Allen has provided one more helpful example of replication (even a published 
lack of replication) in psychological science with the Bulevich et al (2006) 
study. What seems to be being confounded here is the scientific evaluation of 
an original finding as opposed to what appears in the popular press. I don't 
think we can count on the popular press to wait for replication. News is called 
news because it is new so the first time a particular finding occurs will be 
when it is news whereas scientists will wait for replication. I do realize that 
individual scientists don't help much with the press releases they issue on 
their findings. However, other scientists, who may not be so credulous on 
scientific topics as the press, are still likely to replicate, if only to build 
on the new finding. If it isn't replicable, there won't be much follow-up on it 
in the scientific literature (even if failures to replicate aren't published). 

So, instead of looking to the press for the importance of replication in 
science, we should look to the number of reference citations eventually given 
to a new work. If a phenomenon can't be replicated, it isn't going to have much 
impact on science (although it may have an impact on the popular understanding 
of science as disseminated by the media).

Rick

Dr. Rick Froman, Chair
Division of Humanities and Social Sciences
John Brown University
Siloam Springs, AR  72761
[email protected]
________________________________________
From: Allen Esterson [[email protected]]
Sent: Sunday, November 06, 2011 4:25 AM
To: Teaching in the Psychological Sciences (TIPS)
Subject: Re:[tips] Stapel's faking of social psychology data

Presumably in response to my writing this:
> My impression is that this is frequently not the case in psychology,
> with results of studies sometimes being widely cited regardless of
> whether they have been replicated.

Jim Clark provided a justified corrective:

>Is replication that uncommon in psychology?  Just a couple of
observations

>1.  the recently cited paper noting that effects get weaker in
>subsequent studies would require replications even to conduct the
analysis

>2.  meta-analysis requires multiple replications of some effect (e.g.,
> gender differences); otherwise again not possible to do a
meta-analysis

>3.  I can think of many, many areas where there are repeated
demonstrations
>of certain effects (e.g., serial position curve, DRM false memory
effect,
>caregiver-infant attachment, ...).

I agree with virtually everything Jim wrote [see below]. I would just
add two comments. I should have made clear that what I had in mind was
the limited field of social psychology, where results of studies
(typically based on questionnaires) are on occasion publicised and
widely accepted without regard to replication. But what I also had in
mind was an occasional experimental claim in psychological science that
is widely publicised as proven regardless of replication. For instance,
the claim by Anderson and Green (2001) to have experimentally validated
Freudian repression mechanisms:
http://www.nature.com/nature/journal/v410/n6826/abs/410366a0.html

A University of Oregon press release (no longer available online)
reported:

"Our findings are consistent with Freud's notions of suppression and
repression, but go a long ways towards demystifying the process," says
the paper's co-author Michael Anderson, an assistant professor of
psychology at the University of Oregon. "Our work allows Freud's idea
to be understood in terms of widely accepted mechanisms of cognitive
control that apply in a broader range of circumstances."

As a result of the press release this paper was widely heralded in the
press as vindication of Freudian repression. Leaving aside the
criticism that the kind of innocuous memory suppression claimed in the
paper bore little relation to Freudian repression, passing virtually
unnoticed was this report of failure to replicate the results:

 Bulevichet al (2006)

Anderson and Green (2001) had subjects learn paired associates and then
selectively suppress responses to some of them. They reported a
decrease in final cued recall for responses that subjects had been
instructed not to think of and explained their data as resulting from
cognitive suppression, a laboratory analog of repression. We report
three experiments designed to replicate the suppression/repression
results. […] None of our experiments showed reliable suppression
effects with either the same or independent-probe tests. Suppression is
apparently not a robust experimental phenomenon in the think/no-think
paradigm.
http://www.psych.wustl.edu/coglab/publications/bulevich2006.pdf

Allen Esterson
Former lecturer, Science Department
 SouthwarkCollege, London
[email protected]
http://www.esterson.org

@fsulist.frostburg.edu
---
You are currently subscribed to tips as: [email protected].
To unsubscribe click here: 
http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5&n=T&l=tips&o=13940
or send a blank email to 
leave-13940-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu

Reply via email to