Reporting bias occurs when only some trial outcomes are reported. This can occur within a trial, so that researchers looked at many outcomes, but only report the results of some. It can also occur with entire trials, so that some research that is done never gets published (also referred to as publication bias). In general, positive results are much more likely to be reported than negative results.
Time lag bias is another type of reporting bias that occurs when positive or “exciting” results are published quicker than negative or less exciting results.
The location that research ends up published can also result in bias. Positive and “more exciting” research tends to end up in larger journals, which are read and cited more often, potentially leading to a skewed sense of the literature.
Duplicate publication of the same trial results is another type of publication bias.
A common, related problem in systematic reviews is language bias. Authors who only read English frequently limit their searches to English language research, which may not capture all important studies in an area.
For a great book that covers this topic, check out Bad Pharma by Ben Goldacre.
The impact of outcome reporting bias in randomised controlled trials on a cohort of systematic reviews. Kirkham J.J. et.al. BMJ. 2010; 340:c365