Thread by @steiny

Thought experiment time. Imagine that you're a funder, and you can only ask a funding applicant three questions. After the third answer, you're going to have to make a funding decision, yes or no. What three questions would you ask?

Delighted to see so many people in the mood for a thought experiment tonight!

It's only just occurred to me that if I was a mad [data] scientist your replies would make for a promising training set for my machine learning model 😀

Here's the thought process that inspired this thread.

In social science research, people talk a lot about the different predictive powers of answers to different questions. Here's a fun article on the predictive power of questions to ask on first dates theblog.okcupid.com/the-best-quest…

I was thinking that if you had a pool of historic applications, and you quality scored every answer to every question, and then you looked for correlations with the final grant decision (yes/no), you might be able to find some questions that are very strong predictors

My thought experiment that started this thread was me asking you what you think are probably the questions with the strongest prediction value, but without me having to do an enormous data gathering and statistics exercise 😄