Was Herodotus Correct?
William B. Hansen
Published online: 3 May 2011
Society for Prevention Research 2011
Ideas about replication have been with us for a long time.
On the one hand, the ability to reproduce findings is an
essential tenet of science. The same methods should
produce similar results. On the other hand, we are faced
with the conundrum first suggested by Herodotus who is
credited with the thought that replication may not be
possible. He framed his idea as a truism: One cannot step
into the same stream twice. His logic was that both the
stream and the person stepping into it were ever changing.
Hence, replication is not possible.
So, on the one hand, it is impossible for a trained researcher
to ever think replication of anything but a worthwhile pursuit
because our science cannot progress and establish its scientific
veracity without it. But, on the other hand, our expectations
should be low because the conditions that are required to
fulfill the mandate for replication cannot be satisfied.
Being devoted to advancing the science of prevention, I
want replications to succeed. However, I am faced with the
inevitable truth learned from my own and others’
experience that, with the possible exception of identifying
approaches that do not work, the replication of findings is
elusive in the best of cases. Beyond the trivial, if
replication is to have meaning, it must be informative
and promote progress in the field.
I’ve been invited to offer a commentary on the article
published in this volume titled “Replication in Prevention
Science.” I cannot decry the intent of the paper–to foster an
increased dedication to pursuing replication and to suggest
ways in which replication should be pursued. However, all
these are worthy of an anti-thesis. My intent is to provide
one. I take as my point of departure my interpretation of
what was implied by Herodotus.
Can there be replication in prevention science? Prevention
research has been funded and pursued systematically for over
35 years. Narrowly considered, prevention research focuses
on the ability of public health, psychosocial, environmental
and ecological interventions to deter the onset of undesirable
behaviors and the occurrence of negative outcomes. As such,
a primary focus of research has been on testing the efficacy
and effectiveness of interventions. More broadly considered,
the field must include research on epidemiology and etiology
that inform intervention design.
There is ample evidence from the latter two areas of
research that suggest epidemiologic and etiologic findings
can be more or less replicated. For example, Monitoring the
Future (Johnston, Bachman, O’Malley & Schulenberg,
2009), which has been able to replicate its methods
annually since 1975 has consistently shown that as
adolescents grow older an increasing proportion use drugs.
However, there is annual variation in the prevalence of use
of all drugs and year-to-year changes in drug use trends, the
causes of which are not fully understood. Broadly defined,
the methods produce replicable findings. Narrowly defined,
there are many changes afoot from year to year; even the
brightest among us cannot project next year’s outcomes or
when trends will reverse course.
The irony with pursuing replication is that the greatest
body of evidence reflects what we know about programs
that do not work. For example, numerous studies
demonstrated that the original DARE intervention failed
to have a meaningful preventive impact (Ennett, Tobler,
Ringwalt, & Flewelling, 1994). If a replication of non-
findings is the best our field can do, there is low value to
replication. My guess is that it will not be possible to find
W. B. Hansen (*)
420-A Gallimore Dairy Road,
Greensboro, NC 27409, USA
Prev Sci (2011) 12:118–120