A seminal paper by Roediger and McDermott (1995) observed that studying (similar to thinking about) materials was associated with a high likelihood of false recall and false recognition of materials similar to the studied items. This research has spawned hundreds of follow-up studies, none of which address your question entirely but converge on a plausible explanation: As you try to recall the specifics of the event, you likely recall a number of things that actually occurred (e.g., you were drinking rum, there was a pretty girl in the corner, and it was getting late). Considering these true specifics, however, according to this line of research, may introduce associated memories which never actually occurred (e.g., the pretty girl winked at you). Think these thoughts long enough and you'll have what's called a source monitoring error -- that is, you'll conflate thinking about the pretty girl winking with a memory that she actually did wink. That might be why you feel as if you had fewer false memories when you don't consider specifics -- there are fewer opportunities to make false associations. (It's also interesting to note that repeated testing, while improving "correct" memory, also enhances false memory. I can provide more links if you're interested.) If you're interested in other things that cause false memory (at least in these laboratory studies), here's a link.
Most of this is research my group is associated with, generally because I'm most familiar with that literature. I like this line of studies (called DRM studies) because they show false memories can be generated in a matter of moments (or "the morning after"). Loftus's work is fascinating, too, because she uses more naturalistic materials, those more pertinent to real life. Definitely two sides of the same coin.
Leave a comment if I can provide more info.