Probability of success licensed vs. in-house projects

The results are in Table 4. As we did earlier, we show both the results for the US companies running US trials, and those obtained using the full sample. The two samples show no qualitative difference in the results. Also, to avoid any potential bias in the data, the sample used in this and the next sections does not include the compounds developed by the universities (but it still includes compounds originated by imiversities and licensed to companies). This reduces our full sample fi-om 2078 to 2036 projects. 

(Success Compound licensed in Clinical I) Prob.(Success Internally developed compounds) 

The Table also shows that the difference in the probability of success between licensed and internally developed compounds is quite pronounced. While one cannot exclude that our sample is biased in favor of the collaborative projects, this evidence would be consistent with the view that licensed 

---