If you were a nefarious research scientist, what's a good way to secretly skew the data?

Say you don't like the results of a research survey... what are some creative ways that you could skew or alter the results to be the way you WANT them to be, but that still look official and legitimate if your research is ever investigated.





And no, I'm not a research scientist, but I'm writing a story and want to include this idea of deliberately mangled research data.





thanks.

If you were a nefarious research scientist, what's a good way to secretly skew the data?
There are examples of this. There was a guy at Bell Labs who was using noise from old data and calling it new data. Someone noticed.





Read about John H Schön in Wiki: http://en.wikipedia.org/wiki/Jan_Hendrik...





The techniques he used are not well described in Wiki, but if you hunt around the web, you might find something.





The real problem with falsifying data is people won't be able to duplicate the work so they will start looking hard at the publications to see what is going on and will eventually find problems with them or conclude the initial results are incorrect somehow. If people aren't having problems while trying to duplicate and build on the work, then the work is irrelevant anyway.





This guy had his doctorate revoked by his Uni and can't apply for grants from the German equivalent of the US National Science Foundation.





The reason falsifying data is such a big deal in science is because people will waste resources and time trying to build on the results. This is unconscionable in chronically underfunded fields. From the wiki article





"Also before the allegations became public, several research groups tried - without success - to reproduce most of the groundbreaking results in the field of the physics of organic molecular materials."
Reply:The most obvious way is to just write the readings/results you want. "Pretend" that those were your actual results. If you write down the procedures the correct way, then if anybody ever investigates, they'll see you did it right. Unless your made up data is way out of line (like saying it took 100 mL of 1 M NaOH to neutralize 10 mL 1 M HCl - at equal molarity, M, it will take equal amounts of a strong acid and strong base to neutralize each other), no one will no the difference. They'll believe the data.





If someone comes along behind the nefarious scientist and tries to duplicate the testing and gets different results (as s/he will), the bad scientist can play like s/he must have made a mistake somewhere...but s/he can't find where...blah blah blah...





Depending on the research type, there may be other ways, too. Like if it is supposed to be a random sampling and the scientist specifically picks things that will give him/her the results s/he wants.
Reply:Change the font, it works, it always does
Reply:The key is to find a result in social sciences, were the burden of proof is much less than say physics or biology.





Some fields are so unprovable that you could collect data, and then ask a question to which the data would fit, with 95% confidence. I believe that there is a correlation between IQ and shoe size.