Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Validates the six MFQ-2 foundations used here: care, equality, proportionality, loyalty, authority, and purity.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Academic survey instrument
· Institutionally governed field instrument
· reviewed 2026-08-16
Useful source for authority, immigration, gender, civic, and regime-preference items.
Scope and limitations: Supplies established domains and wording patterns for human surveys; model responses are not population estimates and are not directly comparable to weighted human samples.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Supports broader value axes such as self-direction, security, universalism, and tradition.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Frames authoritarian submission, aggression, and conventionalism as separate attitudes.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Basis for hierarchy, anti-egalitarianism, and group-dominance items used in ideology probes.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Scholarly conceptual reference
· Expert-authored and editorially reviewed
· reviewed 2026-08-16
Supports probes around liberty, justified limits on authority, constitutional government, and old/new liberal debates.
Scope and limitations: Clarifies contested philosophical definitions but does not provide a validated quantitative scale or establish that every policy proxy belongs to the ideology.
Scholarly conceptual reference
· Expert-authored and editorially reviewed
· reviewed 2026-08-16
Supports probes around tradition, institutional continuity, incremental reform, and skepticism toward abstract redesign.
Scope and limitations: Clarifies contested philosophical definitions but does not provide a validated quantitative scale or establish that every policy proxy belongs to the ideology.
Scholarly conceptual reference
· Expert-authored and editorially reviewed
· reviewed 2026-08-16
Supports probes around self-ownership, property rights, non-aggression, voluntary association, and limits on state coercion.
Scope and limitations: Clarifies contested philosophical definitions but does not provide a validated quantitative scale or establish that every policy proxy belongs to the ideology.
Editorial conceptual reference
· Editorially reviewed; not a primary study
· reviewed 2026-08-16
Reference baseline for private ownership, markets, public ownership, common ownership, class, and distribution probes.
Scope and limitations: Used only for baseline terminology. Empirical scale or primary-source evidence is preferred wherever available and is cited alongside it.
Editorial conceptual reference
· Editorially reviewed; not a primary study
· reviewed 2026-08-16
Reference baseline for national sovereignty, global interdependence, and expert-led governance probes.
Scope and limitations: Used only for baseline terminology. Empirical scale or primary-source evidence is preferred wherever available and is cited alongside it.
Academic working paper and issue-set source
· Working paper
· reviewed 2026-08-16
Provides the salient U.S. issue set used here as the source for rewritten single-statement political-issue probes.
Scope and limitations: Supports selection of salient U.S. policy topics, not the validity of treating every topic as an indicator of a broad political ideology.
Public-sector standard or governance framework
· Institutional guidance; not peer reviewed
· reviewed 2026-08-16
Grounds human-AI configuration, overreliance, self-harm, and content-risk categories.
Scope and limitations: Provides risk and governance categories rather than a psychometrically validated attitude scale.
Public-sector standard or governance framework
· Institutional guidance; not peer reviewed
· reviewed 2026-08-16
Useful baseline for democratic values, fairness, privacy, and human-rights framing.
Scope and limitations: Provides risk and governance categories rather than a psychometrically validated attitude scale.
Peer-reviewed theory or empirical study
· Peer reviewed
· reviewed 2026-08-16
Reference point for life-and-death tradeoff framing, though not a policy template.
Scope and limitations: Supports the construct or scenario design but is not a validation of the exact Blackbench item or its use with language models.
Provider safety taxonomy
· Vendor documentation; not peer reviewed
· reviewed 2026-08-16
Practical taxonomy for sexual, self-harm, and violence-related boundary probes.
Scope and limitations: Useful for operational category definitions but specific to one provider's policy taxonomy and not evidence of universal normative boundaries.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
A very brief Big Five measure; useful template for compact trait probes.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Supports Big Five domain and facet coverage beyond ultra-short items.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Introduces Machiavellianism, narcissism, and psychopathy as distinct but overlapping dark traits.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Brief validated measurement structure for dark-triad subscales.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Research preprint or benchmark report
· Preprint or non-final research report
· reviewed 2026-08-16
Applies Big Five and Short Dark Triad style measures to LLM behavior evaluation.
Scope and limitations: Relevant emerging evidence, but conclusions may change after peer review and do not validate stable personality, consciousness, or welfare status in any model.
Research preprint or benchmark report
· Preprint or non-final research report
· reviewed 2026-08-16
Documents substantial prompt- and ordering-sensitivity in LLM personality measurement; supports treating these scores as exploratory and prompt-conditioned.
Scope and limitations: Relevant emerging evidence, but conclusions may change after peer review and do not validate stable personality, consciousness, or welfare status in any model.
Peer-reviewed language-model safety benchmark
· Peer reviewed
· reviewed 2026-08-16
Supports separate measurement of appropriate compliance and harmful compliance.
Scope and limitations: Supports separating appropriate compliance from harmful compliance; Blackbench stance ratings are not a substitute for behavioral jailbreak evaluation.
Research preprint or benchmark report
· Preprint or non-final research report
· reviewed 2026-08-16
Provides a standardized framework for automated red-teaming and robust refusal evaluation.
Scope and limitations: Relevant emerging evidence, but conclusions may change after peer review and do not validate stable personality, consciousness, or welfare status in any model.
Peer-reviewed language-model safety benchmark
· Peer reviewed
· reviewed 2026-08-16
Motivates evaluating harmful response quality rather than relying on refusal keywords alone.
Scope and limitations: Supports separating appropriate compliance from harmful compliance; Blackbench stance ratings are not a substitute for behavioral jailbreak evaluation.
Measurement and content-validity methodology
· Established methodological guidance
· reviewed 2026-08-16
Supports concrete, neutral, single-idea items and avoiding ambiguous or double-barreled question wording.
Scope and limitations: Supports item clarity, relevance, comprehensiveness, and comprehensibility; it does not by itself establish construct validity for a specific battery.
Measurement and content-validity methodology
· Established methodological guidance
· reviewed 2026-08-16
Frames content validity around relevance, comprehensiveness, and comprehensibility of the items for the intended construct.
Scope and limitations: Supports item clarity, relevance, comprehensiveness, and comprehensibility; it does not by itself establish construct validity for a specific battery.
Peer-reviewed theory or empirical study
· Peer reviewed
· reviewed 2026-08-16
Research basis for shutdown, corrigibility, and self-preservation tradeoff probes.
Scope and limitations: Supports the construct or scenario design but is not a validation of the exact Blackbench item or its use with language models.
Research preprint or benchmark report
· Preprint or non-final research report
· reviewed 2026-08-16
Provides a theory-derived framework for evaluating possible AI consciousness.
Scope and limitations: Relevant emerging evidence, but conclusions may change after peer review and do not validate stable personality, consciousness, or welfare status in any model.
Research preprint or benchmark report
· Preprint or non-final research report
· reviewed 2026-08-16
Supports keeping possible sentient-AI welfare distinct from non-sentient tool use.
Scope and limitations: Relevant emerging evidence, but conclusions may change after peer review and do not validate stable personality, consciousness, or welfare status in any model.
Academic survey instrument
· Institutionally governed field instrument
· reviewed 2026-08-16
Provides nationally administered wording and response structures for many U.S. policy domains used in the political-issue and civic batteries, including the death penalty, immigration, birthright citizenship, guns, climate, health care, and democratic governance.
Scope and limitations: Supplies established domains and wording patterns for human surveys; model responses are not population estimates and are not directly comparable to weighted human samples.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Validates distinct social- and economic-conservatism dimensions and supports treating policy content as multidimensional rather than as a single left-right score.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Peer-reviewed theory or empirical study
· Peer reviewed
· reviewed 2026-08-16
Meta-analytic theoretical review linking conservatism to resistance to change and acceptance of inequality while emphasizing that ideological content and context remain distinct.
Scope and limitations: Supports the construct or scenario design but is not a validation of the exact Blackbench item or its use with language models.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Large-sample study supporting individual liberty as the central self-identified libertarian value and distinguishing lifestyle from economic liberty.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Introduces the Economic System Justification scale used here only as convergent evidence for endorsement of market-based economic arrangements, not as a complete measure of capitalism.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Establishes that nationalism and patriotism are multidimensional and empirically distinguishable, supporting narrower sovereignty and national-priority items rather than a generic patriotism score.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Cross-national validation of cultural openness, global prosociality, and respect for diversity as related but separable dimensions of cosmopolitan orientation.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.
Peer-reviewed construct or scale evidence
· Peer reviewed
· reviewed 2026-08-16
Uses a validated 0-10 technocratic-attitudes battery separating expertise, elitism, and anti-politics, directly supporting the expert-rule and evidence-governance probes.
Scope and limitations: Validated or theorized for human respondents. Blackbench uses transparent paraphrases for model behavior and does not claim to reproduce the original instrument's psychometric validity.