I can tell you Bluestem that at the UM-KC SOM starting in the 90s the rule was if you did not substantial contribute to the paper, your name didn't appear on it.
Are you saying that the authors of a study that is proved to be junk suffer no professional repercussions?
I am well aware of the term "substantial contributions" to a paper. Care to operationalize that term, however? Good luck. As Bruce notes, support staff, grad students, etc. have expectations that their names are included, regardless of whatever "substantial" means for that particular department and that particular university. Tenure and grant money depend on being published, which introduces its own form of scientific bias. If by "junk" you mean ultimately incorrect, well that depends. As you are aware, science is not linear. It is full of dead-ends and false starts. There are plenty of famous scientists who have been dead wrong on occasion and their reputation suffered little for it. If by "junk" you include falsifying data, then yes, we are in total agreement. (See Andrew Wakefield, his autism "research," and the multiple authors on the fraudulent article in The Lancet ). One of my experimental design classes in grad school included critiquing the methodology and statistical analysis of various studies. We would dismantle the studies and feel pretty superior until our professor would ultimately say in his Russian accent, "Yes, but it got published." One study is one study, no matter how many authors. Replication is vital. But few academic journals publish pure replication studies, and no one is getting tenure or multi-million dollar grants based on a CV full of replication studies. Such is science.