Is there any research on the prevalence of academic theft?

Is there any research on the question: how common is academic theft, really? Such as surveys of people having experienced (or committed!) such theft according to an appropriate definition, possibly compared to peoples perception as to the risks.

See the related articles:

De Vries, Raymond, Melissa S. Anderson, and Brian C. Martinson. "Normal misbehavior: Scientists talk about the ethics of research." Journal of empirical research on human research ethics: JERHRE 1.1 (2006): 43.

and

Martinson, Brian C., Melissa S. Anderson, and Raymond De Vries. "Scientists behaving badly." Nature 435.7043 (2005): 737-738.

Among a sample of 3,247 NIH-funded scientists in the United States, asked about the behavior "Using another's ideas without permission or giving due credit":

• 1.4% said they themselves have engaged in this behavior within the last three years
• 45.7% agreed with the statement, "I have observed or had other direct evidence of this behavior among my professional colleagues including postdoctoral associates, within the last three years."

Please read the article for methodology, limitations, etc.

Stealing ideas is difficult because you have a victim you have stolen from and presumably they know (or will know) that you stole from them.

Fraud is much more common, much easier to do, and much harder to prove—unless you do something really stupid like re-use the same image multiple times in various unrelated papers, something the people who get caught always seem to do. (Note 1)

• Jan Hendrik Schön - Physics
• Haruko Obokata - Biomedical sciences

Also common, as Yiuin states, is people (notably PIs and supervisors) taking the credit for their underlings'/minions' work.

Note 1: This means that the either all the frauds are stupid and re-use images and get caught, or that the frauds who don't re-use images are rarely caught.

Not quite an answer, but too long for a comment. In order to quantify how common academic theft is, one needs to define theft. You attempt to define it as:

By academic theft, I am not talking about plagiarism, but rather about stealing research ideas before anything is published

Now consider the following scenario. Alice has been carrying out research on topic X on and off for years. She has a number of nice research findings that she hasn't gotten around to publishing and hasn't shared the findings with anyone. Alice finally decides to start focusing on topic X and publish her existing results. Unbeknownst to Alice, Bob has just started working on topic X as she begin trying to publish her results. I don't think anyone would argue that Alice has engaged in academic theft. The issue becomes what would have to change for Alice to have engaged in academic theft.

What if Alice found out about Bob's intentions from Carol (or maybe Bob himself) and that changed Alice's research direction?

What if Alice only starts doing the research after she hears about what Bob is doing?

What if Bob presents a novel approach to topic X and Alice runs with the approach faster/further than Bob, but Alice is careful to always credit Bob with the new approach? What if Bob presented the new approach N years ago (choose your N)?

In order to quantify how often academic theft occurs, one needs to define what theft is.