- Human error: Humans make mistakes. It is a simple fact of life, and many of the mistakes are unintentional.
- Mistaken assumptions: Any endeavor presupposes a number of assumptions, many of which we do not realize we are making. Some of those assumptions may be wrong. If we are explicit about our assumptions we have a better chance of identifying which ones are mistaken. Unfortunately, we cannot be explicit about all of our assumptions.
- Mistaken observations: For whatever reasons, data can be recorded incorrectly. It is hard enough to tell when we make a mistake, it is often harder to say why someone else made a mistake, particularly when that individual is no longer around to ask.
- Wrong filter: We may be looking at a problem from a certain point of view. We may think we have the data explained but we may be looking at a problem the wrong way.
- Academic fraud: This is where individuals involved in research either invent their data or egregiously distort it so that it fits their theory. (Plagiarism is a separate case because the data may not be fraudulent, only the claim that it represents the author's or authors' own work.)
[Addendum: Another report on this incident with thoughts about how the system encourages academic fraud and works against its discovery. The follow-up is a bit more meaty than the initial one.]