Academic integrity and honesty

I have been listening to Richard Feynman’s autobiography on his life experiences.

Just near the end, he ranted on about the necessity of being completely honest about one’s research; present not only the good bits but everything, thus leaving the readers to determine the merits of one’s claim.

To be honest

It is hard to do. There is always the temptation when conducting an experiment to explain away the outliers as statistically insignificant. To stay within one’s own expectation, injecting bias into one’s own result. This is by no means dishonest in the current academic definition. We are not falsifying anything, merely presenting only what strengthens our case.

He described the Millikan phenomenon. He measured the charge in an electron, but got it wrong slightly by using the wrong value for (I think it was) air density. Using the same setup, subsequent researchers of course used the right values, but then got different results. Since they all believed that Millikan is the expert, and therefore must be right, these researchers would always justify ditching some data until the final number is just close enough to but slightly bigger than Millikan’s. Then later researchers would increase the value further until the right one is reached. So instead of an obvious jump to the correct value, it increased gradually.

To try to explain everything, not just what is convenient

Malcolm Gladwell talked about the err in omitting the outliers instead of focusing on them. The reason for doing so is to uncover the reason for such extra-ordinary performance in whatever one is measuring, with the intention to replicate such outlier until it becomes the norm. We are too focused on the mean, the standard deviation, and a nicely shaped Gaussian distribution, and often completely ignore those at the tail ends of such a distribution. It may just happen that these tail ends represent unforeseen opportunities that we routinely brush off.

And not to glorify

I think there is an ego associated with being able to write massive mathematical equations, and being able to publish them. So much that we try to make equations out of English sentences when it is perfectly clear what the sentences mean already. I think this is part of the reason why Feynman was such a good lecturer; he saw more value in understanding than in unnecessary complexity. Though in a way, he may have glorified himself in the book; wrote about himself in a way that made him seem like the least pretentious guy ever.

Not easy principles to live by when publications are good, and frequently cited publications are better…