Replicating Scientific Studies is Really, Really Hard

By ivan wanjiku / 1485351171
(Photo : Wikimedia commons) Replicating scientific findings is more like trying to play a complicated board game without all the instructions or even all the parts

Replication is a powerful tool for researchers to check science and get humanity closer to the facts. Scientists pick an experiment that has been completed and test whether its findings hold up by reproducing it.

The consensus is that if the results are the same as the original, then the previous results were correct and can be relied on.

If the results do not match up, then the first finding must be flawed or false.

But now, scientists have found that the replication of studies does not work like that. As scientists reproduce more experiments, they are learning that it is not always possible to get clear answers on the reliability of the previous findings.

Taking the latest conclusions from the Reproducibility Project: Cancer Biology, where researchers focused on reproducing scientific experiments from the highest-impact papers on cancer biology published from 2010 to 2012.

Not even one of the replications by the scientists fully confirmed the original findings.

To pull off the replication of these cancer biology studies, the scientists were forced to contact the authors ask them to share what they may have omitted from their method sections.

The authors were also asked to share any unpublished data that may have helped them arrive at their conclusions.

The researchers then came up with a plan to reproduce the scientific studies and got their papers reviewed by the original authors, statisticians and experts in the relevant field.

"Sometimes we would learn that we were doing the wrong experiment, and we'd go back and make the changes," said Tim Errington, who led the cancer reproducibility project.

 It turns out that replicating the results of scientific studies can be extremely difficult.

For the replication of scientific studies to be easier, researchers are obliged to be totally transparent about their process.