MCH said:
I'd like to see some evidence, or reasoning, that it will work.
It's a standard technique used in biology. Well, more accurately, it's the underlying premise for several such techniques. I'm sorry, but if you aren't familiar with these statistical methods I can't take your claims about the validity of anything seriously. It's basic, freshman stuff. Look up tag and release sampling sometime; it's an example of the application of these principles.
Oh I have methods for detecting fraud in research if not in publications; they are the methods used everywhere else for detecting any kind of mismatch, by mistake or deliberate.
So no, you in fact DO NOT have any method for determining undetected frauds in scientific publications. We can therefore dismiss the notion that you have any better sense of this data than has been presented in this thread.
This means that you have nothing upon which to base your arguments.
Thank you for refuting yourself.
Let's go on to: what can we use to estimate likely fraud rates?
I have answered that repeatedly. You have ignored it repeatedly--and you have demonstrated that you're ignorant of standard methods for estimating this sort of unknown data. You really should think about that a bit.
Datasets on fraud rates in business is a datasets available for science
No, they're not. They're datasets
for business. By definition. If you play this fast and loose with terminology, prepare to have your accedemic career cut very short.
When I say you need data for science, I mean that if you wish to discuss rates of fraud in scientific publications you are obliged to provide data on rates of fraud in scientific publications. Providing rates of fraud for other things is irrelevant. It's off topic. We actually have data for rates of fraud in science--not, perhaps, the best data, but sufficient to draw conclusions (again, the sampling methods will help provide estimates for what we're missing). Ignoring that data in favor of datasets less related to the topic at hand is dishonest. If you want to discuss undetected frauds in scientific publications, you must provided date on the rate of undetected frauds in scientific publications.
I'm not sure why this is so difficult: If you wish to discuss X, you need data on X. This is basic stuff.
I'm not dismissing data about detected fraud, I'm simply saying that it's not related to actual fraud.
That's absurd on the face of it. Obviously they are related--detected frauds are a subset of committed frauds. The issue you and I are having is that you presume that undetected rates are much higher than detected rates, and I'm demanding you actually provide some justification for that belief.
And again you're making claims about what's been presented that are not relevant or are not true: no-one's made any claim upstream as to why scientists would be less fraudulant than most people.
This is a lie.
Here's proof.
Perpetual Student said:
Science is a unique human endeavor in that its purpose is to expand knowledge and understanding. By its very nature, its end result is both public and open. Those who dare commit fraud in science risk a high probability of exposure because thousands of scientists throughout the world are motivated to scrutinize new results and expand on new findings. Simply put, other than intellectual curiosity, a major motive behind discovering something new is to have the world acknowledge one’s success in having done so. It’s very difficult for fraud to be successful.
In most other areas, fraud is secretive and not subject to universal scrutiny and the reward is almost always financial. Without any supporting data, it stands to reason that fraud is much less prevalent in the pursuit of science when compared to other professional activities.
Hired hacks of corporations who spew stuff like anti-global warming babble, the safety of tobacco use, the benefits of herbal remedies, etc. are just ordinary frauds pretending to be scientific -- even though they actually may have some credentials. There are incompetent hacks in every field.
So, so far, the idea that research fraud rates will be similar to normal people's scientifically tested fraud rates is likely a closer dataset than detected fraud rates. [snipped parantheticals]
Nothing more than baseless assertions which ignore all available data in favor of supporting a pet hypothesis. The paranetheticals were nothing more than self-serving justifications for ignoring data and facts you don't like.
Bollocks. I mean, really? I'd notice a 5% change in my body temperature.
Is your body temperature metal? If no, than your statement is irrelevant. I was giving a few numbers I happened to know off the top of my head regarding human perception, to illustrate what range "commonplace" and "detectable" should fall into.