What level of work merits authorship on an academic paper?

I know this topic is pretty specific, but I think we have enough academics and people familiar with academia to make it a good discussion.

A couple experiments in my lab have finished the arduous data collection process and are finally on their way to becoming papers. This has gotten me thinking about authorship: how much work do you need to do to deserve to be an author?

When I was an undergrad RA, there were multiple projects on which I did all or the bulk of the data collection, but it was made clear to me that authorship was not an option for RAs. I assumed it was because we weren’t involved with developing the experiments intellectually.

But now that I’m a grad student, I find that we all tend to collaborate and bounce ideas off each other, and sometimes one person’s, “hey, have you thought about…?” will be the key to turning data from garbage to publishable or even very exciting. But does that one idea really merit authorship? How much work and what kind of work do you really need to do to merit authorship?