Science Isn't Broken

Science Isn't Broken

Organisation: FiveThirtyEight (United States)

Publication Date: 04/10/2016

Size of team/newsroom:large


The temptation to push and tweak the underlying numerical values of a scientific experiment until they meet a desired goal – known in the scientific community as “p-hacking” – is sometimes so great that many researchers cannot resist it. In an extensive investigation, science writer Christie Aschwanden wrote that there were so many examples of this kind of cheating that it might appear that science is “broken” – and yet it is not. P-hacking instead shows how difficult science is, and – as the author wrote in an insight that made this article widely read and admired in the scientific and lay communities alike – it might represent a way to inch forward to the truth. The article was accompanied by a remarkable interactive graphic by Ritchie King that helped readers understand the concept of p-values and how they can be changed to produce different results. The interactive let readers adjust the inputs of an invented study testing whether Republicans or Democrats were better for the economy. By changing the inputs -- such as including or excluding recessions -- it was possible to claim that either party had a more positive impact on the economy, with a significance of p < 0.05.

What makes this project innovative? What was its impact?

P-values and p-hacking are dry, hard-to-describe subjects, and it was important to find a way to get readers engaged in the story. This interactive did just that, providing a fun-to-use entry point that perfectly conveyed the point the piece was trying to make. The graphic also urged users on by letting them know when their p-value was “almost” publishable, mimicking the sort of pressures that scientists can run into when their research is close to passing significance tests. The calculator-like, graph paper design kept it light and fun -- this was essentially a mini-game. This project is reflective of FiveThirtyEight’s aim to make difficult statistical subjects approachable to the everyday reader.

Technologies used for this project:

The interactive graphic was built using d3, with background data analysis in R.
Follow this project

Comments (0)

You have to be connected to contribute

You have to be connected to follow

Leave this project and no longer be informed about this project

By joining this project, you will be informed by email when an update or a new contribution is posted on the website.

Thank you for your active participation !

The GEN Community Team