Read on Twitter
Sunday advice for the budding/ junior medical researcher:
Before you get too wedded to your project, consider What if everything my PI has told me is *wrong?
[thread]
*not original word choice
Before you get too wedded to your project, consider What if everything my PI has told me is *wrong?
[thread]
*not original word choice
Each week, there is yet another paper that follows the contours of a popular medical narrative that has serious deficiencies
These papers don't do too well on this forum.
Occasionally, authors are surprised at the criticism.
And for good reason!...
These papers don't do too well on this forum.
Occasionally, authors are surprised at the criticism.
And for good reason!...
Until the point the paper was launched in the world, they had been living in the like-minded bubble.
The PI was like-minded
The collaborators were like-minded.
The editor was like-minded
Even the reviewers...
The PI was like-minded
The collaborators were like-minded.
The editor was like-minded
Even the reviewers...
Could be divided into those who criticized and rejected the paper, and were wrong and obviously jealous, and the final 3 enlightened reviewers who swooned, saw the light, and accepted the paper. Yay!
then it went on twitter, and got evicerated...
then it went on twitter, and got evicerated...
The reason this happens is that often times PIs are wrong, and the junior person did not entertain that possibility.
What can you do to avoid it?
What can you do to avoid it?
Step 1: PI wants to do project X
Go home and read a lot of things about the topic. Read what the field thinks. Read what folks who aren't your PI think. Do a deep background reading.
If a project takes 100 hours, 10 should be pre-reading
Consider what if PI is wrong?
Go home and read a lot of things about the topic. Read what the field thinks. Read what folks who aren't your PI think. Do a deep background reading.
If a project takes 100 hours, 10 should be pre-reading
Consider what if PI is wrong?
Step 2: Ask PI the unresolved questions-- why are we doing this? How would you interpret this result, or that result?
They have to convince you first!
They have to convince you first!
Step 3: Think for yourself. Does the method actually answer the question the PI has set out? What biases might the PI be missing? What might someone with methods training say?
Step 4: If you are not satisfied, make a polite excuse and move on.
One common class of medical questions that are often flawed are retrospective analyses that try to answer causal questions
Your PI might have a good Q, but probably has no idea of how to answer it.
Your PI might have a good Q, but probably has no idea of how to answer it.
Another common error is doing work in a 'hot field' where the result is meant to preach to the choir.
Biomarker matched cancer drugs are better than non-biomarker drugs. -- that's a sexy hypothesis that a lot of folks believe in, but can you rigorously interrogate it?
Biomarker matched cancer drugs are better than non-biomarker drugs. -- that's a sexy hypothesis that a lot of folks believe in, but can you rigorously interrogate it?
Another common error is the virtue error. The result of your paper is a 'virtuous' result. It reaches a conclusion that is aligned with our sense of right or wrong.
ITs commendable to help the underdog!
But be careful you are not the victim of wishful thinking.
ITs commendable to help the underdog!
But be careful you are not the victim of wishful thinking.
A virtuous paper can still suffer from serious methodologic problems. You can be right for the wrong reason. Or wrong, but still right-- you know what I mean.
Just a few thoughts on how to get the most of your research experience. If you are specifically curious about cancer medicine, I would direct you to section 3 of Malignant https://www.amazon.com/Malignant-Policy-Evidence-People-Cancer/dp/1421437635/
