Compare sources by scope, timing and method before treating them as contradictory. Two records can answer different questions. Synthesis should preserve what each supports and identify what neither establishes.
Worked example
One record counts entries and another counts unique visitors. Different totals may result because the units differ, not because one record is false.
Original fiction: Entries and Visitors
Two fictional records describe a workshop day. Record A lists 48 entries through the door; it counts a person again after each return from outside. Record B lists 31 distinct registered visitors; each registered person appears once. Neither record identifies all unregistered visitors. A draft report calls the totals contradictory. A reviewer asks the writer to distinguish entry events from people before comparing them.