Thoughts on 358 Article "Science Isn't Broken"


The article first introduces the concept of p-hacking - researchers could tweak variables in the study or other things to reach the convention 0.05 significant level, in order to get their research published. Then she explains that scientists mildly tweak p-value often because of personal biased or some other incentives, and only a very small portion researchers are actually fabricating data and hack the peer review system.

Then the author introduces some positive changes in the science self-correction mechanisms - an online blog Retraction Watch that audits other researches, and i other good internet research platform provides frameworks and support for good and open science.

At the end the author has a small reflection that could be summarized as: science is not broken, even though it is probably much messier and harder than we think it is.


As more and more people are joining the academia, or doing research(somehow positively correlated to the increasing undergraduates), I think the academia is going to be messier. The proportion of people who are “cheating the system” will probably increase, even under the control of the growing self-correction mechanism.

More different personalities, interests and background adds diversity but also other things. One of my friend could do well on a True and False question by simply marking all every extra topic mentioned in the lecture “True” in the midterms and test. He is no doubt a smart person and motivated to get a phd seems for higher paying jobs, but I am not sure if his interest is geared enough toward scientific advancement. I think I may be over-worrying since similar problem is probably been in the academia for a long time(thinking about the human-side drama in the old days: controversy between Newton and Leibniz, and Ronald Risher and the Lung Cancer controversy)

But I do advocate and hope universities and colleges could update their coursework at a more adaptive rate(a better self-correction mechanism on Higher Education). Because eventually some undergraduates are going to become researchers. For instance, despite hearing many voices criticizing the misuse of p-values(such as the ASA’s Statement of p-Values: Context, Process, and Purpose and Nate Silver’s Signal and the Noise), I still see courses related to p-value testing is still being taught with almost the same materials. At least I am expecting to see some professors would put sufficient focus warning students the limitations of the p-values. With good chance, I think this is happening but I am just ignorant about it.

Table of Contents