Build an evidence dossier
Separate what five sources measure before comparing their claims.
One programme, five types of evidence
This is a fictional Pontechiaro case. Four free repair workshops ran on four Saturdays in May 2025. An administrative register records 96 entries, 61 distinct registration codes, and 68 object forms. An organiser’s public note calls the entries “almost 100 citizens.” A voluntary questionnaire has 37 replies, including 31 people who might attend another series. Finance records €1,460 in invoices but no value for volunteer time. One resident reports a transport difficulty. These sources can all matter without answering the same question. Start a claim–source–unit–limit table.
- Il registro conta novantasei ingressi, non persone diverse.· The register counts 96 entries, not different people.
- La nota pubblica parla di «quasi cento cittadini».· The public note speaks of “almost 100 citizens.”
- Il questionario raccoglie trentasette risposte volontarie.· The questionnaire collects 37 voluntary replies.
- Il promemoria finanziario elenca millequattrocentosessanta euro di fatture.· The finance memo lists €1,460 in invoices.
A conflict can be a category change
The register says 42 of 68 objects worked at the end of a session, 16 could not be repaired there, and 10 awaited parts or later checks. This is not a one-month durability result. The press note’s “more than 60%” is correct for 42/68 immediate outcomes, but can mislead if it suggests lasting repair. The resident’s message raises a transport question; it does not measure the whole city. Before you reconcile claims, identify whether they disagree on fact, denominator, timeframe, or interpretation.
- Quarantadue oggetti funzionavano al termine della sessione.· Forty-two objects worked at the end of the session.
- Dieci schede restavano in attesa di pezzi o verifiche.· Ten object forms awaited parts or checks.
- Una percentuale corretta può usare una cornice fuorviante.· A correct percentage can use a misleading frame.
- Una testimonianza individuale suggerisce una verifica mirata.· One person’s account suggests a targeted check.