The only ethics training I got in grad school was effectively "don't photoshop your data" the end.

It's not just photoshop.

A thread. (1/n)

@AcademicChatter #AcademicChatter
We rarely talk about all the murky ways that data are manipulated, cherrypicked & curated to *support* hypotheses, rather than truly *test* hypotheses. (2/n)
Of course, there are those who photoshop Western Blots, and microscopy images. There are those who blatantly change (or invent) data values.

I'm pretty sure we're all in agreement that this is unethical - that's clear.

But not all matters of research ethics are so clear. (3/n)
Let's talk about microscopy: what about a scenario where someone only images the subset of cells with the desired phenotype that supports their hypothesis?

Clearly, this doesn't reflect the reality of the biology being studied. It's highly unethical and hard to detect. (4/n)
It's nearly impossible for reviewers to detect this kind of misrepresentation, unless they repeat the experiments themselves (and even then, that likely isn't enough to prove that the microscopist cherrypicked data). (5/n)
Conversations about reliable and reproducible science should start before the studies start, and not a single minute later.

We always talk about ethics at the end: when we're preparing a paper. What we really ought to be doing is talking about it from the beginning. (6/n)
Conversations about ethics in research are often treated as 'over there' conversations. Ex: "wow, did you see what person from THAT uni in THAT lab did?"

Research ethics is a RIGHT HERE conversation.

We're doing the science RIGHT HERE & we're prone to the same errors. (7/n)
Research ethics isn't a once-a-year-at-the-required-seminar type of conversation, either.

It's an ongoing and evolving conversation that should permeate all of your discussions of your data: in lab meeting, at the bench, on your e-mail, with your labmates etc. (8/n)
There's a lot of focus on those who commit blatant research fraud & I'm not here to make any excuses for that behaviour.

But our problem is much bigger than these individuals.

(9/n)
What I'm saying is: the key to research reproducibility is more than not photoshopping your data. It's a deep and evolving understanding of research ethics as it pertains to your work and each particular technique/experiment/analysis you use. (10/n)
None of us walk into the lab knowing how to execute the most ethical experiment using every single available technique/analysis.

Ethical behaviours, like unethical ones, are learned. (11/n)
Whenever possible, have an open dialogue about research ethics. It's not a taboo conversation, it's a required one.(12/n)
You can follow @thoughtsofaphd.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: