From the article cited:
"In a survey published June 9 in the journal Nature, about 1.5
percent of 3,247 researchers who responded admitted to
falsification or plagiarism."

These two charges are totally different! Falsification is
contrary to the most basic elements of science, while plagiarism
is merely a human failing.

Gee, I was lucky in my scientific career. No one ever put a
pascal of pressure on me to exceed my abilities, or that of my
experiments or equipment. All I needed was curiosity, a fondness
for economy of explanation, and the belief that if I couldn't
explain anything I was doing even to my youngest child I didn't
really understand it myself.

I like the part that says that according to research by Prof. David
Wright "foreign nationals who learned somewhat different scientific
standards" is one of the "basic reasons" of "why scientists cheat".

Does that imply that Canadian nationals are the culprits and the basic
reason of why scientists cheat? :-) Seriously, I doubt cheating is
tolerated any more lightly in any non-US system that produces research
scientists.

That said, I openly admit that I fake data all the time, except that I
call it "simulations". "Real" data is such a hindrance anyway.

1) It is expensive. Many people won't spit for $20 to give a DNA
sample not to mention give blood to benefit ungodly research they
don't trust.

2) You'd have to be careful not to overstep consent, protect privacy,
keep the lab notebook, ... In short, become a part-time accountant.

3) With real data, you never know "the truth", whereas with samples
taken from a simulated population you're the one who specified all the
parameters. So you know how often what you're finding is what you're
supposed to find and thus determine how well your method works.

4) Any results obtained with a real data set are subject to
statistical sampling whereas with simulations, the distribution of
results can be examined over repeated samples produced under exactly
the same conditions.