Author: Dr. Jonathan Berman | Category: Research Integrity | May 20, 2016
Doyle can be fairly regarded as a smart guy. He was a physician and wrote over
60 books. He invented the Sherlock Holmes character that has endured to this
day in the public consciousness as a paragon of reason.
Many familiar with
Doyle’s works of fiction may not realize that one of his chief interests was
the spiritualist movement of the day and psychic phenomena.
In 1919-1920, Doyle became involved in the investigation of a pair of photographs that had
begun circulating in the spiritualist community.
These photographs seemed to
show two young women, Frances Griffiths and Elsie Wright playing with fairies.
These images were later admitted to have been hoaxed by tracing drawings onto
cardboard standups, which was cheaper than a copy of photoshop in 1917.
Doyle latched onto the photographs as positive proof supporting his beliefs and in 1920 he commissioned three more photographs from the women (using a provided camera with secretly marked photographic plates), and upon examining the results
published an article in The Strand arguing that the evidence warranted a prima
facie acceptance of the photographs until proved otherwise (a reverse of
the normal scientific principle of placing the burden of proof on the
As easy as
it is for us to mock and criticize Doyle today, and he received his share of
criticism at the time as well, there are parallels between the mistake the
Doyle made and scientists who make the same mistakes to this day. Doyle was
looking for evidence to confirm his beliefs, rather than evidence to disconfirm
The cases of
bad science that make the news are the rarest and most extreme examples
(otherwise they would not make the news). It isn’t data falsification we have
to worry about so much as the subtle ways that biases creep into our data and
interpretation and fool us in the same way that Doyle was fooled.
In 2015, the Reproducibility Project published replications of over 100 psychological studies and found only
39 percent to be replicable. The problem isn’t isolated to psychology. Although the
cost of replication in biomedical sciences is much higher, earlier studies have
shown similar rates of low reproducibility.
problems are deeper than simple experimenter bias. The majority of biological
publications either misuse or improperly interpret statistics.The National Institutes of Health (NIH) continues to fund a center devoted to throwing money at medical
treatments that don’t meet the scientific standards for evidence.
these are genuine mistakes, but the prevalence of journals that will not report
negative or confirmatory results, or accept results only on the basis of an
arbitrary p-value cutoff suggest a systemic unwillingness to look at the
systemic problems in science.
In the early 20th
century, Doyle maintained a friendship with the skeptic Harry Houdini (also
noted as a magician). Houdini traveled the world exposing the tricks used by
spirit mediums. Eventually, the séances that were in vogue in the Victorian
parlor withered in the light of skeptical investigation and retreated to the
status of fringe diversion.
are driven by all of the same fears, worries, and hidden vanities that drive
all human behavior. We are asked to so often to justify our desire to participate
in science that we internalize a mantra about the love of discovery, puzzle
solving, and human health.
Perhaps we would be better off to internalize a
reflection on our personal motivations, and how they influence our results. If
some day your friend tells you that data interpretation is akin to believing in
fairies, don’t double down on your position.
At least hear her out because
mostly Houdini was right.
Dr. Jonathan Berman is a postdoctoral fellow at the University of Texas Health Science Center San Antonio in the laboratory of James Stockand and in the Department of Physiology. Dr. Berman's research has focused on the molecular origins of hypertension, and will expand to the physiology whole body magnesium homeostasis and associated diseases.
Copyright © 2018 The University of Texas Health Science Center at San Antonio
Links provided from the UTHSCSA pages to other websites do not constitute or imply an endorsement of those sites, their content, or products and services associated with those sites.