RESEARCH IN DEVELOPMENT · SEPTEMBER 2026

Start with a question.
Earn the answer.

Does similarity in how partners solve problems explain later relationship outcomes—and does it contribute more than other relationship facets?

What existing research tells us

Weidmann and colleagues studied 1,294 couples and found that personality similarity was not robustly associated with relationship or life satisfaction. Their discussion points to situation perceptions, values, attitudes, and goals as directions for further work. The study was cross-sectional and its predominantly older, long-established sample does not establish premarital predictive validity. Read Weidmann et al. (2023).

Across 43 longitudinal datasets, Joel and colleagues found that relationship-specific perceptions predicted relationship quality, while change in quality was largely unpredictable from the baseline self-report variables they examined. That is a reason to test new models carefully and compare them with strong existing baselines. Read Joel et al. (2020).

These findings motivate the project. They do not validate its proposed questions.

A focused hypothesis, with a broader comparison

Our central hypothesis is that partners’ problem-solving similarity may be the strongest contributor to prediction among the facets studied. A stronger version proposes that other facets add little useful predictive information once this similarity is known. Neither claim is established. The design must allow the results to challenge both.

The new core module has 24 original candidate questions across four areas: decision choices, plan evaluation, judgments, and interpretations of events. Both partners would independently answer the same third-person scenarios with the same information. Agreement describes similarity in these responses; it does not establish that either answer is correct.

A separate 60-question broader module retains the original twelve facets with expanded, context-rich scenarios: problem framing, information use, uncertainty, planning, follow-through, flexibility, negotiation, conflict repair, support, autonomy, resource fairness, and shared direction. Some overlap with the core is expected and must be examined rather than assumed away.

All 84 distinct questions are unvalidated candidates. They are not a finalized test, and their response options are categories with no established numerical order or scoring key. Review the full question library, or jump to its problem-solving core and broader comparison.

Original scenario questions will be considered alongside established measures of communication, dyadic coping, personality, and relationship satisfaction. Existing measures will retain their original wording and scoring where required; any adaptation will need permission and fresh validation.

Situations with meaningful trade-offs

The current candidates place decisions in concrete situations: a trip with a fixed budget, an accident at a formal gathering, competing commitments, or an ambiguous message. Each includes enough shared information to make the trade-off clear. A question targets one action, judgment, interpretation, or priority at a time. When two questions share an event, their common context must be accounted for in analysis.

Choosing a plausible response to a hypothetical situation may differ from acting under real pressure. The study must examine that gap. Detailed wording alone does not demonstrate that an item reveals a stable personality trait or predicts a relationship outcome.

Four sessions, balanced forms

The draft design uses four forms across approximately two weeks. Each planned session contains 26 candidate questions: eight core and eighteen broader. Across four sessions, that is 104 question presentations per person, drawn from 84 distinct questions. Brief state questions about mood and stress are additional. The pilot will test burden and timing.

Both partners receive the same form on each occasion and answer independently. Form order will be balanced across couples so that a particular form is not always confused with a particular day. Repeated anchor questions link sessions; they are repeated observations, not new independent questions or additional couples.

Consistency is evidence about reliability. It does not prove that a self-description is accurate, that a preference is correct, or that a couple will thrive. Those are separate research questions.

Compare the full questionnaire with the focused module

The planned comparison uses the same couples, outcomes, follow-up times, and evaluation partitions. A baseline model will be compared with three additions: the core problem-solving module only, the broader module only, and both modules together. This asks whether the core adds useful prediction and whether the broader module adds information beyond it.

Restricting analysis to the core questions is an item-domain analysis, not a subgroup of participants. Any comparisons between participant groups would require separate, prespecified hypotheses and adequate sample sizes.

The analysis must distinguish partner similarity from each partner’s response tendencies and from common answer choices. It must account for partners being paired, questions being repeated, and multiple comparisons. A nonsignificant result for another facet would not establish that it has no effect. Claims that the core is sufficient require predefined margins for a meaningfully small loss in prediction and independent validation.

Outcomes defined before analysis

The proposed primary endpoint is each partner's relationship satisfaction at 24 months among relationships still ongoing. Relationship dissolution will be analyzed separately. Six-, twelve-, and thirty-six-month follow-ups will help characterize change and timing. Separation will not be assigned an artificial satisfaction score.

A lasting relationship and a satisfying relationship are different outcomes. Neither alone is a complete definition of a successful marriage.

Validation milestones

1. Content and comprehension: expert assessment and cognitive interviews to test whether questions mean what we intend.
2. Psychometric pilot: estimate reliability, item behavior, form comparability, and state variation; examine whether measurement works comparably across relevant groups.
3. Prospective cohort: test preregistered hypotheses with both partners and follow-up outcomes.
4. External validation: test any frozen prediction model in an independent cohort, including calibration and uncertainty.

Both partners must remain in the same development or validation partition. Sample size will be justified using simulations and the number of outcome events, rather than a generic participants-per-question rule.

Participation and privacy

Before recruitment, the study requires an accountable investigator, ethics review, separate informed consent, a participant support process, a retention policy, and a secure research data system. Raw partner responses will not be shared with the other partner by default. Research participation will remain voluntary.

This public prototype does not collect questionnaire responses. Read the prototype privacy notice.

Methods that guide the plan

• COSMIN measurement standards
• TRIPOD reporting guidance for prediction models
• Canada's TCPS 2 research ethics policy
• Communication Patterns Questionnaire validation
• Dyadic Coping Inventory validation

This is a project overview, not a published protocol or a systematic review. The research repository retains the detailed evidence inventory, search limitations, draft analysis plan, and outstanding decisions.