Editorial Policy
How PolityLens handles data, builds scores, uses AI, and stays nonpartisan. These are not aspirational values. They are operational rules we follow on every page, for every official, regardless of party.
Our Mission
PolityLens exists to make government transparent and officials accountable. We are not a media outlet, a campaign tool, or an advocacy organization. We do not take positions on policy and we do not editorialize. We present verified data, clearly sourced, and let users draw their own conclusions.
Core Principles
Every data point has an identified official government source. News articles and third-party analysis are never primary sources.
We show the record. We do not label it. We show a voting record but don't call a member consistent or hypocritical; we show lobbying alignment but don't call it corrupt or coincidental.
Every official receives identical treatment regardless of party. If a rule applies to a Democrat it applies to a Republican.
When data is unavailable we display the gap honestly, without editorializing what it means.
No crowdsourced data: Wikipedia, Wikidata, and the like are never used as primary sources for the record. The one narrow exception is a low-stakes biographical field, a date of birth for some state legislators, which an aggregator may fill only when no official source exists. It is labeled as such and never overwrites an official source.
Scores & Metrics
Every score is computed from a documented formula. Full definitions live on the methodology page.
Share of roll-call votes where the member cast Yea / Nay / Aye / No. Present, Not Voting, and unrecorded count as missed. Voice votes are excluded.
Share of substantive (non-procedural) votes matching the majority of the member's own party, counted only where the member was present and each party had a clear majority. The full definition is on the methodology page.
Share of votes matching the White House's stated position in an official Statement of Administration Policy (SAP). Only votes on bills with a matching SAP are counted. Applied identically regardless of which party holds the White House.
How often a member's votes matched the position an organization lobbied for, shown as voting overlap with the underlying votes and a clear disclaimer: this is correlation, not proof of a financial relationship or quid pro quo. We present it as alignment, never as influence.
Sourced from FEC filings. We display what was reported. We do not speculate about unreported or dark money beyond disclosing that it exists.
AI-Generated Content
It translates official documents into plain English and describes what a law does. It does not assess whether a policy is good or bad.
Summaries describe what changes for each group. They do not label outcomes as positive or negative.
Every AI summary displays the model name, version, and the date the analysis was produced.
The AI is always given the official source document (bill text, Federal Register entry), never news articles or third-party analysis.
Omnibus bills are analyzed section by section. The summary notes that identified provisions are the most significant within each section, not necessarily the full document.
What We Cover
Congress
Voting records from official House Clerk and Senate roll-call XML files. We display how a member voted, not what it means. Metrics are computed from documented formulas disclosed on the methodology page.
White House
Executive orders from the Federal Register: the order, its stated purpose as written, and its status. We do not characterize legality, constitutionality, or policy merit. SAPs are parsed to extract the administration's position using consistent, documented rules.
State Legislators & Governors
Voting records, bill sponsorships, and committee activity are sourced directly from official state legislative data feeds. We apply the same attendance and alignment metrics used for federal members wherever the underlying data supports it. Coverage expands state by state as direct data pipelines are built.