Respondents may misremember details, interpret questions differently or give an answer that reflects well on them. These effects can influence surveys about technology use, performance and perceived benefits.
A self-report can still provide useful insight into perceptions. To study actual outcomes, combine it with independent records or clearly distinguish reported experience from verified results.
ELI5
For example, a manager might remember a new tool saving hours every week because several examples stand out. A time log could show the typical saving was smaller.
The survey tells you how the manager sees the tool. The time log helps check how much work actually changed.
Does a large survey remove self-report bias?
No. More responses can improve precision, but they cannot eliminate a systematic tendency to overstate or understate an outcome.
How can researchers reduce this bias?
They can ask specific questions about recent events, protect anonymity where appropriate and compare responses with independent evidence.

