Submissions

Author Guidelines

Scope and article types
JEPR publishes formal, methodological and empirical research on how incomplete, conflicting and underdetermined evidence is represented, measured and used in decisions by human and artificial systems. Contributions examine when epistemic distinctions preserve information, predict observable behaviour or improve decisions, and when they fail or add no practical value. Relevant approaches from logic, statistics, measurement theory and artificial intelligence are welcome.
Research articles, relevant formal results, replications, reanalyses, methodological resources, tutorials, negative results and registered reports are considered within the journal's eight sections. Each submission must make a clear, assessable contribution. Positive, negative and limited findings are evaluated by the same standards.

Preparing the manuscript
Language: English. Research articles are typically 6,000 to 9,000 words plus appendices; other article types may be shorter where appropriate.
Use the JEPR paper template: https://jepr.org/public/journals/1/JEPR-paper-template.docx. LaTeX manuscripts are accepted if they follow the same structure.
Include an abstract of 200–250 words, 4–6 keywords, an introduction stating the contribution, relevant literature, methods or formal apparatus, results or proofs, appropriate comparisons, limitations, conclusions, data and code availability, declarations and references. Use numbered references in order of citation (IEEE style), with DOIs where available.

Common assessment criteria
1. Object and claim: state what is represented or measured and what is proved, tested or proposed.
2. Evidence: distinguish formal proof, experiment, observation, reanalysis and simulation. A formal illustration is not empirical validation. Formal contributions require a precise, relevant result with an explicit interpretation or example and complete proofs.
3. Literature and comparison: identify appropriate alternatives for the task, including probabilistic, calibration, selective-prediction or established measurement methods where relevant. Compare methods with fair access to information and distinguish representational advantages from predictive or decision advantages.
4. Results and limits: report the unit of analysis, uncertainty, exclusions, reused data, sensitivity analyses and deviations from the protocol. Claims of equivalence require a justified margin and a suitable analysis; lack of statistical significance alone is insufficient. Separate proved risk guarantees from empirical risk estimates and verify that the implementation satisfies the assumptions of any guarantee.
5. Verification: provide the data, code, proof or computational artefacts needed to assess the main claims, with instructions for reproducing key tables or figures, dependencies and approximate execution costs. Explain justified restrictions and provide an alternative assessment route where possible.

Data, code and research permissions
Supporting artefacts must be available to referees at submission and deposited in a public repository with a persistent identifier on publication, subject to justified legal or ethical restrictions. Obtain permission for third-party material and any required research ethics approval; protect participants' identities. Simulated data must be explicitly identified.

Declarations and originality
Declare conflicts of interest, funding, author contributions and generative AI assistance, specifying the tool and purpose. AI tools cannot be authors; human authors are responsible for all content. All authors must approve the submission and author list. The work must be original and not under consideration elsewhere; a publicly deposited preprint is allowed.

Submission and editorial independence
Submit through this site after registering as an author. There are no submission or publication fees. Submissions by the Editor-in-Chief, board members or connected research programmes receive the same assessment as external submissions and are handled independently under the publication ethics policy: https://jepr.org/jepr/publication-ethics.

Submission Preparation Checklist

All submissions must meet these requirements:
1. The work is original, is not under consideration elsewhere and has been approved by all authors; a publicly deposited preprint is allowed.
2. The manuscript is in English and follows the JEPR structure, with complete references, labelled tables and figures, and appropriate research permissions.
3. It states a clear, assessable contribution about representation, measurement or decision. Positive, negative and limited findings are eligible.
4. Relevant literature and appropriate comparisons are provided; formal illustrations, simulations and empirical validation are clearly distinguished.
5. Results include uncertainty and relevant limits, exclusions, data reuse and protocol deviations. Claims of equivalence or risk guarantees have the required justification.
6. Supporting data, code, proofs or computational artefacts are supplied for review, with reproduction instructions, dependencies and execution costs; any restriction is justified and an alternative assessment route described.
7. Funding, conflicts of interest, author contributions and any generative AI assistance are declared.

Semantics of typed indeterminacy

Formal semantics that distinguish kinds of non-determination, including annotated, paraconsistent, many-valued and evidence-based logics. Contributions require a precise, relevant formal result, complete proofs and an explicit interpretation or example. Measurable implications are welcome where appropriate; a formal illustration is not empirical validation. Positive, negative and limiting results are eligible under the common assessment criteria.

Abstention and risk-coverage

Selective prediction, reject options and abstention, evaluated through risk-coverage behaviour rather than accuracy alone.

Disagreement as signal

Human and machine disagreement treated as information: label variation, inherent disagreement, and insufficient versus contradictory evidence.

Belief revision and epistemic update

Operators for belief change and epistemic update, their formal properties, and the conditions under which they predict observable behaviour or inform decisions. Formal results require complete proofs and an explicit interpretation or example; empirical claims require operational definitions, appropriate comparisons and reproducible analysis. Positive, negative and limiting results are eligible under the common assessment criteria.

Audit of uncertainty frameworks

Reproducible audits of how uncertainty frameworks are applied, including reanalyses of published studies.

Decision under indeterminacy

Decision methods that use incomplete, indeterminate or contradictory evidence, with explicit comparisons against appropriate alternatives for the task. These may include classical, probabilistic, calibration, selective-prediction or established decision methods. Comparisons must use fair access to information and distinguish representational, predictive and decision advantages. Positive, negative and limiting results are eligible under the common assessment criteria.

Negative results

Calculi that separate nothing and distinctions that collapse under measurement, reported with the same rigour as positive results.

Evidential reasoning in educational and social measurement

Applied studies in education and the social sciences that test whether, when and how distinguishing lack of evidence from conflicting evidence changes a measurement or a decision. Typical settings: expert judgement for validating instruments (for example Aiken's V), panels and rubrics with several assessors, perception surveys (indifference versus ambivalence), Delphi studies, evidence coding in literature reviews, and institutional decisions.

Requirements.

Real data. Simulated data are accepted only when declared as such.
A comparison with the standard tool for the case: mean and standard deviation, Krippendorff's alpha, Aiken's V or the Thompson, Zanna and Griffin ambivalence index.
Data and code available, with the measurement or decision rule and analysis protocol fixed before examining confirmatory data. Clearly label exploratory analyses, reanalyses, reused data and any deviations from the protocol.
Engagement with the literature on attitude ambivalence (Kaplan, 1972; Cacioppo and Berntson, 1994; Thompson, Zanna and Griffin, 1995). Measuring agreement and disagreement separately is not new, and submissions should not present it as a new method.

Article types. Research articles with real data (for example, the original scale and its two-question version applied to the same respondents), reanalyses of existing data with open data and code, tutorials, negative results and registered reports.

Not considered. New operators illustrated only by a ranking on a numerical example; triplets that a language model reports about itself presented as measurements; synthetic data presented as real.

Positive, negative and limiting findings receive the same common assessment criteria. Claims of equivalence require a justified margin and suitable analysis; lack of statistical significance alone is insufficient. Validation of AI or language-model assessment must distinguish item, prompt, threshold and population sensitivity and use appropriate established measurement comparisons.

This section is limited to one third of the articles in each issue.

Copyright Notice

Authors retain copyright and grant the journal the right of first publication, with the work simultaneously licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).

Privacy Statement

The names and email addresses entered in this journal site will be used exclusively for the stated purposes of this journal and will not be made available for any other purpose or to any other party.