Methodology version 1.1 · Congressional Voting Score Protocol version 0.1, pilot
November, 2026
The Civic Outcomes Standard (COS) is sherafy.com’s method for evaluating public choices. We state the outcomes we value, investigate what the available choices are likely to produce, and explain the evidence and tradeoffs behind our judgment. Party affiliation, appealing intentions and an impressive policy title do not establish public benefit.
COS applies to elections, ballot measures, officeholders and public policy. It combines an explicit public-interest baseline with evidence that readers can inspect and challenge. The congressional protocol below applies those principles to a limited voting-alignment measure; it is a pilot method, not a validated rating of a politician’s overall worth.
Our public-interest baseline
We favor conditions that allow people to live healthy, safe, free, materially secure and meaningful lives, while protecting the institutions and resources future generations will need. Growth, productive enterprise and innovation matter alongside opportunity, civil liberties, environmental conditions and accountable government.
Neither spending nor cutting spending demonstrates success by itself. Markets and public institutions must be evaluated by their results, costs and capacity. Concentrated hardship, barriers to opportunity and excessive concentrations of power deserve examination even when an aggregate indicator improves.
These are stated values, not discoveries produced by a formula. Nonpartisanship means applying the same evidentiary standards regardless of whose position benefits. Readers may disagree with our priorities and still use the underlying analysis.
COS retains the evidence discipline it draws from the Objectivity AI framework developed by Fabled Sky Research: trace claims to their basis, distinguish knowledge from inference, and make conclusions revisable. Our Editorial Standards govern the published work.
Establish what the evidence can support
Separate six kinds of statement:
- Verified fact: a specific proposition supported by an appropriate, checked record.
- Well-supported finding: a conclusion sustained by credible research or converging evidence, with its scope stated.
- Attributed or disputed claim: an assertion whose truth remains unestablished or materially contested. An allegation is not a finding.
- Reasonable inference: a conclusion drawn from supported premises; identify the connecting reasoning and strongest alternative.
- Unknown: a question the available evidence cannot responsibly resolve.
- Value judgment: a choice about priorities, acceptable risks or competing interests.
Confidence is separate from these categories. Describe it as high, moderate, low or indeterminate, with a reason. High confidence requires direct or convergent, relevant evidence with no unresolved weakness likely to reverse the finding. Moderate confidence allows material limitations while retaining a supported direction. Low confidence means fragile support; indeterminate means no defensible direction. These labels are judgments, not numerical probabilities.
Match sources to questions. Official records and applicable legal text establish actions and rules. Relevant research, including credible systematic reviews, informs causal effects. Evidence about setting, population and administrative capacity determines whether findings transfer. An original statement establishes what its author said, not whether it is true.
Assess methods, relevance, independence and limitations rather than applying a universal source ranking. Repetition of one claim is not independent corroboration. Interested sources can supply valid evidence; their factual claims still need scrutiny. Our Source Evaluation & Evidence policy explains this approach.
Compare the choices people actually face
Identify the decision, the responsible institution, its powers, and the applicable text or policy version. Explain the causal path from the proposed action to the claimed result.
Compare the proposal with a realistic alternative, including continuation of existing policy. For a vote, establish what adopting or rejecting the exact question would do. A No vote does not automatically enact a better proposal; a Yes vote does not guarantee successful implementation. Claims that a different bargain was available require evidence.
Separate passage, implementation, observed outcomes and attributable effects. The GAO distinction between outcome and impact evaluation matters: a change after a program begins is not, by itself, proof that the program caused it.
Examine all ten civic lenses
Consider every lens, recording why it is material or not applicable. Give consequential effects space; do not manufacture detail for an irrelevant category.
| Lens | Central question |
|---|---|
| Human welfare | What changes in health, safety, housing, education, access, environmental exposure or daily life, and how severely? |
| Distribution and inequality | Who gains and loses across incomes, communities and generations? Are benefits or burdens concentrated? |
| Civil liberties and equal treatment | Are rights, due process, privacy and equal treatment protected? Is coercion justified and subject to review? |
| Economic and material effects | What happens to jobs, wages, prices, productivity, competition, investment and household security? |
| Fiscal reality and opportunity cost | What resources are required, who finances them, and what other uses are displaced? |
| Institutional integrity and democratic accountability | What changes in transparency, oversight, corruption risk, lawful constraints and the public’s ability to hold power accountable? |
| Evidence of effectiveness | Does credible evidence support the mechanism, at the relevant scale and in the relevant setting? |
| Implementation and administrative capacity | Can the responsible institution deliver, fund and enforce the policy, and can people actually access it? |
| Unintended consequences and behavioral response | How might people, firms and institutions adapt, evade, substitute or exploit the rules? |
| Reversibility, resilience and future lock-in | How durable are the effects, how costly is reversal, and what happens if assumptions fail? |
These are overlapping analytical lenses, not ten independent point totals. Record each underlying consequence once and cross-reference it where relevant. For example, lower medical bills can affect welfare, household finances and distribution without constituting three separate benefits.
For each material effect, identify direction, likely magnitude, affected population, distribution, timing, evidence and confidence. Use numbers where justified; explain important effects that cannot responsibly be quantified. The OECD well-being framework provides useful context for examining conditions beyond aggregate economic output.
Reach and test a judgment
Start with legal and rights constraints. Distinguish the law as it stands from an argument to change it. Do not assume that aggregate gains justify an arbitrary deprivation of rights. When rights conflict, explain necessity, proportionality, alternatives and safeguards; unresolved legal status must remain visible.
Then compare the material benefits and harms, their distribution, feasibility and uncertainty. State the considerations that decide the case and the value judgment that gives them priority. For a mixed package, assess the whole choice rather than selecting its most attractive provision.
The available conclusions are favor the proposal, oppose it, or unresolved. For a binary vote, these become favor Yes, favor No, or unresolved. Distinguish unresolved evidence from unresolved value conflicts. A supported conclusion need not wait for certainty, but neither should an unresolved case be forced into a binary verdict.
Apply four checks:
- Party switch: would the rule change if political identities were reversed?
- Strongest opposition: does the conclusion survive the strongest credible objection?
- Falsifiability: which new evidence or changed assumption would alter the empirical conclusion? Identify moral commitments separately.
- Evidence symmetry: scrutinize competing claims consistently without giving unequal evidence equal weight.
Demand a stronger justification as coercion, cost or irreversible harm increases. Apply that burden to inaction too. Where plausible assumptions change the preferred choice, disclose the sensitivity. The Green Book’s appraisal methods provide useful techniques for comparing options and testing uncertainty; they do not determine COS’s values.
Adapt the method to the decision
Candidates and officeholders: examine authority, documented conduct, competence, integrity and governing constraints. Distinguish promises from achievements, allegations from findings, and an individual’s contribution from collective results. An endorsement is a broader judgment than agreement on selected votes.
Ballot measures: begin with the operative text, financing, enforcement and consequences of passage or failure. Test campaign claims against the mechanism and realistic alternatives.
Judges: examine legal competence, reasoning, ethics, procedural fairness, independence and respect for the office’s limits. Disagreement with a ruling’s policy effect does not establish judicial misconduct.
Campaign finance: report funding and material interests accurately. Separate contributions received by campaigns from outside spending, employee donations from organizational contributions, and documented relationships from inferred influence. Money can warrant scrutiny without proving that a particular payment caused a particular action. Funding does not automatically add or subtract voting-alignment points.
Evaluate historical decisions without hiding hindsight
Label whether an assessment asks what was reasonably supported at the time or what later evidence reveals.
For decision-time analysis, identify an evidence cutoff at the decision and use information reasonably available then. Later retrieval of an authentic contemporary record is acceptable; later discoveries and outcome studies belong in a separate retrospective review. If the historical evidence cannot be reconstructed adequately, say so.
Applying COS today to earlier decisions is a retrospective application of the stated methodology. It is not a claim that the official used COS or knew every relevant source. Later evidence can reveal whether expectations were justified without silently changing the historical evidence set.
Use numerical summaries within their limits
A numerical summary is permissible when readers can inspect its selection rules, inputs, formula and exclusions. It must describe a defined quantity rather than imply an objective ranking of human or political worth.
COS Voting Alignment measures the share of assessed votes cast that agree with COS’s published positions. It does not measure policy impact, honesty, independence, competence or agreement with constituents. A claim about constituent preferences requires separate, representative evidence. A score of 100% means complete alignment on the included votes, not a perfect public servant.
Keep policy importance, evidence confidence, participation and funding separate. Equal vote weights measure frequency of alignment, not equal social consequences. Readers must be able to inspect the judgments behind the arithmetic. The pilot rules below do not authorize general politician rankings.
Keep the record open to correction
Every assessment needs its decision, sources, material effects, counterargument, uncertainty, conclusion, decisive tradeoff and version. Preserve source dates and exact locations, consistent with our Citation & Attribution Policy.
Record author, reviewer, review status and material conflicts. AI may assist with discovery, extraction and checking; an accountable human editor approves public assessments. Do not describe self-review or agreement between AI outputs as independent human validation.
Readers may challenge a fact, interpretation, omission or rule. Evaluate the evidence and publish consequential corrections with the reason, date and affected conclusions. Distinguish factual corrections, later developments and methodology changes. Preserve earlier versions and identify any recalculated scores. Neither criticism nor political inconvenience is a reason to conceal a correction.
Appendix A: Congressional Voting Score Protocol v0.1
Status: experimental pilot. No validated public rating system or completed pilot is claimed. COS v1.1 governs; this protocol specifies its voting application.
Scope and historical window
Begin with Senate records for the 117th, 118th and 119th Congresses: January 3, 2021, through each release’s stated cutoff. Include former as well as serving members, using their actual service dates. Congress boundaries follow the Senate’s session record; do not truncate a Congress at December 31 if it continued into January.
The House is a later extension requiring its own procedural checks. Report each Congress separately. Do not publish a combined six-year headline score or cross-chamber ranking under v0.1.
Eligible decisions and selection
Build a complete inventory of recorded final-passage votes on bills and joint resolutions, including recorded concurrence in final legislative text and adoption of conference reports, whether successful or unsuccessful. Exclude nominations, treaties, impeachment, veto overrides, amendments considered separately, budget resolutions, cloture and other procedural motions from v0.1 scoring. Preserve their records and exclusion reasons for later protocols.
This narrow scope omits important exercises of power, including measures blocked before final passage. It cannot represent the full legislative record. Voice votes and unanimous-consent actions do not establish an individual Yes/No position; do not infer one. See the official guide to congressional votes.
Use one scoring unit per measure per chamber per Congress: the latest eligible roll call as of the release cutoff. Group separately numbered measures together only when they present the same operative policy choice without material differences; record the text comparison and use the latest eligible roll call in that group. Shared subjects alone do not justify grouping. Publish superseded roll calls and grouping decisions. Later replacement of a scoring unit requires a new release and explanation.
Assign one primary policy area to each unit before viewing member scores, using a published category list. Record secondary areas without counting the vote again. Do not select cases because they help or hurt a particular member. For the pilot, freeze a list of 30 units, ten per Congress, selected to test different policy areas and difficult tradeoffs. Publish the selection rationale and inventory; this is a purposive test set, not a representative sample. Any extra procedural test cases remain outside the calculation.
Assessments and evidence
Identify the chamber, Congress, session, roll-call number, date, official question, applicable text and decision group. Verify member identifiers and service dates; preserve original vote categories and normalize Yea/Aye and Nay/No consistently.
Complete the assessment form below using evidence available at the vote. Record Yes and No consequences separately. Assign favor Yes, favor No, or unresolved, plus confidence and an explanation. Score only approved Yes/No assessments with at least moderate confidence in their decisive factual premises. A numerical confidence weight is not used. Unresolved tradeoffs or premises that cannot support a direction remain unscored.
One approved assessment applies to every eligible member on that roll call. Drafts, disputed assessments awaiting resolution and missing records do not enter published calculations.
Calculation and required display
For each member and Congress, define:
- A: aligned Yes/No votes cast on included, approved scoring units.
- N: opposed Yes/No votes cast on those units.
- M: official Present or Not Voting records on those units while the member was serving and entitled to vote.
- U: missing or unverified member records on those units while eligible.
Voting alignment = 100 × A ÷ (A + N).
Assessed-vote participation = 100 × (A + N) ÷ (A + N + M).
If a denominator is zero, show not available, never zero percent. If U is greater than zero, withhold the affected member’s percentage summaries until reconciled. Service before swearing-in or after departure creates no eligible opportunity. Absence does not count as opposition; retain the actual official category and any verified context.
Show whole percentages and underlying counts. For example, 18 aligned votes and 6 opposed votes produce 75% alignment. If there were also 3 absences, assessed-vote participation is 24/27, or 89%. This is a hypothetical arithmetic example, not a politician’s result.
Every displayed pilot result must include the experimental label, Congress, included dates, A/N/M counts, participation, policy-area distribution, and a link to all included assessments. Separately show the eligible inventory size, how many units were assessed, how many were scored, and how many remain unresolved or pending. Review coverage and member participation are different denominators.
V0.1 sets no validated minimum vote count. Pilot results may appear as clearly labeled worked examples; profile headline scores and leaderboards remain withheld until a later protocol publishes tested minimum coverage, participation and comparability rules. A selective set of votes must not be presented as comprehensive performance.
Finance displayed alongside voting
Display separate two-year reporting windows: 2019–2020, 2021–2022, 2023–2024 and 2025–2026 through the latest verified filing coverage. The first captures the 2020 election period; it is not part of the legislative scoring window. For senators, these windows are not necessarily complete campaign histories for their six-year terms.
Use verified candidate-to-committee mappings and identify which committees are included. Preserve FEC category definitions, reporting dates, amendments, refunds and transfer treatment. Do not double count transfers between included committees, label all other-committee receipts as corporate PAC money, or treat unitemized receipts as a verified donor count. Missing data are not zero. Outside spending is a separate display. Finance does not enter the alignment formula.
Pilot review and release controls
Before any production score release, two human reviewers independently assess at least ten pilot units spanning the three Congresses and difficult cases. Publish agreement, disagreements, their resolution and resulting rule changes; agreement establishes consistency, not truth. Recheck changed rules against all affected pilot assessments.
Test every member-to-vote mapping in the pilot, verify calculations independently, and document how results change when each unit or primary policy area is removed. Report the resulting ranges as sensitivity checks, not statistical confidence intervals. Examine exclusions and absences for systematic distortions. Do not claim population validity from a purposive sample.
Publish a pilot report and a revised protocol setting release thresholds before enabling profile scores. Save each release’s source snapshot, cutoff, assessment versions, selected units and calculation version. Separate the date records were fetched from the date through which editorial assessment is complete. Data corrections trigger logged recalculation; new judgments require review before inclusion. Retain earlier releases and make failures or stale coverage visible.
Appendix B: Reusable COS assessment form
Copy this form for each decision. Use “not applicable” with a reason rather than leaving material questions unanswered. The metadata is part of the assessment record; private contact details are not.
Identity and scope
- Assessment ID / revision / status: Draft, reviewed, approved or superseded.
- COS version / applicable protocol version:
- Author / reviewer / review dates / material conflicts:
- Decision / institution / jurisdiction / powers:
- Decision date / evidence cutoff / research access date:
- Assessment type: Decision-time or subsequent outcome review.
- For votes: Congress, chamber, session, roll call, measure, text version, official URLs, decision group and primary policy area.
- Selection: Eligible, excluded or superseded; rationale, related roll calls and grouping evidence.
Choice and evidence
- Actual alternatives: What changes under each option? For votes, explain Yes and No separately.
- Baseline and mechanism: What would otherwise happen, and how would the action change it?
- Sources: For each source, record ID, title, URL, issuing body, date, exact passage/table, supported claim, independence and limitations. For historical decisions, explain availability at the time.
- Claim status: Identify facts, findings, attributed/disputed claims, inferences, unknowns and value judgments.
Ten-lens review
| Lens | Material effect or reason not applicable | Evidence and confidence |
|---|---|---|
| Human welfare | ||
| Distribution and inequality | ||
| Civil liberties and equal treatment | ||
| Economic and material effects | ||
| Fiscal reality and opportunity cost | ||
| Institutional integrity and democratic accountability | ||
| Evidence of effectiveness | ||
| Implementation and administrative capacity | ||
| Unintended consequences and behavioral response | ||
| Reversibility, resilience and future lock-in |
For each material effect, include direction, magnitude or its limits, affected groups, distribution, time horizon and the source. Cross-reference overlapping effects instead of counting them twice.
Judgment and checks
- Strongest case for / strongest case against:
- Rights or legal constraints / decisive tradeoff / value priority:
- Party-switch and evidence-symmetry checks:
- What changes the conclusion / sensitivity to plausible assumptions:
- Conclusion: Favor proposal/Yes, oppose proposal/No, or unresolved.
- Confidence and reason / unresolved-evidence or values issue:
- Scoring disposition: Include, exclude or hold; explain under the protocol. No fractional credit for unresolved cases.
- Public explanation: One short paragraph connecting the decisive evidence to the conclusion and its limits.
- Review and revision: Reviewer decision, disagreements, correction type, affected releases and superseded assessment link.
What changed in version 1.1
Version 1.1 preserves COS’s public-interest baseline, ten lenses, counterfactual reasoning and nonpartisan evidence standards. It clarifies source selection, tradeoff decisions, historical evidence, numerical-summary limits and correction records. It adds an experimental congressional protocol and a reusable form. Earlier assessments retain their original methodology version unless explicitly reassessed; this revision does not silently recalculate past conclusions.
References and Further Reading
Evidence and public-policy foundations
- GAO — Evidence-Based Policymaking: Practices to Help Manage and Assess the Results of Federal Efforts. Guidance on developing and using evidence to assess public efforts.
- GAO — Federal Programs: Assessing and Improving Effectiveness. Distinguishes process, outcome and impact evaluation.
- OECD — Measuring Well-being and Progress. A multidimensional approach to living conditions, distribution and future well-being.
- OMB — Circular A-94, November 9, 2023 revision. The edition referenced by COS v1; retained for analytical background, not as a claim about currently operative federal requirements or discount rates.
- HM Treasury — The Green Book 2026. Methods for appraising alternatives, costs, risks and uncertainty.
- Stanford Encyclopedia of Philosophy — Public Justification. Philosophical background on reasons for exercising political power.
- Stanford Encyclopedia of Philosophy — The Common Good. Background on shared civic interests and institutions.
Congressional and financial records
- U.S. Senate — Dates of Sessions of the Congress. Official session chronology for assigning records to Congresses.
- U.S. Senate — How to Find Congressional Votes. Official vote sources and limits on individual attribution.
- FEC — Committee Summary File Description. Reporting periods and financial categories for contextual funding displays.
These sources inform the method; they do not endorse COS or validate its scoring protocol. Each application must identify its methodology version, evidence cutoff and any relevant changes in law, records or research.
Methodology version: 1.0 — October 2026
Applies to: candidates, ballot measures, judges, elected offices, public-policy questions, and sherafy.com election guides
Most political guides begin too late.
They begin with candidates.
Or parties.
Or endorsements.
Or a proposition number followed by a list of organizations telling you how to vote.
The Sherafy Civic Outcomes Standard begins somewhere else:
What are we actually trying to accomplish for the people who have to live under the result?
That sounds almost embarrassingly obvious. It is not how a great deal of political analysis works.
In ordinary political life, people often begin with identity and reason backward. Democrats know the Democratic position. Republicans know the Republican position. Advocacy organizations know the outcome they were created to advocate for. Campaigns know what their candidate needs voters to believe.
Then everyone goes looking for facts.
The Civic Outcomes Standard reverses the order.
We declare the values first. We establish what the government action or elected office can actually do. We identify the relevant evidence. We separate established facts from estimates, claims, inference, and unknowns. We compare the choice with realistic alternatives. We examine who receives the benefits, who bears the costs, how strong the causal evidence is, whether the institution can actually implement the proposal, and what can go wrong.
Only then do we describe what the evidence implies.
The framework is not politically neutral in the sense of having no values. No serious political analysis is value-free.
It is neutral in a more useful sense:
No party, ideology, candidate, institution, corporation, union, activist group, or government agency receives a presumption that its preferred mechanism is correct.
A policy associated with the left does not become effective because its purpose is compassionate.
A policy associated with the right does not become efficient because its purpose is fiscal restraint.
A government program is not successful because it spends money on an important problem.
A tax cut is not successful merely because taxpayers keep more money.
A regulation is not successful because the harm it targets is real.
Deregulation is not successful because compliance costs are real.
Intent is evidence about intent.
It is not evidence about outcomes.
That distinction is the foundation of the Civic Outcomes Standard.
The name: Civic Outcomes Standard
We chose Civic Outcomes Standard, or COS, because the name states what the method is supposed to do without requiring a voter to adopt a partisan identity.
Civic means the analysis concerns the shared institutions, rules, resources, rights, responsibilities, and decisions through which people live together.
Outcomes means we care about what happens in the real world, not merely what a proposal is called, which coalition supports it, or what its authors intended.
Standard means the same evidentiary rules apply when the result helps our prior beliefs and when it hurts them.
This methodology is adapted from the Objectivity AI framework developed by Fabled Sky Research and extended by sherafy.com specifically for elections, public policy, and civic decision-making.
Objectivity AI contributes the core evidentiary discipline: distinguish verified facts, disputed claims, reasonable inference, and unknowns; follow evidence backward toward primary sources; do not manufacture symmetry where the evidence is unequal; and do not manufacture certainty where the evidence is incomplete.
The Civic Outcomes Standard adds a public-policy layer:
Even when the facts are known, how do we evaluate what those facts mean for the people who must live with the decision?
That requires an explicit theory of what counts as a desirable civic outcome.
The framework’s public-interest baseline
The Civic Outcomes Standard begins from a deliberately broad proposition:
A healthy political system should improve people’s ability to live safe, healthy, free, materially secure, meaningful lives while preserving the institutions and resources that allow future generations to do the same.
That baseline is intentionally wider than GDP.
Economic growth matters enormously. So do wages, housing, employment, productivity, prices, investment, and entrepreneurship.
But a society can become richer on paper while housing becomes inaccessible, preventable illness rises, public institutions deteriorate, political power becomes increasingly concentrated, or the gains from growth accrue almost entirely to a narrow segment of the population.
Likewise, a society can pursue equality so mechanically that it damages productive capacity, freedom, innovation, or the very tax base required to sustain public goods.
The objective is not equality at any cost.
It is not growth at any cost.
It is broadly shared human flourishing within a society capable of sustaining it.
The OECD’s Well-being Framework reflects a similar analytical insight: economic output alone does not tell us whether life is improving, for whom it is improving, or whether those gains can be sustained. The OECD therefore examines material conditions, quality of life, inequalities in outcomes, and the resources needed for future well-being.
Our civic baseline therefore includes several commitments:
- People should have a meaningful opportunity to build stable and productive lives.
- Preventable suffering is a legitimate concern of public policy.
- Civil liberties and equal treatment matter independently of aggregate economic benefit.
- Public safety matters, including protection from both private violence and unjustified state coercion.
- Prosperity should be broadly accessible rather than structurally reserved for people who already possess wealth or political power.
- Extreme inequality matters when it damages health, mobility, opportunity, social stability, market competition, or political equality.
- Government should be capable of delivering the things it promises.
- Public resources should be treated as scarce even when the purpose of spending is admirable.
- Markets should be evaluated by how well they produce useful competition and human welfare, not treated as inherently virtuous or inherently exploitative.
- Environmental costs, infrastructure depletion, debt, institutional damage, and other burdens cannot simply be shifted to people who are too young to vote today.
- Political power should remain accountable, contestable, transparent, and constrained by law.
A reader does not have to weight every one of those goals exactly as sherafy.com does.
That is important.
The point of an explicit baseline is not to smuggle ideology into analysis. It is to prevent ideology from being smuggled in invisibly.
If a reader cares more about one value and less about another, the analysis should still be useful because the underlying effects remain visible.
The reader can change the weights.
They should not have to reconstruct the facts.
Objectivity is not the absence of values
This deserves its own section because political discussions routinely confuse objectivity with moral emptiness.
Evidence cannot tell us what ultimate goals we must care about.
A spreadsheet cannot prove that liberty matters.
A randomized trial cannot establish that human dignity has moral worth.
An econometric model cannot determine the proper balance between privacy and security without some prior judgment about the importance of each.
Those are normative questions.
But once a society decides that it cares about an outcome, evidence becomes indispensable.
If we care about reducing homelessness, evidence can help determine which interventions actually reduce homelessness.
If we care about reducing crime, evidence can help distinguish policies that reduce victimization from policies that merely increase punishment.
If we care about educational opportunity, evidence can help determine whether a program improves student outcomes and for whom.
If we care about housing affordability, evidence can help determine whether a subsidy increases effective purchasing power, whether its benefits capitalize into prices, whether new construction increases supply, and which households ultimately gain.
If we care about environmental protection, evidence can help measure the damage, compare interventions, identify externalities, and estimate what a regulation costs relative to the harm it prevents.
The philosophical tradition of public justification asks, in different forms, how the exercise of political power can be justified to the people subject to it. The related idea of the common good asks citizens to reason from interests and institutions they share despite substantial disagreement over politics, religion, and the good life.
The Civic Outcomes Standard is not a new theory of political philosophy.
It is a practical editorial method built for the place where those problems become concrete:
There is an election. There are actual choices. What do we know about what each one is likely to do?
The standard in one sentence
The full methodology can be summarized this way:
Define the real choice, state the public goals openly, establish the baseline, follow the strongest available evidence, compare realistic alternatives, examine the effects across ten civic lenses, identify uncertainty and tradeoffs, and never substitute partisan identity for causal analysis.
Everything else follows from that.
The evidence protocol
Before sherafy.com evaluates consequences, we first determine what we actually know.
Every consequential factual claim should fall into one of the following categories.
Verified fact
Directly supported by reliable evidence such as statutory text, official records, audited financial information, high-quality datasets, court records, original research, documented votes, or other strong primary evidence.
Well-supported finding
Supported by a substantial body of credible evidence but not literally provable in the way a date, vote, expenditure, or statutory provision may be.
This category often applies to scientific, economic, or social-science conclusions.
Disputed claim
A material claim for which credible disagreement exists or for which an interested party’s assertion has not been independently established.
The fact that a campaign says something is a verified fact about what the campaign says.
It does not make the underlying allegation verified.
Reasonable inference
A conclusion not directly documented but supported by evidence such as incentives, chronology, behavior, comparable outcomes, institutional structure, financial relationships, or multiple corroborating facts.
Inference must be labeled as inference.
Unknown
The available evidence is insufficient to support a responsible conclusion.
“Unknown” is an answer.
It is often the most honest one.
Value judgment
A conclusion that depends substantially on moral priorities or on how competing legitimate values are weighted.
Value judgments should not be disguised as empirical facts.
Evidence quality and confidence are separate from political desirability
For material findings, sherafy.com should identify confidence as high, moderate, low, or indeterminate when doing so meaningfully helps the reader.
That confidence describes the evidence.
It does not describe whether we like the result.
A policy can have:
- strong evidence of a small beneficial effect;
- weak evidence of a potentially large beneficial effect;
- strong evidence of a serious tradeoff;
- substantial uncertainty in both directions.
Those are very different propositions.
We do not collapse them into one number.
Evidence hierarchy
The source closest to the fact generally deserves the most weight, although no source is automatically infallible.
The preferred order is:
- Legal text and primary public records
- Original datasets, audits, financial records, and administrative data
- Original scientific and social-science research
- Systematic reviews and high-quality evidence syntheses
- Court opinions and filings for legal questions
- Independent institutional analysis
- High-quality original journalism
- Specialist reporting and credible expert analysis
- Campaigns, advocacy groups, unions, trade groups, corporations, political parties, and government press releases as evidence of what those actors claim
- Unverified social-media assertions and anonymous claims as leads, not conclusions
A source’s institutional prestige is not a substitute for evaluating its method.
Likewise, a source’s political affiliation does not automatically invalidate accurate evidence.
We ask what the source actually establishes.
The counterfactual rule: compared with what?
One of the most common failures in policy argument is comparing an imperfect reality with an imaginary perfect alternative.
A housing program costs money.
Compared with what?
A criminal-justice reform creates risk.
Compared with what?
A regulation imposes compliance costs.
Compared with what?
Doing nothing is not necessarily free.
Maintaining the status quo is itself a policy choice with costs, benefits, risks, and distributional effects.
The correct comparison is therefore not:
Is this proposal good?
It is:
What is likely to happen under this proposal compared with the most realistic alternative?
The U.S. Government Accountability Office distinguishes among process, outcome, and impact evaluation. Impact evaluation asks the crucial causal question: what effect did a program have compared with what would have happened in its absence?
That counterfactual discipline is mandatory under the Civic Outcomes Standard.
The ten substantive lenses
Every significant election choice should be examined through the same ten lenses.
Not every lens will matter equally in every race.
A county assessor and a governor do not exercise the same powers. A judicial retention election is not a bond measure. A school-board race should not be forced through questions relevant only to foreign policy.
But the analyst must consider every lens and explain why a lens is material or immaterial rather than silently ignoring inconvenient effects.
1. Human welfare
What happens to people’s actual lives?
This is the broadest lens and the reason the others exist.
Relevant indicators may include:
- mortality and life expectancy;
- physical and mental health;
- crime and victimization;
- housing stability;
- food and energy security;
- educational attainment;
- environmental exposure;
- transportation access;
- family stability;
- disability access;
- time burdens;
- economic insecurity;
- subjective well-being when reliable measures exist.
The framework asks both how many people are affected and how severely.
Saving one person from catastrophic harm can matter more than providing a trivial convenience to thousands.
Likewise, a program serving a sympathetic population is not assumed effective merely because its intended beneficiaries have legitimate needs.
Need establishes the importance of the problem. It does not establish the effectiveness of the solution.
2. Distribution and inequality
Who gains, who pays, and where do the effects land?
A policy that increases total economic output can still impose severe losses on people least capable of absorbing them.
A policy described as redistributive can also produce benefits that are captured by landlords, incumbent firms, asset owners, contractors, or higher-income households rather than the people named in the legislation.
We therefore examine:
- income groups;
- wealth groups;
- renters and homeowners;
- workers and employers;
- consumers and producers;
- geographic communities;
- racial or demographic disparities when materially relevant;
- age and generation;
- people with disabilities;
- small firms versus incumbent firms;
- concentrated benefits versus diffuse costs;
- concentrated costs versus diffuse benefits.
The OECD’s well-being work explicitly examines both aggregate outcomes and how those outcomes are distributed across populations.
The Civic Outcomes Standard adopts the same principle:
An average can improve while millions of people become worse off.
Distribution is not an optional footnote.
3. Civil liberties and equal treatment
What power is being exercised over people, and what protects them from abuse?
This lens examines:
- due process;
- equal protection;
- privacy;
- speech;
- association;
- religious liberty;
- bodily autonomy;
- property rights;
- search and surveillance;
- detention and punishment;
- voting and political participation;
- discrimination;
- access to courts;
- procedural fairness.
Not every right is absolute, and rights can conflict.
Public safety can justify some coercion.
Tax systems necessarily compel payment.
Regulations constrain conduct.
Courts restrict liberty after lawful convictions.
The question is not whether government power exists.
It is whether the power is lawful, necessary, proportionate, reviewable, and protected against predictable abuse.
As coercion becomes more severe or difficult to reverse, the evidentiary and procedural burden should rise.
4. Economic and material effects
What happens to production, prices, jobs, wages, housing, competition, and household finances?
Economic analysis is not synonymous with asking whether businesses like a policy.
Nor is economics synonymous with maximizing GDP.
We examine:
- employment;
- wages and compensation;
- consumer prices;
- housing costs and supply;
- productivity;
- investment;
- innovation;
- market entry;
- competition;
- monopoly or monopsony power;
- entrepreneurship;
- labor mobility;
- household disposable income;
- insurance and risk;
- externalities;
- long-run productive capacity.
A worker protection can create compliance costs and still produce net social value.
A tax reduction can increase investment and still worsen a fiscal problem.
A regulation can correct a market failure or entrench incumbents.
A subsidy can solve an access problem or simply bid up the price of a constrained asset.
There is no ideological shortcut.
We trace the mechanism.
5. Fiscal reality and opportunity cost
What does this cost, who ultimately finances it, and what else could the same resources do?
Public money is finite.
That remains true when the proposed expenditure is morally compelling.
Fiscal analysis includes:
- direct public cost;
- administrative cost;
- debt service;
- revenue source;
- tax incidence;
- fees and indirect charges;
- long-term liabilities;
- pension or benefit commitments;
- maintenance obligations;
- displacement of other spending;
- cost per achieved outcome;
- existing spending on the same problem;
- realistic opportunities to improve efficiency.
This lens also rejects a common rhetorical trick from both sides.
Government spending is not automatically a benefit merely because money was spent.
Government spending is also not automatically waste merely because the number is large.
The question is:
What public value is produced per unit of scarce resources, compared with available alternatives?
OMB’s benefit-cost guidance requires analysts to identify benefits and costs, account for uncertainty, and examine distributional effects. Even when not every consequence can responsibly be monetized, making costs, benefits, assumptions, and uncertainty explicit improves analysis.
6. Institutional integrity and democratic accountability
What does the choice do to the institutions through which society governs itself?
A policy can produce an attractive short-term outcome while creating a dangerous institutional precedent.
This lens examines:
- corruption risk;
- conflicts of interest;
- transparency;
- auditability;
- campaign-finance relationships;
- procurement integrity;
- oversight;
- checks and balances;
- separation of powers;
- rule of law;
- political capture;
- independent review;
- public-record access;
- enforcement discretion;
- accountability for failure;
- democratic responsiveness.
This does not mean every procedural violation outweighs every substantive outcome.
It means institutional damage has a real cost.
A society repeatedly willing to break its own safeguards for causes it considers urgent will eventually create powers that can be used by people it does not trust.
7. Evidence of effectiveness
Is there a credible reason to believe the proposed mechanism will produce its claimed outcome?
This is where good intentions most often meet reality.
We ask:
- What is the theory of change?
- What causal link connects the action to the promised result?
- Has something similar been tried?
- What happened?
- Are the comparison cases genuinely comparable?
- Is the evidence experimental, quasi-experimental, observational, modeled, anecdotal, or purely theoretical?
- Has the result replicated?
- Are effect sizes practically meaningful?
- Does the evidence apply at the proposed scale?
- What does the strongest contrary evidence show?
A policy does not receive credit because its goal is important.
The more important the goal, the more important it is to determine whether the mechanism works.
8. Implementation and administrative capacity
Can the responsible institution actually do this?
A law on paper and a functioning program are different things.
Implementation analysis asks:
- Which agency or official is responsible?
- Does that institution have legal authority?
- Does it have the staffing and expertise?
- What technology or infrastructure is required?
- What procurement is necessary?
- How long will implementation realistically take?
- What compliance burden falls on citizens or businesses?
- How will eligibility be determined?
- How will fraud and error be controlled?
- Can the program reach the people it is designed to serve?
- Is enforcement realistic?
- What happens when the system encounters edge cases?
GAO’s evidence-based policymaking work emphasizes not only whether programs achieve results but whether agencies build evidence, monitor performance, and use what they learn to improve implementation.
The Civic Outcomes Standard treats administrative competence as a policy variable, not an afterthought.
9. Unintended consequences and behavioral response
What happens after people adapt to the rule?
People respond to incentives.
Businesses change prices and practices.
Agencies learn how to protect their budgets.
Criminal networks route around enforcement.
Consumers substitute.
Developers redesign projects.
Interest groups discover loopholes.
Recipients alter behavior.
Politicians exploit ambiguities.
The question is not merely what a law commands.
It is what the system is likely to do once the law exists.
We look for:
- price pass-through;
- avoidance and evasion;
- black markets;
- regulatory arbitrage;
- displacement effects;
- rebound effects;
- chilling effects;
- moral hazard;
- rent-seeking;
- administrative gaming;
- capture;
- perverse incentives;
- effects on supply;
- effects on people just outside an eligibility threshold.
The United Kingdom’s Green Book appraisal framework explicitly requires serious treatment of risk, uncertainty, distributional effects, sensitivity to assumptions, and the well-documented tendency for project appraisals to be overly optimistic.
That is a useful discipline regardless of jurisdiction.
The promised case is not the only case we analyze.
10. Reversibility, resilience, and future lock-in
If we are wrong, how hard is this to fix?
This final lens changes the burden of proof.
Some decisions are easily revised.
Others create:
- constitutional lock-in;
- irreversible environmental effects;
- large sunk costs;
- long-lived debt;
- entrenched bureaucracies;
- legal precedents;
- permanent surveillance infrastructure;
- dependency on a funding stream;
- land-use patterns that persist for decades;
- technology systems that are expensive to replace.
Under uncertainty, flexibility has value.
That does not mean temporary policy is always better.
Stable rules can encourage investment, protect rights, and prevent constant political interference.
But when evidence is weak and the cost of reversal is enormous, the standard should demand a stronger case than it would for a limited pilot, sunset provision, or easily amended statute.
The 2026 Green Book specifically discusses sequential decision-making, irreversible or costly-to-reverse choices, and the value of preserving flexibility as new information arrives.
Our rule is straightforward:
The more irreversible the decision, the less comfortable we should be pretending uncertainty does not exist.
Four cross-cutting tests
The ten lenses tell us what to examine.
Four additional tests tell us how to keep the analysis honest.
The party-switch test
Imagine the same policy, evidence, executive power, court theory, or fiscal maneuver proposed by the opposing party.
Would we apply the same standard?
If not, why not?
Sometimes context genuinely changes the analysis.
But if the only changing variable is the partisan label, we have found a bias problem.
The strongest-opposition test
We do not rebut the stupidest argument against our preferred position.
We find the strongest one.
If serious critics raise an empirical concern, we investigate it.
If the concern survives scrutiny, we say so.
If it does not, we explain why.
A guide that can defeat only social-media caricatures is not rigorous.
The falsifiability test
What evidence would materially change our assessment?
If there is no possible answer, then the conclusion is not functioning as an empirical judgment.
It may be a moral commitment, which should be labeled honestly.
The symmetry test
Objectivity does not require equal treatment of unequal evidence.
If one claim is supported by extensive high-quality evidence and another by speculation, presenting them as equivalent is not neutrality.
It is distortion.
The Civic Outcomes Standard therefore rejects both partisan favoritism and false balance.
Candidate analysis requires a different tool than policy analysis
Candidates are not propositions.
A candidate brings an uncertain bundle of future decisions, appointments, management choices, negotiations, emergencies, and political incentives.
For candidate races, the Civic Outcomes Standard therefore adds several specific questions.
What can the office actually do?
Campaigns routinely promise actions outside the legal or practical authority of the office being sought.
We begin with jurisdiction.
What has the candidate actually done?
When a meaningful record exists, behavior receives more weight than rhetoric.
We examine votes, executive decisions, budgets, litigation, management history, audits, official findings, public records, and measurable outcomes in context.
Does the candidate demonstrate administrative capacity?
Ideological alignment cannot compensate for an inability to run the institution.
Competence is not politically glamorous.
It is enormously important.
What powers would this person control?
Appointments, prosecutorial discretion, emergency authority, contracting, enforcement, vetoes, rulemaking, agenda setting, and control of information can matter more than campaign slogans.
Are there material integrity concerns?
Conflicts of interest, corruption findings, deception, misuse of authority, ethics violations, contempt for lawful oversight, or documented patterns of misconduct are relevant because officeholders exercise power over people who did not vote for them.
Allegations are not treated as findings.
Findings are not softened into allegations.
What does the candidate’s coalition tell us?
Endorsements and donors are contextual evidence, not automatic proof of what a candidate will do.
But a candidate’s advisers, funders, political allies, institutional backers, and likely appointees can reveal incentives and governing constraints that campaign websites omit.
Ballot measures require legal analysis before political analysis
Ballot measures are especially vulnerable to slogan-based reasoning because voters often encounter only a title, a television advertisement, and a paragraph in a voter guide.
For every significant measure, sherafy.com should establish:
- What does the legal text actually change?
- What happens if it passes?
- What happens if it fails?
- What is the realistic status-quo baseline?
- Who pays directly and indirectly?
- Who is expected to benefit?
- What does the nonpartisan fiscal analysis say?
- What evidence supports the policy mechanism?
- Who funded the campaign, and what material interests do they have?
- What are the strongest arguments on both sides?
- What is uncertain?
- Can ordinary legislation repair mistakes, or does the measure create constitutional or financial lock-in?
A ballot measure should never receive a favorable presumption because its title contains words like schools, safety, rights, taxpayer protection, affordability, accountability, victims, or environment.
Those words identify values.
The legal text determines the mechanism.
Judges require a judicial standard, not a partisan standard
Judicial elections and retention votes create a special problem.
A judge’s role is not to produce whichever policy outcome a voter prefers.
For judges, the Civic Outcomes Standard emphasizes:
- ethical record;
- conflicts of interest;
- disciplinary findings;
- legal competence;
- fidelity to governing law;
- procedural fairness;
- due process;
- quality and clarity of legal reasoning;
- appropriate treatment of precedent;
- independence from improper political pressure;
- respect for the limits of judicial authority;
- professional evaluations where methodology is credible.
A ruling that produces an outcome we dislike is not, by itself, evidence of judicial misconduct or incompetence.
Likewise, “following the law” cannot be used as a magic phrase to immunize demonstrably poor reasoning or ethical problems.
The evidence still has to be examined.
Campaign money is context, not a verdict
Money matters in politics.
It can purchase access, advertising, organizational capacity, legal talent, polling, consultants, and the ability to repeat a message until it becomes familiar.
Financial interests can also reveal who expects to gain or lose from a policy.
For those reasons, campaign funding belongs in serious election analysis.
But the Civic Outcomes Standard rejects a lazy inference:
Bad person funded this, therefore the policy is bad.
A pharmaceutical company can support a policy that is medically sound.
A labor union can support a policy that genuinely improves working conditions.
A billionaire can oppose a badly designed tax.
A police association can make a valid public-safety argument.
An environmental group can be wrong about an environmental mechanism.
Interested parties do not lose the ability to be factually correct.
Their interests tell us to scrutinize the claim.
They do not tell us what the answer must be.
We do not convert politics into a fake precision score
A great temptation in evidence-driven analysis is to make everything quantitative.
Give each candidate 82.4 points.
Give a proposition 73 out of 100.
Declare a winner.
That looks objective.
Often it is merely subjectivity with decimals.
A numerical score requires weights.
How much is privacy worth relative to a projected 0.3 percent reduction in some cost?
How many points should a candidate lose for an ethics concern compared with administrative competence?
How should a modest aggregate economic gain be weighed against a concentrated loss imposed on a vulnerable population?
Those are partly value judgments.
We will not hide them inside arithmetic.
Instead, sherafy.com election analysis should report, when the evidence permits:
- direction of effect;
- estimated magnitude;
- population affected;
- distribution of benefits and costs;
- time horizon;
- evidence quality;
- confidence;
- material uncertainty;
- important value conflicts;
- reversibility.
Readers should be able to see the machinery.
The burden of proof rises with power, cost, and irreversibility
Not every political decision requires the same amount of evidence.
A one-year pilot program with a small budget is different from a permanent constitutional amendment.
A minor licensing adjustment is different from imprisonment.
A reversible administrative rule is different from issuing billions of dollars in long-term debt.
The Civic Outcomes Standard therefore uses a graduated burden of proof.
The case should become stronger as a proposal becomes:
- more coercive;
- more expensive;
- more difficult to reverse;
- more constitutionally significant;
- more dependent on uncertain forecasts;
- more likely to affect fundamental rights;
- more likely to create large irreversible harms if its assumptions are wrong.
This is not an argument for paralysis.
Doing nothing can itself produce large and irreversible harm.
The burden applies to the choice among alternatives, not only to action.
Status quo is not the neutral option
This rule matters enormously.
People often demand exhaustive proof before changing an existing system while requiring almost no evidence that the existing system deserves to continue.
That is status-quo bias masquerading as caution.
If the current system is producing preventable deaths, unaffordable housing, failing infrastructure, excessive incarceration, pollution, fiscal instability, monopoly pricing, educational failure, or rights violations, maintaining it has consequences too.
The correct comparison is:
change versus continuation, with both evaluated seriously.
How sherafy.com will present uncertainty
Political campaigns hate uncertainty.
Our methodology requires it.
Where evidence is strong, we should say it is strong.
Where estimates depend heavily on assumptions, show the assumptions.
Where credible studies conflict, explain why.
Where a forecast could plausibly move in several directions, give the range rather than selecting the number most favorable to a narrative.
Where an effect cannot currently be known, say so.
The UK Green Book’s use of sensitivity analysis is useful here: ask what happens when important assumptions change, and identify the point at which a conclusion would change.
For sherafy.com, the reader-facing version is simple:
What would have to be different for this conclusion to stop being true?
That question is often more informative than another paragraph of political argument.
The final analysis record
When the Civic Outcomes Standard is applied to a candidate, proposition, or major policy choice, the finished analysis should allow a reader to recover the reasoning.
At minimum, it should contain:
The actual choice
What legally or institutionally changes?
The baseline
What happens without the proposed change?
The stated goal
What problem is the candidate or measure trying to solve?
The mechanism
How is the proposal supposed to cause the desired outcome?
The evidence
What supports or undermines that causal story?
The ten-lens review
What are the material effects on welfare, distribution, rights, economics, fiscal conditions, institutions, effectiveness, implementation, unintended consequences, and reversibility?
Who benefits and who bears the cost
Not merely who is named in the campaign.
Strongest case for
The best serious argument, not campaign fluff.
Strongest case against
The best serious criticism, not partisan caricature.
Conflicts and interested parties
Including funding when material.
Unknowns
What cannot responsibly be established.
Confidence
How strongly does the available evidence support the important findings?
What would change the analysis
The assumptions or new evidence that could materially alter the assessment.
That record allows disagreement without forcing the reader to start from scratch.
Who this standard is for
This standard is not only for people who call themselves progressive.
It is not only for Democrats.
It is not a centrist project either.
Centrism can become its own shortcut if it assumes the truth must lie halfway between two political coalitions.
Sometimes one side is substantially more correct.
Sometimes neither is.
Sometimes the disagreement is mostly about values.
Sometimes people share the same goal and have radically different beliefs about causation.
The Civic Outcomes Standard is for the voter who can say:
I have values. I want society to work. I also want to know whether the thing being sold to me actually works.
That voter might be liberal.
Conservative.
Libertarian.
Social democratic.
Moderate.
Independent.
Politically homeless.
Or simply tired of discovering that every “voter guide” is an endorsement sheet wearing glasses.
People who disagree politically often still want many of the same basic things: physical safety, economic security, opportunity for their children, decent infrastructure, freedom from arbitrary power, access to competent institutions, communities that function, and some confidence that tomorrow will not be worse because today’s leaders consumed everything in reach.
They disagree over priorities.
They disagree over causes.
They disagree over what government should do.
They disagree over what markets can solve.
They disagree over risk.
Those disagreements are real.
The purpose of the framework is not to erase them.
It is to locate them accurately.
If two people agree on the evidence but weight liberty and security differently, that is a values disagreement.
If they agree on the goal but disagree about whether rent control reduces housing supply, that is an empirical disagreement.
If they agree about both but disagree about whether a city can administer a complicated program competently, that is an implementation disagreement.
Those conversations become much more productive once we stop pretending they are all the same kind of disagreement.
What this standard is not
It is not an algorithm that can replace judgment.
It is not a machine for proving that sherafy.com‘s politics are correct.
It is not “both sides” journalism.
It is not anti-government.
It is not pro-government.
It is not pro-market or anti-market.
It is not a demand that every human value be converted into dollars.
It is not a requirement that uncertainty prevent action.
And it is not a promise of perfect objectivity.
Human beings choose the questions.
Human beings choose what outcomes matter.
Human beings interpret evidence.
Human beings can make mistakes.
The purpose of a standard is not to pretend those limitations disappear.
It is to make the reasoning auditable enough that mistakes can be found.
That may be the most important feature of the entire project.
A political methodology should contain a mechanism for discovering that its authors were wrong.
Otherwise it is not an analytical framework.
It is an ideology.
Our commitment
Every sherafy.com election guide built under the Civic Outcomes Standard should be held to the same promise:
We will state our values instead of hiding them.
We will not treat party affiliation as evidence.
We will distinguish goals from mechanisms.
We will compare proposals with realistic alternatives, including the real costs of the status quo.
We will follow claims back toward the strongest available evidence.
We will distinguish verified facts, disputed claims, inference, uncertainty, and value judgments.
We will examine who actually benefits and who actually pays.
We will count fiscal costs even when we support the goal.
We will count human and institutional costs even when they do not fit neatly in a spreadsheet.
We will scrutinize implementation rather than assuming legislation becomes reality when the governor signs it or voters approve it.
We will look for second-order effects and incentives.
We will demand more evidence as coercion, cost, risk, and irreversibility increase.
We will not create false balance when the evidence is lopsided.
We will not create false certainty when it is not.
We will correct the record when better evidence changes the answer.
And above all:
No political tribe gets evidentiary credit for good intentions.
The people living under public policy deserve something better than that.
They deserve to know what the choices actually are, what the strongest evidence says, where the tradeoffs fall, what remains uncertain, and what each decision is likely to mean outside the campaign advertisement.
That is the Civic Outcomes Standard.
And that is the standard we will use.
Methodology Foundations and Further Reading
Evidence-Based Policy and Program Evaluation
U.S. Government Accountability Office — Evidence-Based Policymaking: Practices to Help Manage and Assess the Results of Federal Efforts
GAO’s framework for developing, assessing, using, and continuously improving the evidence that government agencies rely on to understand whether programs achieve intended results.U.S. Government Accountability Office — Federal Programs: Assessing and Improving Effectiveness
A useful distinction among process, outcome, and impact evaluations, including the counterfactual question of what would have happened without the program.
Social Outcomes, Distribution, and Future Well-Being
- OECD — Measuring Well-being and Progress
The OECD’s multidimensional framework for evaluating current well-being, inequalities in outcomes, and the resources necessary to sustain well-being for future generations.
Costs, Benefits, Uncertainty, and Distribution
U.S. Office of Management and Budget — Circular A-94: Guidelines and Discount Rates for Benefit-Cost Analysis of Federal Programs
Federal analytical guidance addressing benefit-cost and cost-effectiveness analysis, uncertainty, sensitivity analysis, and distributional effects.HM Treasury — The Green Book 2026
Detailed public-policy appraisal guidance addressing social value, distribution, risk, uncertainty, optimism bias, sensitivity analysis, implementation, and the value of flexibility where decisions are difficult to reverse.
Public Justification and the Common Good
Stanford Encyclopedia of Philosophy — Public Justification
Background on the idea that exercises of political power require reasons capable of public justification rather than mere factional preference.Stanford Encyclopedia of Philosophy — The Common Good
A philosophical overview of shared civic interests and the institutions through which members of a political community reason about common concerns.
Editorial note: The Civic Outcomes Standard is a living sherafy.com methodology. When the standard materially changes, future election analyses should identify the version used so readers can understand the rules under which an earlier conclusion was reached.


