namelesstavern.net

Apparently, quite a bit.

http://chronicle.com/daily/2006/04/2006040704n.htm

[quote]Raymond G. De Vries, an associate professor of medical education at the University of Michigan at Ann Arbor, and three colleagues last year reported surveying more than 3,000 scientists about whether they had ever engaged in misbehavior, such as changing a study because of pressure from a source of funds, or failing to present data that contradict one’s own research. One-third of the scientists acknowledged they had committed some form of research misbehavior.

“Why are people engaging in these behaviors that they feel uncomfortable about?â€

So being scientiffically proven has an actual 1/3 chance of being wrong in some fashion?

Maybe this survey part of that 1/3!

to compound this (and I think anyone who's done a lot of database work will be able to verify) it's really easy to make 1 bad assumption, and base a lot of work of that assumption


easy example is to pull numbers from a database in a query of some sort
set 1 bad parameter, or simply overlook 1, and you can have all your data tainted, and no realize it

I think the survey was about scientists knowingly tainting the findings or parameters to get the outcome they wanted or were trying to prove.

yes, and as a related tangent, the amount of conclusions based on faulty non falsified data is likely as high

or higher

How many people can really look at their work objectively and not add their personal biases unintentionally?

All the science you ever needed is in the bible. Everyone knows you can't fuck with that, since it's all true.