Scoring deliberation practices without moral rankings

Research context

Method-bias and construct-development reviews distinguish measurement questions from a score’s tidy appearance.

Machiavelli development analysis

Each MED domain has ten original statements on a one-to-five agreement scale, and the current software requires all sixty answers before producing a report. Two opposite-keyed items per domain use six minus the response; the mean of all ten keyed answers is displayed as round((mean − 1) × 25). There is no current skip option or eight-item scoring threshold. An optional-omission version with an eight-of-ten rule could be investigated separately, but it is neither implemented nor validated. Researchers would need to test whether omitted stakeholder or review situations change the process being represented. Declining to complete a form is not evidence of indifference to ethics.

The meaning of a larger score is narrow: stronger endorsement of the proposed process habits. It does not establish that the participant reaches better conclusions, behaves more ethically, or possesses greater moral worth. A high option-exploration score could reflect thoughtful flexibility or an aspirational description. A lower process-recording score could reflect informal work where written records are uncommon. The report should invite context before interpreting either pattern, and should never convert a domain mean into a percentage probability of ethical conduct.

A total score would be particularly misleading. It would imply that all six practices form a single quantity and that strengths in one can compensate for omissions in another. Those assumptions are untested. It could also encourage people to maximize procedural activity even when a simple decision is appropriate. The proposed report should keep domains separate, explain their boundaries, and offer one optional practice. There is no ethical intelligence quotient, pass mark, certification level, or normative rank in the development version.

Precision must be established empirically. A long questionnaire can still contain ambiguous or redundant items, and a high internal-consistency estimate can coexist with poor construct coverage. Future studies should examine domain reliability with uncertainty, test alternative structures, and investigate whether reverse items produce a wording effect. Repeated scores should not be interpreted as growth until stability and meaningful-change questions have been studied. A participant may answer differently after learning the framework’s vocabulary without changing their decision process in practice.

The technical record should preserve the exact version, item key, missingness policy, and interpretation text. Automated checks can confirm that the scoring software handles reversed items and incomplete domains correctly. They cannot establish that the score measures the intended process. Any future external relationship should identify the criterion, sample, method, and uncertainty. Until those studies exist, MED’s scores are descriptive summaries within an experimental reflection tool, and the report should make that status as visible as the chart itself.

The arithmetic should also be demonstrated with a fictional response example rather than a real participant’s data. Showing how one opposite-keyed answer contributes to a domain can improve transparency without suggesting that the example represents a norm or a desirable response pattern.

A score explanation should identify what the instrument leaves unobserved: the actual decision, the information available at the time, the obligations involved, and the effects on others. Keeping these omissions visible makes it harder for a reader to mistake a self-report summary for an objective evaluation of ethical competence.