This usually happens with the passage of time and professional police officers are well aware of it. Research has shown that it is relatively easy to introduce false memory. Media coverage or written recollections by other witnesses - if reviewed by a witness - could suggest "facts" to the witness. Witnesses will often discuss the occurrence amongst themselves and false memories can result. The most reliable statements are those taken almost immediately following any incident, statements taken months after an event are usually less reliable and open to question.
I agree with the first part of that, and I'd agree more or less with the last part - except that there is no way of knowing what was actually said as opposed to what got typed up in the summary statements.
What I did read, but still haven't been able to find again, was advice to witnesses of any crime to write down their recollections to try to preserve the initial memories. On the other hand, I can't see how that's always true.
To take a different setting, for some reason I sometimes get cold-callers from research institutes and if I find them genuine and if I have a few minutes, I usually oblige. I know that I've sometimes answered questions, then realised later that what I'd said wasn't quite correct. If they'd rung me back, I'd have corrected what I'd said, but they never do.
Sometimes I get annoyed with research questions (social or marketing, for example) which generally only offer a few options. A lot of the time, I find that answers don't necessarily fit with the options available, which ends up skewing the results. Some of the more interesting ones, IMO, were those that didn't stop at the set multiple-choice answers, but also offered an opportunity for interviewees to explain to what degree they agreed or didn't and why.
