Publication bias might happen when studies aren’t published because they report negative results, or they aren’t statistically significant. The author may be discouraged to continue the research, or journal editors/reviewers may not want to accept the research. However, if the study was designed well, then these studies should be published.
A good example of this are studies funded by pharmaceutical companies, which may not publish data if the results are negative because that would impact sales of their drug.
As a result, studies that are only positive may be published, and this can be evaluated using a funnel plot.