Is there a reproducibility crisis in science? - Matt Anticole
-
0:07 - 0:11In 2011, a team of physicists reported
a startling discovery: -
0:11 - 0:14neutrinos traveled faster
than the speed of light -
0:14 - 0:16by 60 billionths of a second
-
0:16 - 0:21in their 730 kilometer trip from Geneva
to a detector in Italy. -
0:21 - 0:25Despite six months of double checking,
the bizarre discovery refused to yield. -
0:25 - 0:28But rather than celebrating
a physics revolution, -
0:28 - 0:30the researchers published a cautious paper
-
0:30 - 0:35arguing for continued research in an
effort to explain the observed anomaly. -
0:35 - 0:41In time, the error was tracked to a single
incorrectly connected fiber optic cable. -
0:41 - 0:46This example reminds us that real
science is more than static textbooks. -
0:46 - 0:50Instead, researchers around the world
are continuously publishing -
0:50 - 0:51their latest discoveries
-
0:51 - 0:55with each paper adding
to the scientific conversation. -
0:55 - 0:57Published studies
can motivate future research, -
0:57 - 0:59inspire new products,
-
0:59 - 1:01and inform government policy.
-
1:01 - 1:05So it's important that we have confidence
in the published results. -
1:05 - 1:07If their conclusions are wrong,
-
1:07 - 1:08we risk time,
-
1:08 - 1:09resources,
-
1:09 - 1:12and even our health in the pursuit
of false leads. -
1:12 - 1:14When findings are significant,
-
1:14 - 1:17they are frequently double-checked
by other researchers, -
1:17 - 1:19either by reanalyzing the data
-
1:19 - 1:22or by redoing the entire experiment.
-
1:22 - 1:25For example, it took repeated
investigation of the CERN data -
1:25 - 1:29before the timing error was tracked down.
-
1:29 - 1:33Unfortunately, there are currently neither
the resources nor professional incentives -
1:33 - 1:39to double check the more than 1 million
scientific papers published annually. -
1:39 - 1:43Even when papers are challenged,
the results are not reassuring. -
1:43 - 1:46Recent studies that examined dozens
of published pharmaceutical papers -
1:46 - 1:51managed to replicate the results of
less than 25% of them. -
1:51 - 1:55And similar results have been found
in other scientific disciplines. -
1:55 - 1:58There are a variety of sources
for irreproducible results. -
1:58 - 2:04Errors could hide in their original
design, execution, or analysis of the data. -
2:04 - 2:05Unknown factors,
-
2:05 - 2:08such as patients' undisclosed condition
in a medical study, -
2:08 - 2:12can produce results that are
not repeatable in new test subjects. -
2:12 - 2:16And sometimes, the second research group
can't reproduce the original results -
2:16 - 2:20simply because they don't know
exactly what the original group did. -
2:20 - 2:24However, some problems might stem
from systematic decisions -
2:24 - 2:26in how we do science.
-
2:26 - 2:27Researchers,
-
2:27 - 2:29the institutions that employ them,
-
2:29 - 2:31and the scientific journals
that publish findings -
2:31 - 2:35are expected to produce
big results frequently. -
2:35 - 2:37Important papers can advance careers,
-
2:37 - 2:39generate media interest,
-
2:39 - 2:41and secure essential funding,
-
2:41 - 2:45so there's slim motivation for researchers
to challenge their own exciting results. -
2:45 - 2:48In addition, little incentive exists
-
2:48 - 2:52to publish results unsupportive
of the expected hypothesis. -
2:52 - 2:55That results in a deluge of agreement
between what was expected -
2:55 - 2:57and what was found.
-
2:57 - 3:00In rare occasions, this can even lead
to deliberate fabrication, -
3:00 - 3:05such as in 2013, when a researcher
spiked rabbit blood with human blood -
3:05 - 3:09to give false evidence that
his HIV vaccine was working. -
3:09 - 3:11The publish or perish mindset
-
3:11 - 3:15can also compromise academic journals'
traditional peer-review processes -
3:15 - 3:17which are safety checks
-
3:17 - 3:20where experts examine submitted papers
for potential shortcomings. -
3:20 - 3:22The current system,
-
3:22 - 3:24which might involve only one
or two reviewers, -
3:24 - 3:26can be woefully ineffective.
-
3:26 - 3:29That was demonstrated in a 1998 study
-
3:29 - 3:33where eight weaknesses were deliberately
inserted into papers, -
3:33 - 3:36but only around 25%
were caught upon review. -
3:36 - 3:41Many scientists are working toward
improving reproducibility in their fields. -
3:41 - 3:43There's a push to make researchers
raw data, -
3:43 - 3:45experimental procedures,
-
3:45 - 3:48and analytical techniques
more openly available -
3:48 - 3:51in order to ease replication efforts.
-
3:51 - 3:53The peer review process can also
be strengthened -
3:53 - 3:57to more efficiently weed out weak papers
prior to publication. -
3:57 - 4:00And we could temper the pressure
to find big results -
4:00 - 4:04by publishing more papers that fail
to confirm the original hypothesis, -
4:04 - 4:09an event that happens far more than
current scientific literature suggests. -
4:09 - 4:12Science always has, and always will,
encounter some false starts -
4:12 - 4:15as part of the collective acquisition
of new knowledge. -
4:15 - 4:18Finding ways to improve
the reproducibility of our results -
4:18 - 4:22can help us weed out those false starts
more effectively, -
4:22 - 4:25keeping us moving steadily toward
exciting new discoveries.
- Title:
- Is there a reproducibility crisis in science? - Matt Anticole
- Description:
-
View full lesson: http://ed.ted.com/lessons/is-there-a-reproducibility-crisis-in-science-matt-anticole
Published scientific studies can motivate research, inspire products, and inform policy. However, recent studies that examined dozens of published pharmaceutical papers managed to replicate the results of less than 25% of them — and similar results have been found in other scientific disciplines. How do we combat this crisis of scientific irreproducibility? Matt Anticole investigates.
Lesson by Matt Anticole, animation by Brett Underhill.
- Video Language:
- English
- Team:
- closed TED
- Project:
- TED-Ed
- Duration:
- 04:47
Jessica Ruby approved English subtitles for 1608 04 A Anticole Matthew ScientificIrreproducibility (correct version) | ||
Jessica Ruby accepted English subtitles for 1608 04 A Anticole Matthew ScientificIrreproducibility (correct version) | ||
Jessica Ruby edited English subtitles for 1608 04 A Anticole Matthew ScientificIrreproducibility (correct version) | ||
Jennifer Cody edited English subtitles for 1608 04 A Anticole Matthew ScientificIrreproducibility (correct version) |