What would you do/investigate with 1 mil dollars?

1

So my supervisor has made a “teamwork” day in which one of the ideas is to think about what we (as a group of Ph.D. students) would do if we had 1 mio dollars for research. In our group, none of us are really statisticians. We work in the field of oncology. But I am, however, the one doing most modelling and stuff, so my idea should somehow be in the realm of modelling.
I think the “assignment” is quite funny, but I have obviously not been in the game of modelling and statistics as long as some of you. Not even close. So I was wondering what you might think would be a cool/fun/interesting thing to investigate from a modelling perspective.
I mean, doing a RCT and testing some kind of metric, with the correct statistical approach, and so on, is maybe a bit boring in this sense. That’s probably what most of you are consulted about anyways. But like, if you had 1 mio dollars to do whatever you wanted (again, within modelling/statistics), what would you do/investigate ?

2

As someone who has come through a similar path, I think the best advice I can give is not to put the cart before the horse. Good research projects (or a series of good projects if you’re allowed) start with good research questions, and role as the analyst/statistician is to help design a study that doesn’t need to be saved with fancy models

I would recommend you grab some recent articles that are relevant in your field and see what people think the important questions are. Maybe it’s an analysis type question that would lend itself to a simulation study or meta-epidemiological work, but more likely its something where the benefit will come from thoughtful design (and that’s probably what your prof wants you to think about). So start with a research question, come up with a design that does the best job at answering that question, and then propose the analysis that is appropriate for that design.

Some other things to think about (ie for budgeting):

  1. Statistical consultation
  2. Data management (i.e. database development, forms, entry, etc)
  3. Information specialist (librarian) to support literature review (ie so you can think about which variables are prognostic of outcome)
  4. Trial staff (if applicable)
  5. feasibility
  6. ethics
1 Like
3

I would look at Shannon Brownlee’s book Overtreated, which will give you some super interesting investigative subjects. For example, I don’t think there is enough literature on Overdiagnosis. Pick a oncology treatment.

1 Like