Preferences over Relative Income within the Household: Pre-Analysis Plan

We conduct representative online surveys in Germany to study preferences over relative income within couples. Participants complete a background module and are then assigned to one of three survey modules: (i) a qualitative satisfaction survey, (ii) a quantitative choice survey experiment, or (iii) a vignette-based beliefs survey experiment. These modules are designed to study the qualitative form, quantitative magnitude, and potential mechanisms of preferences over relative income within couples.

2026-04-24

2026-05-08

Primary outcomes are module-specific. In the qualitative satisfaction survey, the primary outcome is stated satisfaction with hypothetical within-couple relative income splits. In the quantitative choice survey, the primary outcome is an indicator for choosing the equal-split option in each binary choice item. In the vignette-based beliefs survey, the primary outcomes are stated beliefs about the consequences of different within-couple relative income configurations, including beliefs about housework and childcare, joint leisure time, marital satisfaction, and divorce risk.

In the qualitative module, respondents report satisfaction on a 7-point Likert scale for 11 hypothetical income configurations; the main analysis standardizes this satisfaction measure and studies standardized satisfaction as a function of the respondent’s own share in household income. In the quantitative module, the outcome equals one if a respondent chooses the equal-split option for a given scenario item. These item-level choices are used to estimate preference coefficients on total household income and on relative income when the respondent is the secondary earner and when the respondent is the primary earner. In the vignette module, outcomes are stated beliefs about secondary impacts and costs of different relative-income configurations.

econdary outcomes include respondents’ beliefs in the quantitative satisfaction survey about what the researchers intend to find in the study and respondents’ perceptions of prevailing gender norms regarding the relative earnings of wives and husbands.

These secondary outcomes are collected after the quantitative choice items and are used descriptively and as manipulation checks.

The study consists of online survey experiments conducted in Germany. After completing a background information survey module, respondents are randomly assigned to one of three survey modules: a qualitative satisfaction survey, a quantitative choice survey experiment, or a vignette-based beliefs survey experiment. In the qualitative module, respondents indicate how satisfied they would be with 11 hypothetical within-couple income configurations while holding total household income fixed. In the quantitative module, respondents complete 14 binary choice items in which they are asked to pick between two hypothetical income situations in which total household and partners' income vary. One of the choices always includes an unequal relative income split whereas the other choice always includes a 50--50 split. Within this module, respondents are additionally randomized into a main arm (baseline treatment), a ceteris paribus treatment arm, a positive demand treatment arm, or a negative demand treatment arm. In the vignette module, respondents evaluate hypothetical couples and report beliefs about outcomes such as housework, childcare, time together, marital satisfaction, and divorce risk. Further details are provided in the pre-analysis plan.

Not available

Randomization done by computer within the online survey. Respondents are randomly assigned at survey start to one of the three survey modules. Within the quantitative module, respondents are randomly assigned to one of four treatment arms, and scenario parameters are randomized at the respondent level.

Individual respondent. Randomization occurs at the respondent level for survey-module assignment and treatment-arm assignment within the quantitative module. Additional scenario parameters are randomized at the respondent level.

No

3,000 individual respondents.

3,000 respondents overall: 1,500 in the quantitative choice survey; 750 in the qualitative satisfaction survey; and 750 in the vignette-based beliefs survey.

Quantitative choice survey: 1,500 respondents total (with Main arm: 750; Ceteris paribus treatment: 250; Positive demand treatment: 250; Negative demand treatment: 250); Qualitative satisfaction survey: 750; Vignette-based beliefs survey: 750

Pre-Analysis Plan