Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

In my college statistics class, we learned MatLab and R using data from the Challenger to recreate why the engineers thought the o-ring might fail and raised the warning. I don't remember the specifics of all the stats, but it was fascinating and the data is publicly available.

Here's a blog that shows some of the analysis (from google search): https://byuistats.github.io/Statistics-Notebook/Analyses/Log...



I wonder what kind of effect that analysis would have on the launch decision. Here's how I would expect it to go:

Engineer: The logistic regression model shows there is a 99.96% chance of failure at 31deg

Manager: How good is the model?

Engineer: p=0.1157

Manager: So it's not statistically significant by our standards. In other words, it's not a very good model.

Engineer: Yeah, but still.

Manager: Well, what does the model say the temperature needs to be before we're more likely to have a successful launch than not?

Engineer: 64.8 degrees.

Manager: I can't defend pushing the schedule to the right that far based on a statistically insignificant model.

(It doesn't mean the manager was right, but I think it highlights hindsight bias. The model only seems good in hindsight because it aligns with the outcome we witnessed and not because it's a really good model. The managers at the time didn't have the benefit of hindsight.)




Consider applying for YC's Fall 2026 batch! Applications are open till July 27.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: