Re: Program-survival bias / Zebra’s law
>>The answer is so obvious, but maybe not amenable to a simple statistical analysis. I wondered if they checked success rates versus the size to the company.<<
Obvious to iHubbers, but perhaps not so obvious to the academics who conducted the study!:-)!
It is indeed a pity they didn’t check to see if company size was a variable with independent predictive value. After reading your post, I was wondering what the best operational definition of company size would be. I think market cap (or the more precise “enterprise value”) would be a reasonable proxy, but my choice would be a variable that is not about size per se but a composite and size and execution: the number of NDA/BLA applications from the company approved by the FDA or EMEA during the n years preceding the data point in question (pick a value of n).
Market Data 
Markets 
AI
