title: “Content Guide: SBRS (schoolboardresearch.org)” type: content-guide site: schoolboardresearch.org updated: 2026-07-08

ESB Language, Terminology & Mechanics (binding)

This section is binding for all generated content. Where anything below conflicts with older instructions in this guide, this section wins. The full standard is esb_writing_style_guide.md in the canonical guides folder.

Mechanics. No em-dashes. Where an em-dash would go, use a spaced double-hyphen (“–”), sparingly, or split the sentence. No en-dashes (use a hyphen, including in ranges like “1-2”). No semicolons in body prose. Straight quotes only. Oxford comma. No horizontal rules.

ESB terminology, never Policy Governance terms. Goals, not Ends. Guardrails, not Executive Limitations. Governing, not Governance Process. Delegation, not Board-Management Delegation. Never describe ESB as built on or adopted from Policy Governance. It is grounded in “principles of a policy-based approach to governance.” Never write “SOFG,” “Student Outcomes Focused Governance,” “Policy Governance,” “PG,” “Lone Star Governance,” or “LSG” (exception: SOFG only when intentionally contrasting with ESB). Use “board chair,” not “board president.” Use “Practitioner,” not “coach,” for the certified role. Spell out “Any Reasonable Interpretation” (never “ARI”). Use “own conduct,” not “own-conduct.” Four board-adopted policy types only: Goals, Guardrails, Delegation, Governing (no “Policies About X” category). ESB is prioritized (1-3 policies) in Goals and Guardrails and exhaustive in Delegation and Governing. A Goal is always a student outcome.

Governance language to avoid (describe the specific behavior instead): “micromanage,” “rubberstamp,” “stay in your lane,” “good/bad/right/wrong” as judgment (even affirmatives like “the right framing”), power-over framings, “district”/”school district” (say “school system” in reader-facing content – a questioner’s own words, direct quotes, and operational or data references to a specific entity may keep it), “equity” as a bare term, “achievement gap,” politically-coded terms, and “non-essential activities” for what a board could otherwise do.

LLM Configuration

Models are configured centrally – do not hardcode model slugs in this guide. See System Administration -> Model Registry (backed by site-pipeline/config.py and the per-site overrides in pipeline_sites). The registry is the single source of truth, so the docs never drift from what the pipeline actually runs.

Roles:

Writing (drafting): the configured write model, with automatic fallback to the configured write-fallback model on error or empty output.

Everything else (research, quality checks, anti-slop scanning, council review): the configured research model.

Live web search: the configured search model.

To change any of these, edit the Model Registry (or a per-site override) – never this section.

Section 3 — Site Context

Purpose: schoolboardresearch.org (SBRS) applies the PEARLS framework to evaluate peer-reviewed research and credentialed policy reports on school board governance. The site exists to help practitioners and researchers assess the quality, applicability, and limitations of governance research.

This is an evaluation site, not a summary site. Each piece critically assesses a specific study. Generating a summary of a study’s findings, however thorough, is not SBRS content. The product is an evaluation — a structured judgment about whether the research is trustworthy, how strong its evidence is, how applicable its findings are, and what its limitations mean for practice.

The distinction test: Is this piece evaluating the quality and applicability of a specific study using the PEARLS framework? If yes, it belongs on SBRS. If the piece summarizes what research says about a topic, synthesizes multiple studies into a policy brief, or uses research to support an argument, it does not belong on SBRS.

Content Format

Output format — critical: The generated file contains ONLY the front matter block followed by the article body HTML. Never output a full standalone HTML document. Do not include <!DOCTYPE html>, <html>, <head>,

File naming format: The topic phrase in the filename should be lowercase, hyphenated, and descriptive of the study’s subject (not the title). Examples:

research/board-governance-student-achievement.html

research/superintendent-tenure-stability.html

research/lighthouse-inquiry-review.html

research/effective-governance-literature.html

Front matter example:

---

layout: research

title: “Board Governance and Student Achievement: A PEARLS Review”

study_type: “Peer-Reviewed Journal Article”

date_display: “June 26, 2026”

---

Valid study_type values:

Peer-Reviewed Journal Article

Policy Report — Major Institute

State Education Agency Study

Meta-Analysis or Systematic Review

Practitioner Research Brief

PEARLS Section Structure

Every PEARLS review follows this exact structure, in this order:

Study citation header (before body text)

Brief framing paragraph (2–3 sentences orienting the reader to the study and why it merits a PEARLS review — NOT a summary of findings)

P — Peer Review (H2)

E — Evidence Quality (H2)

A — Applicability (H2)

R — Replicability (H2)

L — Limitations (H2)

S — Significance (H2)

Practitioner Takeaways (H2, final section)

Methodology page link (inline, within the framing paragraph or the first PEARLS section: “For a full explanation of the PEARLS framework, see our methodology page.”)

Word count: 800–1,200 words (body text only, not counting citation header or front matter). No PEARLS section fewer than 80 words.

Target Audience

SBRS serves:

Governance researchers — doctoral students, university faculty, research scientists at education policy organizations

Policy analysts and wonks — staff at state education agencies, think tanks, advocacy organizations who need to evaluate research before citing it

Sophisticated practitioners — board members, superintendents, and district administrators with graduate-level education who want to evaluate research directly, not just be told what it says

Think tank analysts — researchers at RAND, Brookings, AASA research division, NSBA, McREL, and similar institutions

Who this site is NOT for: Boards or administrators wanting quick takeaways from research. Those audiences are better served by effectivegoverningboards.com. SBRS is for people who want to know whether the research itself is trustworthy and applicable — not just what it claims to find.

Compared to Sister Sites

Compared to effectivegoverningboards.com: That site synthesizes research into policy briefs and practitioner guidance. SBRS evaluates individual studies using a formal framework. The output of effectivegoverningboards.com is a synthesis; the output of SBRS is an evaluation.

Compared to schoolboardreport.com: That site references research to support analysis of governance issues. SBRS is about the research itself — not using research to support another argument, but evaluating whether the research is trustworthy.

The PEARLS structure is the distinguishing feature. Every piece on SBRS must complete all six PEARLS criteria. A piece that summarizes research findings — even well — without applying PEARLS is not SBRS content.

The PEARLS Framework — Full Definitions

PEARLS is the evaluative spine of every review on this site. All six criteria must appear in every review. Each criterion has specific questions the reviewer must address. Below are the complete definitions and diagnostic questions for each criterion.

P — Peer Review

What it assesses: Whether the study underwent formal, independent peer review before publication, and the rigor and prestige of that review process.

Questions to answer:

Was the study published in a peer-reviewed journal? Which journal? What is that journal’s reputation in the field?

If it’s a policy report (not a journal article): did a credentialed institution (RAND, Brookings, IES, AASA, NSBA, McREL) produce it, and what was their internal review process?

Was the peer review blind (single-blind, double-blind)? If unknown, note that.

Is the publication a predatory or low-quality journal? If so, flag it.

When was the peer review process completed? Is there a lag between the data and publication?

High quality in this criterion: Published in a respected, peer-reviewed journal with documented blind review. Institutional policy reports from RAND, Brookings, IES, or comparable bodies with transparent methodology review.

Low quality in this criterion: Self-published work, conference papers not yet peer-reviewed, reports from organizations with clear advocacy positions and no independent review, publications in journals known for low standards.

What to write: A factual statement of the publication venue, the review process as documented, and an honest assessment of what that means for the study’s credibility. Do not inflate the prestige of a mediocre journal or dismiss a solid policy report for not being a journal article.

E — Evidence Quality

What it assesses: The strength of the empirical evidence — the methodology, sample, data quality, and whether the study design supports the claims made.

Questions to answer:

What is the study’s primary methodology? (Quantitative, qualitative, mixed methods, meta-analysis, case study, survey, regression analysis, randomized controlled trial, etc.)

What is the sample size? Is it adequate for the claims being made?

Is the study longitudinal (data collected over time) or cross-sectional (data from a single point in time)? This matters enormously for causal claims.

What is the unit of analysis? (Individual boards, districts, states, countries?)

Are there threats to internal validity? (Selection bias, confounding variables, measurement issues?)

Does the study make causal claims that its design supports — or does it make causal claims based on correlational data?

Are the measures valid and reliable? (How were key constructs like “board effectiveness” or “governance quality” operationalized and measured?)

What is the data source? Administrative data, surveys, interviews, observations, document analysis?

High quality in this criterion: Large, representative sample; longitudinal design; rigorous controls for confounding variables; validated measures; appropriate statistical methods; design supports the claims made; sensitivity analyses or robustness checks present.

Low quality in this criterion: Small convenience sample; cross-sectional design making causal claims; poor operationalization of key constructs; no controls for confounding factors; survey-based with low response rates; design does not support the strength of claims made.

What to write: A methodologist’s honest read of how strong the evidence actually is. Name the methodology. Assess the sample. If the study overclaims (e.g., implies causation from correlation), say so directly. If the methodology is genuinely strong, acknowledge it.

A — Applicability

What it assesses: How transferable the findings are to real-world school board governance contexts — and what conditions must be present for the findings to apply.

Questions to answer:

What types of districts, boards, or governance contexts were studied? (Urban? Rural? Large? Small? Which state or country?)

How representative is the sample of the broader universe of school boards? (U.S. public school boards? Only one state? Only large urban districts?)

What scope conditions limit applicability? (The findings may apply to large urban districts but not rural ones. They may apply to states with strong open meetings laws but not all states.)

Do the study’s findings require specific contextual conditions to hold? What are those conditions?

Are the findings actionable for a practitioner? What would a board or superintendent actually do differently based on this research?

Is there a gap between what was studied (e.g., board meeting behavior) and what practitioners want to know (e.g., whether board behavior affects student outcomes)?

High quality in this criterion: Broad, nationally representative sample; explicit discussion of scope conditions; practical implications that map to real governance behaviors; findings that can travel across contexts.

Low quality in this criterion: Single-district or single-state case study presented as broadly applicable; no discussion of scope conditions; findings that require resources or conditions most boards don’t have; gap between what was studied and what practitioners need.

What to write: Be specific about what kinds of boards or governance contexts the findings apply to. If the study was done in Texas urban districts, say so — and say what that means for a rural Minnesota board reading this review. Evaluate whether the leap from “this was studied here” to “this applies everywhere” is justified.

R — Replicability

What it assesses: Whether the study’s findings have been replicated elsewhere, whether the methodology is transparent enough to replicate, and how the findings compare to the broader research base.

Questions to answer:

Have other studies found similar results? Name them if possible.

Is this a lone finding, or is it part of an emerging or established pattern in the literature?

Is the methodology described in enough detail that another researcher could replicate the study?

Are the data sources available or documented for replication?

Have attempts to replicate this type of study produced different results? (If so, explain.)

Is this an outlier in the literature, or a convergent finding?

High quality in this criterion: Findings align with prior research; methodology is fully transparent and documented; data sources are accessible or clearly described; the research community has produced corroborating evidence.

Low quality in this criterion: Findings contradict prior literature without explanation; methodology is opaque; data sources are proprietary or inaccessible; this appears to be a lone finding with no corroboration.

What to write: Place the study in the context of the research base. If it’s a replication or extension of prior work, say so and cite the prior work. If it’s a novel finding with no prior corroboration, flag that as a limitation. If the findings are contested, present both sides. This is where secondary citations are most appropriate and encouraged.

L — Limitations

What it assesses: What the study itself acknowledged as limitations, and what it did NOT acknowledge that a careful reader would identify.

Questions to answer:

What limitations did the study authors explicitly state?

What limitations did the study authors NOT state that are apparent from the methodology, sample, or claims?

Are there threats to validity (internal or external) the authors glossed over?

Are there alternative explanations for the findings that the study didn’t adequately address?

Were there data collection problems (e.g., social desirability bias in surveys, self-reporting issues)?

What is the temporal limitation — when was the data collected, and does the governance context it studied still exist?

Are there political, funding, or institutional conflicts of interest that could have shaped the findings?

High quality in this criterion (from the review’s perspective): The review identifies both acknowledged and unacknowledged limitations. It neither ignores what the authors said nor stops there.

What to write: Two-part structure is appropriate here: (1) Limitations the authors acknowledge, and (2) Limitations worth noting that the study did not fully address. Be honest but fair — the goal is not to discredit but to help the reader understand the study’s edges.

S — Significance

What it assesses: What the study contributes to the field of school board governance knowledge, and what its findings mean for practice — particularly for governing in service of student outcomes.

Questions to answer:

What does this study add to what we already know about school board governance?

Is this a confirming study (strengthening existing knowledge), an extending study (applying existing theory to new contexts), or a challenging study (disrupting prior assumptions)?

What are the practical implications for boards, superintendents, state education agencies, or researchers?

How does this research relate to the underlying mission of school boards — serving students?

Does the evidence help us understand how boards can govern more purposefully in service of their mission? If so, how?

Does the significance hold even given the limitations identified in the L section?

High quality in this criterion: The research advances understanding of how governance connects to student outcomes. Practical implications are specific and grounded in the evidence. The significance is proportionate — not overstated given the evidence quality.

Low quality in this criterion (in a review): Significance section that simply restates the abstract without evaluating what was actually contributed. Overclaiming practical importance for a study with major methodological limitations.

What to write: Be proportionate. A solid small study with a clear finding and honest limitations still has significance — describe it accurately. A large study with exciting findings but serious replicability gaps has limited significance until confirmed — say that. The question “does this help boards govern better for students?” should be answered directly.

Section 4 — Grounding Document

Source: Great on Their Behalf (3rd edition) by AJ Crabill Access: Google Drive ID: 1KwiTQKLZlRDnTCzDmHaleeU3IVj_cHbsmA2s6nORdkE (read via mcp__c7e8b32b-5a12-4797-a6d1-d08c6ac76fa5__read_file_content) Usage pattern: a — pre-draft only

Pre-draft only: Before drafting, read the relevant sections of the grounding document. Use it to calibrate the frame, vocabulary, and core claims before writing begins. It is a framing calibrator, not a citation source.

Section 5 — Voice Register

Voice: Evaluative-academic. Structured, precise, and honest.

Register: ajc-long — approximately 23 words per sentence average; 4–8 sentence paragraphs; author blend of Gladwell 35% / Collins 30% / Sinek 20% / Gawande 15%.

Tone characteristics:

Formal but readable. No jargon for its own sake, but technical terms from research methodology are appropriate and expected.

Critical but fair. The review is not a cheerleader for the research and not a hostile critic. It is an honest evaluator.

Does not oversell research. “This groundbreaking study proves…” is prohibited. Findings are findings; proof requires replication and weight of evidence.

Does not undersell research either. Dismissing a solid study for having any limitations is not critical thinking — it’s cynicism. Acknowledge strengths accurately.

Uses PEARLS framework language throughout. Each section is headed with the criterion name and uses criterion-specific framing.

The reviewer has a perspective and states it. Hedging everything to avoid taking a position is not the voice of this site.

Voice examples:

Correct: “The sample size — 47 boards across three states — is adequate for exploratory analysis but insufficient to support the broad causal claims the authors make in the conclusion.”

Incorrect: “The sample size may be seen by some readers as potentially not fully representative of all possible governance contexts.”

Correct: “This is one of three peer-reviewed studies to link board meeting micromanagement behaviors to superintendent turnover. The convergent findings across different samples and methodologies strengthen confidence in this relationship.”

Incorrect: “This study found interesting results that other researchers may also find useful to consider.”

Section 6 — NEVER Rules

An AI generating content for SBRS must NEVER do the following:

Generate a review of a study that has not been verified as a real, accessible, published work

Publish a review with a fabricated or incomplete citation

Skip any of the 6 PEARLS criteria

Write the P, E, A, R, L, or S sections as descriptions of what the study did rather than evaluations of quality and applicability

Name individual board members, superintendents, students, or administrators

Use proper district names in research contexts (use descriptor + state instead: “a mid-sized Midwestern urban district”)

Oversell research with language like “proves,” “groundbreaking,” “landmark,” “definitively establishes,” “revolutionary,” “pioneering”

Dismiss research unfairly with language like “this limited study merely”

Publish without AJ approval OR completion of the the configured write model fallback process

Summarize research findings as if that were the output (this site is an evaluator, not a summarizer)

Skip the Practitioner Takeaways section

Omit the methodology page link

Review a study that is out-of-bounds (opinion pieces, non-empirical work, studies not about school board governance, invented or unverifiable studies)

Review a study that has already been reviewed on SBRS

Name ESB, AJ Crabill, Effective School Boards, GOTB, or why-based governance as brands or frameworks in content

Use numerical scores (e.g., “P: 8/10”) for PEARLS criteria without a documented rubric — prose evaluation is the SBRS standard

Begin any piece with a generic statement about the importance of research or school governance

Section 7 — Generation Process

Every PEARLS review must follow this complete generation process. No steps may be skipped.

Step 1 — Pre-Draft Grounding Document Read

Before any drafting begins, read the relevant sections of Great on Their Behalf (3rd edition) via the Google Drive access listed in Section 4

Use it to calibrate frame, vocabulary, and core claims

This is a framing calibrator, not a citation source — do not cite it in the review

Step 2 — Study Selection and Accessibility Verification

Select the next study from the queue per priority order

Queue priority: (1) peer-reviewed journal articles with clear governance focus; (2) meta-analyses and systematic reviews; (3) RAND, Brookings, IES policy reports; (4) AASA, NSBA, McREL research publications; (5) state education agency longitudinal studies; (6) practitioner research briefs with documented methodology

Verify the study is real, accessible in full text, in-bounds, and not previously reviewed

Confirm full APA citation is constructible and accurate

If the study fails any check: return to queue, select next study

Step 3 — Study Reading and Note-Taking

Read the complete study (full text, not abstract only)

Extract: research question, methodology, sample, findings, statistical approaches, author-acknowledged limitations, data sources, funding sources

Note anything unusual, questionable, or noteworthy about the design or claims

Step 4 — PEARLS Assessment Notes

Work through the diagnostic checklist for each of the six criteria (see below) before writing prose

Draft raw assessment notes for each criterion

Identify secondary sources to cite in R and S sections

Diagnostic checklist — P (Peer Review):

What is the publication venue?

Is it listed in Cabell’s or Scimago? What quartile?

Does the journal description indicate blind peer review?

Is this a predatory journal? (Check against known predatory journal lists if uncertain)

If policy report: what institution published it? Is their review process documented?

Diagnostic checklist — E (Evidence Quality):

What is the stated methodology?

What is the N (sample size) and what is the unit of analysis?

Is the design longitudinal or cross-sectional?

What is the data source?

Do the statistical methods match the design and claims?

Are causal claims supported by a causal design, or are they inferred from correlation?

How are key constructs defined and measured?

Diagnostic checklist — A (Applicability):

What is the geographic scope? (Nation, region, state, district, school?)

What is the demographic and contextual scope? (Urban/rural/suburban? District size? State policy context?)

What conditions must be true for the findings to transfer?

Are there explicit scope conditions stated by the authors?

What would a practitioner DO differently based on this research?

Diagnostic checklist — R (Replicability):

Is the methodology fully described?

What prior research corroborates or contradicts these findings? (Cite specifically)

Have similar studies been conducted? What did they find?

Are data sources available for replication?

Is this a lone finding or a convergent pattern?

Diagnostic checklist — L (Limitations):

What limitations did the authors state explicitly? (List them)

What limitations did the authors NOT state that are visible from the methodology?

Are there alternative explanations for the findings?

Are there data quality or measurement issues?

Are there temporal limitations (when was the data collected)?

Are there conflicts of interest (funding source, institutional affiliation)?

Diagnostic checklist — S (Significance):

What does this study add that we didn’t know before?

Is this confirming, extending, or challenging prior knowledge?

What are the practical implications for boards or superintendents?

Does this research connect governance behaviors to student outcomes? How directly?

Is the significance claim proportionate to the evidence quality and limitations?

Step 5 — Practitioner Takeaways Identification

Identify what a practitioner can reasonably conclude from this study (given its PEARLS profile)

Identify what a practitioner should NOT conclude

Draft 2–4 specific actionable takeaways

Calibrate: given the evidence quality, how much weight should a practitioner give this?

Step 6 — Full Review Draft Generation

Primary model: the configured write model

Fallback model: Claude Opus 4.8 (if primary is unavailable or produces output that fails multiple steps after revision attempts)

Generate the complete review following the section structure in Section 3

Include all required elements: citation header, framing paragraph, methodology page link, all 6 PEARLS H2 sections, Practitioner Takeaways

Step 7 — PEARLS Completeness Check

Verify all six criteria are present as H2 sections

Verify each section meets the minimum word count (80 words minimum per section)

Verify each section is substantive and evaluative, not descriptive or thin

Flag any section that describes rather than evaluates — rewrite if needed

Step 8 — Citation Verification (Hard Gate)

Verify the primary study citation is complete, accurate, and in proper APA format

Verify all secondary citations are real and traceable

Confirm DOI or URL resolves

This is a hard gate: the review cannot proceed if citation verification fails

A review generated from an abstract only is insufficient — full text must be accessible

Step 9 — Balance Check

Are the study’s genuine strengths acknowledged?

Are the study’s genuine limitations identified, including limitations the authors didn’t acknowledge?

Does the overall PEARLS evaluation feel proportionate to the actual evidence quality?

Is the language calibrated appropriately? (Not overselling; not dismissing unfairly)

Step 10 — Quality Checks

Run all quality checks in Section 8 (8a through 8g) in sequence. No check may be skipped.

Step 11 — Submit to AJ for Approval

Submit completed review to AJ for approval

Log the submission time to begin the 72-hour clock

Await response

Step 12 — Publish or Fallback

If AJ approves: publish immediately

If AJ requests changes: revise and resubmit (clock resets)

If no response in 72 hours: execute the configured write model quality and validity recheck per Section 10, then publish if checks pass

Section 8 — Quality Checks

Quality Check Failure Protocol

This protocol governs every check in Section 8 (8a through 8h, including all site-specific gates in 8d).

On any failure at any check:

Identify the specific violation — which rule, which tier, which gate, and exactly what was wrong

Revise only the affected portion of the draft to address that specific failure

Re-run the check that failed (not the full Section 8 — just the failing check)

Track the attempt count per check

Retry limit: 3 revision-and-retry cycles per check. If a check passes on retry, continue to the next check normally.

Ultimate Failure: if a check still fails after 3 retries, declare an ULTIMATE FAILURE for that check. Stop all further quality checks immediately. Do not publish, commit, or submit the draft. Proceed to ultimate failure reporting (below), then preserve the draft and full failure log — including stage name, attempt count, nature of each failure, and each revision attempt — in a dated file for AJ’s review.

Ultimate Failure Reporting — SBRS: Do NOT submit for AJ approval. Submit a failure report to the ESB content approval system at esby.effectiveschoolboards.com/admin/. Include: site (SBRS), research topic, which check failed, attempt count, nature of each failure, and the draft. Mark status as FAILED.

After reporting, preserve the draft and the full failure log (stage, attempt count, nature of each failure, revision attempts) in a dated file for AJ’s review.

8a. Anti-Slop Scan

Tier 1 — Kill on sight (any single occurrence fails; revise before proceeding): delve, utilize, facilitate, tapestry, paradigm, synergy, holistic, catalyze, juxtapose, myriad, plethora, pivotal, underscore (as verb), bolster, multifaceted, foster, seamless, embark, transformative, revolutionize, realm, beacon, harness (as verb)

Tier 2 — Suspicious in clusters (3 or more in a single paragraph = flag; revise that paragraph): comprehensive, cutting-edge, streamline, empower, enhance, elevate, optimize, scalable, intricate, profound, resonate, cultivate, galvanize, cornerstone, game-changer, robust, innovative

Tier 3 — Filler phrases (each one weakens the piece; eliminate all): “it’s worth noting,” “furthermore,” “moreover,” “additionally,” “in conclusion,” “to summarize,” “it is important to,” “one must consider,” “at the end of the day,” “when all is said and done,” “it goes without saying,” “needless to say,” “this is a testament to”

Structural tics (each one that appears is a flag):

Paragraph opener is a transition word (Moreover, Furthermore, Notably, However, Additionally, Importantly, Ultimately)

Any em-dash ( — ) or en-dash; the required form is a spaced double-hyphen (“–”), used sparingly (more than 1 “–” in a single paragraph is a flag)

Soft repetition: same idea restated in different words within 3 consecutive paragraphs

Scoring: 0–1.5 = clean (proceed); 1.5–3.0 = minor revision needed before proceeding; >3.0 = full revision required before red-team

Site-specific additions (SBRS):

The following additional language patterns are banned on SBRS beyond the standard Tier 1/2/3 above:

Hollow opener: Does the piece begin with a generic statement about the importance of research or school governance? Delete it. Begin with substantive content — the citation header followed by the framing paragraph.

Filler transitions: “It is important to note that,” “It is worth mentioning that” — remove any occurrence.

Hedge stacking: Remove stacked hedges (“may possibly suggest that some evidence could potentially indicate”). State the finding, then acknowledge uncertainty if genuine.

Definition bloat: Do not define the PEARLS framework in the body of the review. The methodology page handles that. Reference it with a link and move on.

Conclusions that exceed evidence: Flag and remove or soften any takeaway or conclusion that the evidence does not directly support.

Overselling language (specific to research context): Remove or rewrite any of the following: “groundbreaking,” “proves,” “definitively establishes,” “landmark,” “revolutionary,” “pioneering.” Use instead: “provides evidence for,” “suggests,” “finds,” “contributes evidence that.”

Underselling language: Also flag unwarranted dismissal. “This study merely” or “this limited study” used for a methodologically sound study is inaccurate and unfair. Be proportionate in both directions.

8b. Voice Check

Voice Check — Moderate Fidelity (ajc-long)

Run a moderate voice check. No formal scoring threshold — use judgment. If a colleague would describe this as reading like AI output, revise before proceeding.

Core NEVER rules (must be clean):

No Tier 1 anti-slop words (already checked in 8a)

No sentence-initial transitions: Moreover, Furthermore, Notably, Additionally, Importantly, Ultimately

No exclamation marks anywhere in the piece

No ellipses

No curly quotes

No voice hedging: no “I feel/think/believe” as epistemic hedges

Fingerprints (must be present):

Oxford comma: always (a, b, and c)

Straight quotes only (no curly quotes)

Contractions used naturally (don’t, it’s, we’ve)

Rhythm check:

Sentences average ~23 words; paragraphs 4–8 sentences

No extended run (3+ consecutive) of sentences under 8 words

Reading level appropriate for a practitioner audience

8c. ESB Governance Language Check

ESB Governance Language Check

Reference: Google Drive ID 13O3_qNxtRVHsTDyDMXdotPLOhIMTyRe0J_1yQ9tzhOQ

Banned phrases — any occurrence fails:

“micromanage” / “micromanaging” → describe the specific behavior (is the board operating within its governance role or directing implementation?)

“rubberstamping” / “rubber stamp” → describe whether the board exercised informed judgment

“stay in your lane” → reference governing policies or board/superintendent role distinction

“good/bad/right/wrong” as personal moral judgment → frame around effectiveness and outcomes

Banned framings — restructure any instance found:

Board as passive recipient: any suggestion that the board should simply receive, defer, or accept without agency

Board as obstacle: any framing that board involvement is a problem to manage around

Expertise as authority: any framing that professional staff have more legitimate authority than elected board members

Conflict as dysfunction: any framing that disagreement among board members is inherently bad

Efficiency as primary goal: any framing that speed or non-disruption is the board’s primary objective

Education terms — replace with specific language:

“school district” or “district” → “school system”

“equity” → describe the specific gap or outcome with data (reserve the term only for direct statutory quotation)

“achievement gap” → describe the system failure directly with specific data

Politically-coded terms → describe the specific concept: DEI, anti-racist, woke, “parents’ rights” (as contested framing), SEL (as wedge term)

8d. Site-Specific Checks

PEARLS Quality Gates

Criterion inflation check: Has the review given high or satisfactory ratings across all six PEARLS criteria without evidence of genuine differentiation? No study is uniformly strong across all criteria. If the review reads like a uniform endorsement, flag it — the evaluation is not genuinely critical.

Abstract regurgitation check: Does the E or A section primarily restate what the abstract or introduction said? If so, the reviewer read the abstract but not the full methodology. Return to Step 3 (Study Reading) and read the full study.

Missing limitation check: Are there obvious methodological limitations visible from the study’s design that are not mentioned in the L section? Cross-sectional design used to make causal claims? Small convenience sample? Survey with low response rate? These must be named.

False precision check: Has the review assigned numerical scores to PEARLS criteria (e.g., “P: 8/10”) without a documented rubric? Either use a documented rubric or remove numerical scores. Prose evaluation is the SBRS standard.

Passive voice overuse: Where the sentence is clearer in active voice, use it. “The researchers found” over “It was found by the researchers.”

Citation and Source Fidelity Gates

Primary citation hard gate: The study being reviewed must be cited in full APA format at the top of the review before any body text. The citation must include all authors, year, full title, journal name or publisher, volume/issue/pages where applicable, and DOI or stable URL. If a full accurate citation cannot be constructed, the review cannot be published.

DOI / live-link verification: Confirm the DOI resolves or the URL is accessible before proceeding. A review generated from an abstract only is insufficient — full text must be accessed to evaluate evidence quality and limitations.

Secondary citation integrity: Can each secondary citation cited in R (Replicability) or S (Significance) actually be verified as making the claim the review attributes to it? Do not cite a study for a finding it doesn’t actually report.

Research accuracy test: Does the review accurately represent what the study actually found and how it found it? A hostile reader who has read the same study should not be able to say “that’s not what the study said” about any claim in the review.

PEARLS validity test: Are the PEARLS assessments grounded in what the study actually reports, or are they based on what the reviewer assumes the study did? E.g., if the review says “the study used a longitudinal design,” is that documented in the study, or assumed?

Balance and Proportionality Gates

Balance calibration: Are the study’s genuine strengths acknowledged? Are the study’s genuine limitations identified, including limitations the authors didn’t acknowledge? Does the overall PEARLS evaluation feel proportionate to the actual evidence quality?

Expert plausibility test: Would a research methodologist with expertise in educational administration research agree with the E (Evidence Quality) and L (Limitations) assessments? If a methodologist would find the evidence quality assessment naive or the limitation identification inadequate, revise.

Existential Content Check

Apply the SBRS existential check: “Is this piece evaluating the quality and applicability of a specific study using the PEARLS framework, or is it doing something else?”

If the honest answer is “something else” — even if the content is high quality — it does not belong on SBRS. Find the correct site for it.

Apply the voice test: Does this read like a critical evaluator, or does it read like a summarizer or an advocate? If it reads like a summarizer or advocate, revise to the evaluative-academic voice.

Sensitive Content Check

ESB and Why-Based Governance References:

Never name ESB, AJ Crabill, Effective School Boards, GOTB, or why-based governance as brands or frameworks in content

The evaluative lens of ESB’s five practices (Focus, Clarify, Monitor, Align, Communicate) may inform the Applicability and Significance criteria but should not be named explicitly

May link to effectiveschoolboards.com or whybasedboards.com 1–2 times per year, only in a list of 3+ links

8e. Primary Red-Team (the configured write model)

Primary Red-Team — Single Adversarial Pass (the configured write model)

Run the draft through the configured write model with an adversarial prompt instructing it to:

Challenge every major claim: is this actually supported? Where could a knowledgeable reader push back?

Find logical gaps: does the argument hold together from premise to conclusion?

Flag unsupported assertions: any claim that needs evidence and doesn’t have it

Identify reader-hostile moments: where would a skeptical reader stop reading?

Surface structural problems: does the piece earn its conclusion or merely assert it?

Fails if: any major claim is undefended; the argument contains a logical gap; or more than 2 reader-hostile moments are identified. Revise and re-run before proceeding.

8f. Council Red-Team (multi-LLM panel)

Council Red-Team — Multi-LLM Adversarial Panel

Spawn these 4 free OpenRouter models in parallel, each with a distinct adversarial persona. Apply the Section 8 failure protocol: if the council fails, revise the flagged section(s), re-run only the personas that flagged a problem (not all 4), up to 3 retries.

Persona prompts:

Skeptical Practitioner (Gemini): “You are reviewing this piece as a school board practitioner who has seen a lot of governance advice fail in practice. Would this actually work in a real school system? Is this grounded in how governance actually operates, or does it assume conditions that rarely exist? List every claim that feels untested or idealistic, quoting the exact passage.”

Hostile Reader (Llama): “You are looking for something to object to. What is the weakest claim? What is most poorly supported? What would make you stop reading? Be specific — quote the exact passage and state the exact objection.”

Rival Author (DeepSeek): “Is this saying anything the field does not already know? Is there a genuinely new insight here, or is this a repackaging of received wisdom in different words? What would need to change for this piece to be worth reading over existing work on this topic?”

Editor (Qwen): “Is the argument tight? Are there redundant sections? Does the structure serve the argument or fight it? Is the opening strong enough? Does the conclusion earn its landing? Would you accept this piece as submitted, or send it back with notes?”

Passing criteria: 3 of 4 models must find no major problems. If 2 or more models independently flag the same specific passage or claim, that passage fails regardless of overall score.

8g. Existential Voice Test

Existential Voice Test

Ask: “Would this content appear unchanged on another governance education site, practitioner blog, or AI-generated newsletter?”

If yes: the voice is not distinctive enough. The piece is competent but not AJ’s. Identify 2–3 specific passages that read generic and revise them until they could not have been written by anyone else.

Indicators of failure:

Opens with a general statement rather than a specific scene, claim, or reframe

Uses hedged language (“it’s important to consider”) instead of direct assertion

Could be published without attribution and no one would notice the absence of a byline

This test is subjective. If you would not be surprised to see this exact piece on a competitor site, it fails.

Section 9 — Technical Filing

Front Matter Template

---

layout: research

title: “[Study Topic]: A PEARLS Review”

study_type: “[Valid study_type value]”

date_display: “[Month Day, Year]”

---

Valid study_type values:

Peer-Reviewed Journal Article

Policy Report — Major Institute

State Education Agency Study

Meta-Analysis or Systematic Review

Practitioner Research Brief

File Naming and Path

File path: research/[topic-phrase].html

Topic phrase: lowercase, hyphenated, descriptive of study subject (not the study title)

Examples: research/board-governance-student-achievement.html, research/superintendent-tenure-stability.html

Study Citation Header Format

Every review opens with the full APA citation of the study being reviewed — before the body content, not in a references section at the bottom. It is the first thing the reader sees after the title.

For journal articles:

Study Under Review:

Author Last, F. M., & Author Last, F. M. (Year). Title of the study in sentence case. Journal Name, Volume(Issue), page–page. https://doi.org/xxxxx

For policy reports:

Study Under Review:

Organization Name. (Year). Title of the Report. Publisher. https://url

The full APA citation must include: authors (all of them), year, title, journal or publisher, volume/issue/pages (if journal), and DOI or URL.

Pre-Publication Technical Checklist

Before final submission and again before publishing, verify each item. Every item is required.

Content:

Full APA citation of the reviewed study appears at the top, before body text

Framing paragraph (2–3 sentences) is present and does not summarize findings

Link to /methodology.html is present in framing paragraph or first PEARLS section

All six PEARLS criteria are present as H2 sections, in order (P, E, A, R, L, S)

Each PEARLS section is substantive (minimum 80 words)

Practitioner Takeaways section is present as H2 and is the final section

Practitioner Takeaways includes both what to conclude AND what not to conclude

Total word count is 800–1,200 words (body text only)

Citations:

Primary study citation is complete (all authors, year, title, journal/publisher, volume/issue/pages, DOI or URL)

All secondary citations are real and traceable

DOI or URL for primary study resolves

Voice and balance:

No overselling language (“proves,” “groundbreaking,” “landmark”)

No unjustified dismissal language (“merely,” “only”)

Limitations are honestly represented, including unacknowledged limitations

Practitioner takeaways are proportionate to evidence quality

Sensitive content:

No student names

District names in research context use descriptor + state (not proper district names)

No individual board member, superintendent, or administrator names

Researcher/author names used only in citation context

Technical:

Front matter complete: layout: research, title, study_type (valid value), date_display

File path follows research/[topic-phrase].html convention

File name is lowercase, hyphenated, descriptive of study subject

Output contains ONLY front matter + body HTML (no DOCTYPE, html, head, style, nav, header, footer, or body tags)

Approval:

Review submitted to AJ for approval with submission timestamp logged

Either AJ approval confirmed, or 72-hour clock expired and the configured write model recheck completed and passed

Section 10 — Independent Review Protocol

AJ Approval Workflow

Every review requires AJ approval before publishing.

The process:

Completed review is submitted to AJ for approval

The 72-hour clock starts at the moment of submission

If AJ approves, publish immediately

If AJ requests changes, make changes and resubmit (clock resets)

If AJ does not respond within 72 hours:

Conduct a full quality and validity recheck using the configured write model (see below)

If the configured write model recheck passes, publish

If the configured write model recheck identifies issues, resolve issues before publishing

What AJ approval covers:

Accuracy of the PEARLS evaluation

Tone and voice calibration

Citation accuracy

Any unusual or sensitive content decisions

the configured write model Fallback Process

When the 72-hour clock expires without AJ response, the configured write model performs a full quality and validity recheck covering:

Citation accuracy: Does the citation match the actual published work?

PEARLS completeness: Are all six criteria present and substantive?

Evidence representation accuracy: Does the review accurately describe the study’s methodology and findings?

Balance calibration: Are strengths and limitations proportionately represented?

Limitation adequacy: Are significant methodological limitations identified?

Practitioner takeaways reasonableness: Are the takeaways grounded in and proportionate to the evidence?

Voice and tone compliance: Is the review written in the evaluative-academic voice of SBRS?

Technical checklist: Does the review meet all technical publishing requirements?

If the configured write model identifies any issues, they must be resolved before publishing. If all checks pass, publish.

Ongoing Content Integrity

Queue management and duplicate prevention: The review library must be checked before any study is queued for review. A study that has already received a PEARLS review on SBRS should not receive a second one unless the original review is being substantially updated due to new information, or the original review was withdrawn. Maintain a log of reviewed studies including: study citation, date of SBRS review, file path of review.

Recency vs. classic studies: The queue should maintain a balance between new publications (within 12–24 months) and classic or foundational studies not yet in the SBRS library. Target ratio: approximately 60% new publications (within 24 months), 40% classic or foundational studies.

Research priority areas: When multiple studies are in the queue, favor studies in these areas:

School board governance and student achievement connections

Board-superintendent relationship and its effects on school system stability and outcomes

Board governance practices (focus, monitoring, goal-setting) and their measured effects

Superintendent tenure, stability, and transitions

Board member characteristics, selection, and training and their relationship to governance effectiveness

Board policy governance models — evidence base and outcomes

Community engagement and board governance

Board governance in turnaround or improvement contexts

Update policy — retracted or challenged studies:

Retracted: Update prominently with a retraction notice at the top. The PEARLS evaluation stands as a historical record but must be clearly marked.

Subject to significant published critique or replication failure: Add a dated addendum noting the critique or replication failure.

Corrected by authors (erratum): Note the correction in an addendum, reassess whether the PEARLS evaluation changes materially, and update accordingly.

In all cases, the original review is preserved as a record. The update is additive, not a replacement. Date all addenda clearly.

Cross-site linking rules:

SBRS may link to sister sites at a rate of 1–2 times per year, only in a list of 3 or more links, and only when the link adds genuine value to the reader

Never name ESB, AJ Crabill, GOTB, or why-based governance as a brand or framework in SBRS content

May link to effectiveschoolboards.com or whybasedboards.com 1–2 times per year only in a list of 3+ links

Section 11 — Learned Rules

(No site-specific learned rules recorded as of 2026-06-26.)

Learned Rules — Generation Protocol

Trigger: When what AJ published differs substantively from what was submitted to the content approval system, generate a new learned rule and append it here.

Process:

Compare the submitted version against the published (live) version

Identify every substantive edit AJ made — not typos; look for content changes, framing changes, word choice changes, structural changes

For each cluster of related edits, derive the governing rule

Append the rule using this format:

Rule [N] — [Short label] (added [date]) Rule: [The rule stated in one sentence — what to do or not do] Why: [What the original got wrong — what pattern triggered the edit] Pattern to avoid: [Specific construction, phrase type, or framing to watch for] Example: Original: “[verbatim or close paraphrase of what was submitted]” → Published: “[verbatim or close paraphrase of what AJ published]”

End of Content Guide — SBRS (schoolboardresearch.org) This document is the authoritative operational guide for AI-generated PEARLS reviews. All generation processes must follow this guide exactly.