Fourth of July data/statistics link roundup (7/4/2013)

1. An interesting post about how lots of people start out in STEM majors but eventually bail because they are too hard. They recommend either: (1) we better prepare high school students or (2) we make STEM majors easier. I like the idea of making STEM majors more interactive and self-paced. There is a bigger issue here of weed-out classes and barrier classes that deserves a longer discussion (via Alex N.)
2. This is an incredibly interesting FDA proposal to share all clinical data. I didn’t know this, but apparently right now all FDA data is proprietary. That is stunning to me, given the openness that we have say in genomic data - where most data are public. This goes beyond even the alltrials idea of reporting all results. I think we need full open disclosure of data and need to think hard about the privacy/consent implications this may have (via Rima I.).
3. This is a pretty cool data science fellowship program for people who want to transition from academia to industry, post PhD. I have no idea if the program is any good, but certainly the concept is a great one. (via Sherri R.)
4. A paper in Nature Methods about data visualization and understanding the levels of uncertainty in data analysis. I love seeing that journals are recognizing the importance of uncertainty in analysis. Sometimes I feel like the “biggies” want perfect answers with no uncertainty - which never happens.

That’s it, just a short set of links today. Enjoy your 4th!