Scientific confidence and integrity

Posted by Shelby Lake, September 26, 2018

Do you remember when the Sun revolved around the Earth? Probably not, because it never did. But until Copernicus published his evidence-based theory that the Sun is at the center of the solar system (called heliocentrism), most people believed Ptolemy’s theory that the planets and the Sun revolved around the Earth (geocentrism). In the centuries after Copernicus announced his theory, Kepler, Galileo, and other astronomers confirmed and expanded this new theory of heliocentrism.

Ptolemy’s theory may not have been accurate, but that doesn’t necessarily mean the he (and his predecessors) were doing bad science. Ptolemy was a prominent mathematician, astronomer, and geographer whose work influenced centuries of scientific and mathematic thought. The fact that his theory of geocentrism was incorrect is in large part due to the knowledge and tools available to him at the time. Ptolemy proposed his theory based on the best available science. However, as astronomers gathered more data, made more observations, created new tools, and tested new methods, they had to revise their understanding of the solar system to account for new evidence.

Revising our scientific understanding of the world is not an easy or clear-cut process. Right now, science is going through a “reproducibility crisis” in which scholars have found that the findings some scientific studies are difficult or impossible to replicate. There are many reasons for this, including systematic bias in experimental design, statistical analyses, limitations of our current tools and technologies, and, in some cases, issues of integrity.

Recently, the Social Sciences Replication Project attempted to replicate the results of 21 social science studies published in Nature and Science between 2010 and 2015. Only 13 of these replication attempts succeeded, and even in those cases the observed effect was smaller than the original study. One of our annotated papers, A point to reading, was among those replication attempts that did not reproduce the original results.

So what does this mean for the eight studies that couldn’t be replicated? Replication attempts aren’t about proving other scientists wrong, they’re a way for us to figure out how to make science better. What scientists learn from replication studies can increase the reliability and accuracy of our experimental methods in the future. Over time, this means that the way we do science gets better, and that our scientific knowledge gets closer to the truth. This also means that even studies that weren’t replicated help inform our understanding of the world and are an important contribution to their respective fields.

The evolution of our current scientific understanding of the world has been driven by centuries of human collaboration, creativity, accomplishments, and mistakes. Scientific progress depends on the ability to reflect on our scientific practices, learn from our mistakes, and generate new ideas. Ultimately, we must be open to revising our knowledge and understanding of the world in light of new evidence, just like Copernicus did.

If you’re interested in learning more about how science operates in the context of society, check out our new collection: Scientific confidence.