemographic module in the questionnaire.

1. We are interested in the heterogeneity of treatment effects by respondents' prior beliefs on the earnings premia of different educational paths.
2. We are also interested in heterogeneity with respect to a series of perceived mechanisms that may help explain education preferences.
3. We will further test whether preferences for education pathways depend on respondents' time and risk preferences, or their understanding of the information given in the experiment (graph depicting two hypothetical lifetime income trajectories from age 18 to 55).
4. In addition, we will report results separately for the oversample of parents with children aged 0–6.

We will also test heterogeneity of treatment effects by respondents' prior beliefs on the earnings premia of different educational paths. To measure these beliefs, we elicit respondent’s expectations at the beginning of the experiment using a series of survey questions on perceived earnings associated with different educational trajectories. Specifically, respondents are asked:

Question: “First, let's look at how much people with various educational qualifications earn on average.
Individuals without vocational training (e.g., without an apprenticeship, vocational program, or university degree) currently earn an average of around 2,200 euros net per month (for a full-time position). What do you think is the average monthly net income for individuals pursuing the following educational paths at age 20?

No degree/qualification: ___ €
Vocational training: ___ €
Master Craftsman/Technician: ___ €
University degree: ___ €

[same screen]
Now, consider these same individuals at age 40. What do you think is the average monthly net income for 40-year-olds with the following educational backgrounds?

No degree/qualification: ___ €
Vocational training: ___ €
Master Craftsman/Technician: ___ €
University degree: ___ €”

Finally, we include a question on how confident respondents are in their answer. This is displayed on the same screen, where they report from 1 (very unsure) to 10 (very sure) their confidence in accuracy of their responses.

We are also interested in heterogeneity with respect to a series of perceived mechanisms that may help explain education preferences. These include the extent to which respondents agree that an education path imposes a financial burden on the family, enables individuals to earn income early in the career, offers high expected long-term earnings, entails a high risk of dropout or unemployment, provides opportunities for career advancement, facilitates finding employment after completion, aligns socially with family background, confers social status, is expected to be enjoyable, and allows individuals to pursue their own interests.

For heterogeneity by perceived mechanisms, respondents are asked to evaluate the following statements for each educational pathway:

Question: “To what extent do you agree with the following statement:”
• “This educational pathway places a strong financial burden on the family if the child pursues it.”
• “This educational pathway allows individuals to earn their own income early in their career.”
• “This educational pathway leads to a high income in the long run.”
• “There is a high risk of dropping out of this educational pathway.”
• “There is a high risk of becoming unemployed in this educational pathway.”
• “This educational pathway offers good opportunities for career advancement.”
• “After completing this educational pathway, it is easy to find a job quickly.”
• “This educational pathway fits well with my family and my usual social environment.”
• “This educational pathway enjoys high social status.”
• “Most people who pursue this educational pathway enjoy it.”
• “This educational pathway allows individuals to pursue their own interests.”

Responses are collected using a matrix format, where each statement is evaluated separately for vocational education and university education. For each item, respondents indicate their level of agreement on a five-point Likert scale ranging from “strongly agree” to “strongly disagree”.

For the parent sample, we additionally ask a subset of items for their own child:
• “If my child pursues this educational pathway, it would place a strong financial burden on my family.”
• “If my child pursues this educational pathway, it would lead to a high income for him/her in the long run.”
• “If my child pursues this educational pathway, there is a high risk that he/she drops out of it.”
• “If my child pursues this educational pathway, he/she would enjoy it.”
• “If my child pursues this educational pathway, he/she could follow his/her own interests.”

We will further test whether preferences for education pathways depend on respondents' time and risk preferences, or their understanding of the information. We measure time and risk preferences through survey items in the background questionnaire, as well as their understanding of the provided information through a comprehension check.

We measure risk preferences using two complementary questions. The first one elicits a general self-assessment of willingness to take risks. Respondents are asked:

“How would you describe yourself? Are you generally a person who is willing to take risks, or do you try to avoid risks?
Please select a box on the scale, where “0” means ‘not at all willing to take risks’ and “10” means ‘very willing to take risks’. You can use the values in between to indicate intermediate levels of risk willingness.”

The second measure consists of a sequence of hypothetical choice tasks designed to elicit risk preferences in a structured setting. The staircase procedure works as follows. First, each respondent is asked whether they would prefer to receive 160 euros for sure or whether they preferred a 50–50 chance of receiving 300 euros or nothing. In case the respondent opts for the safe choice, the safe amount of money being offered in the second question decreases to 80 euros. If, on the other hand, the respondent opts for the gamble, the safe amount is increased to 240 euros. In the subsequent questions, the amount of certain payment is adjusted upward after a lottery choice and downward after a sure-payment choice, following the same branching logic.

Similarly, we measure respondent’s time preferences (discount rate) using two complementary questions. The first one elicits a general self-assessment of willingness to act. Respondents are asked:

“Are you someone who is generally willing to give something up today in order to benefit from it in the future, or are you not willing to do so?
Please select a box on the scale, where “0” means ‘not at all willing’ and “10” means ‘very willing’. You can use the values in between to indicate intermediate levels.”

The second measure consists of a sequence of hypothetical choice tasks designed to elicit time preferences in a structured setting. Respondents are presented with five interdependent choices between receiving a payment today and receiving a payment in 12 months. The payment today remains fixed across all five situations, while the payment in 12 months varies according to a pre-specified branching sequence.

The staircase procedure works as follows. First, each respondent is asked whether they would prefer to receive €100 today or €154 in 12 months. If the respondent opts for the payment today, the payment in 12 months is adjusted upwards to €185 in the second question. If, on the other hand, the respondent chooses the payment in 12 months, the corresponding payment is adjusted downwards to €125. Working further through the questions follows the same logic.

There are two comprehension checks, where respondents are randomly given either one of them.

Comprehension check 1: Imagine that Leon faces the same decision and must choose between these two career paths (Option 1 and Option 2). It is important to him that he earns a good income later in his professional life, even if this means getting by on very little during the first few years after finishing school.

Based on the previous chart: Which of the two career paths would you recommend to him?

[Treatment screen]:
o Option 1: Vocational education (apprenticeship)
o Option 2: University education

Comprehension check 2: Imagine Paul faces the same decision and has to choose between these two career paths (Option 1 and Option 2). It is particularly important to him to earn as much as possible during the very first years after finishing school (at around age 20).

[Treatment screen]:
o Option 1: Vocational education (apprenticeship)
o Option 2: University education

As respondents complete the comprehension check, they have the option to view the income trajectories graph again if they wish. Moreover, at the end of each comprehension check, a confidence question is included, which is displayed on the same screen, and asks respondents to report from 1 (very unsure) to 10 (very sure) their confidence in accuracy of their responses.

In addition, we will report results separately for the oversample of parents with children aged 0–6. For heterogeneity by parental status, the sample includes an oversample of 1,000 parents with children aged 0–6, which allows for comparisons between the general population and parents of young children.

We conduct the experiment in a sample of 3,000 adults as well as 1,000 parents of children aged 0 to 6. The sampling and polling is done in cooperation with Talk Online Panel Requests, who use an online-access panel to recruit respondents. The questions are answered by respondents independently, without interviewer contact.

Not available

Randomization is carried out by the survey company Talk Online Panel Requests on questionnaire level.

At the individual level

No

The sample is not clustered but at the individual level.

Main sample of 4,000 adults aged 18 and older: 3,000 respondents (general population) and an oversample of 1,000 parents with children aged 0-6.

1,500 respondents in each treatment arm (main sample), 500 respondents in each treatment arm (parents oversample)