Confessions of a science cheat!

#1
Sciborg_S_Patel supplied this link in another thread:

http://nick.brown.free.fr/stapel/FakingScience-20141214.pdf

This is an account by Diederik Stapel of how he ended up faking his psychology research. I think it is worth reading by everyone here who is interested in the honesty of science. So far, I am part way through reading this book, and here is my comment so far, copied from the critiques-of-sciance-as-currently-practiced thread.
One of the most popular and successful research areas in modern social psychology is
called Terror Management Theory. This suggests that the single biggest factor affecting
people’s behavior is the fear of death. People use religion, culture, positive self-image,
stereotypes, preconceptions, and imagined enemies to keep this fundamental dread in
check. According to Terror Management Theory, the conflict between the human instinct
to stay alive for as long as possible and the conscious knowledge that life is finite leads to a
continuous confrontation with the overwhelming fear of dying. We’re all going to die, and
that’s a scary prospect. Although we’re not always conscious of this fear, it’s always there,
ticking away in the background of our lives. People have developed a large number of ways
to reduce the effect of this terror. Because the human fear of death is so powerful and
dominant, simply thinking about the idea of death is sufficient to cause the defense
mechanisms to swing into action and reduce the anxiety. If you ask people to write a few
sentences about their own death, as if by magic, they become more religious, display more
preconceptions, more emotions, more pride in themselves (“I’m a fantastic person”) and
their culture (“U-S-A! We’re number one!”), and more aggressive towards others (“You’re
just a loser”) and other cultures (“Screw those A-rabs”).
71
In the last few decades, a considerable number of elegant experiments have
demonstrated the effects predicted by Terror Management Theory. There have been a few
negative results, but only a few, because it’s one of the most robust theories in social
psychology. And yet, thinking about death doesn’t always cause you to behave in the way
the theory predicts. Social reality is capricious, and conducting experiments is a difficult
art.
For example, consider this story about a group of Dutch researchers who were
interested in investigating the psychological mechanisms behind the effects of Terror
Management, but couldn’t reproduce even the most basic “what happens when you think
about death” results. They tried everything. They ran their experiments with students,
with “normal” people, in small groups in large lecture halls, on the computer, on paper, but
nothing made a difference. Whatever they did, they couldn’t get “thinking about death” to
produce the desired effects.
In desperation, the researchers made a visit to another university that also did a lot
of work on Terror Management Theory and seemed to have no trouble in reproducing the
basic effects at will. They couldn’t find any noticeable difference between the two labs. The
successful research group did everything in exactly the same way as the team that couldn’t
find any effect. But there was one difference between the two groups: the successful
researchers were all young, liberal men from the grunge generation who looked like they
didn’t enjoy life much, and always wore black jeans and black rock band T-shirts to the lab.
Experimenting is an art. You have to catch the right behavior at the right moment with the
right people. To learn about what makes people tick, to be successful and get the plaudits
that go with that success, you have to put a lot of work into the “theater” of the
psychological experiment. All the props and scenery, all the actors and actresses have to be
brought together just so.
I don't think the author considered this was cheating at all, but consider what this is saying (and there are more passages like that),
you try to prove a theory like 'terror management theory' by manipulating other variables until you get the answer you want! Would you believe any research that was performed that way?

To me, the point is that even supposedly valid research is so close to cheating, that the step over the line into actual cheating (not actually doing the experiments at all, in Stapel's case) feels so small.

Yes, Arouet, pre-registering the experiments would stop this sort of behaviour, but of course you could still pre-register an experiment and then make up (or adjust) the data!

David
 
Top