Meta-analysis may make things more confusing

What with possible publication bias (null hypothesis fulfilled is pretty boring) or rare sourcing of actual raw data, how is one supposed to know what's right?