Track what each source measured, when, and with whom. Build a claim supported by both, but do not turn one sample into a universal result. Preserve a limitation when it changes what can reasonably be concluded.
Worked example
A survey of preferences and a count of actual visits answer different questions. A synthesis can compare them, but a stated preference is not identical to observed attendance.
The Workshop Schedule
Source A: In a survey, 60 of 100 current workshop members preferred Saturday sessions. Nonmembers were not surveyed. Source B: During a four-week trial, Saturday sessions averaged 24 attendees and weekday sessions averaged 18. Saturday sessions also offered a new topic, so schedule and topic changed together. Both sources were prepared for a decision about a future schedule.