> Stanford University President resigns over fake data discovered by a student:
> Another fake data by Harvard prof.
A handy trick is to take ones measurements, and they can be measurements of anything at all, and plot them in a distribution.
The universe as a whole tends almost invariably toward "normal distributions" (bell curves). They are so common, we even call them "Normal Distributions".
There's also "Benford's Law" and other similar tools that allow the digits in numbers of individual data points to be checked for abnormal distributions, as people making up numbers (or using random-number generators) are not as good as the real world at creating statistically-verifiable "honest real-world data". The upshot of all this is that cheaters and frauds will never release their data, as it will be a billboard announcing "I Am A Big Fat Fraud!" to anyone with a laptop and a spare 10 minutes to check the data.
Why?
It's almost as basic as the speed of light or Planck's constant - the "Central Limit Theorem" is one of those laws of the universe you can't break no matter how hard you try:
When the number of data points is large (usually 40 to 50 is enough for it to work) the mean X follows a normal distribution, no matter what the distribution of underlying population is.
So, to summarize, math is our friend, and math protects us from liars, cheats, and frauds.
But math scared so many people (due to crappy math teachers), people don't "like" math, when it is honestly the only way to understand anything more complex than a toaster.
And THIS is why all legit papers now link to online "supplemental material" that include downloadable CSV-formatted datasets, to provide firm proof that there was no hanky-panky, cherry-picking, or pulling of numbers out of someone's ear.
And anyone who WON'T immediately supply raw data to anyone who asks? Well, we have a technical and highly scientific term for that sort of fellow.
He's a "Liar".
***********************************************
The BEE-L mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software. For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html
|