3 ms·

If money weren't a concern, I'd definitely want to work on improving science, particularly social science. Right now, there seems to be a lot of medium-hanging

If money weren't a concern, I'd definitely want to work on improving science, particularly social science. Right now, there seems to be a lot of medium-hanging fruit for systematic improvements that don't get built, because there's not much incentive to do so within academia. OSF has done some great work here, and I'd want to start something similar .

workthrowaway27 9y ago
Over the holidays I met my brother's GF's granddad, who is an economist. He said one of the biggest problems he sees in the social sciences is that the tools they have for data analysis (he specifically mentioned Stata, SAS, and SPSS) make it hard to keep track of what transforms you've made to the data, so people get results and then can't reproduce them because they don't remember what sequence of operations they've performed. Related to this, he said most social science people aren't good programmers which hurts their ability to understand their data. So I think you're right, although I suspect it'll be hard to get people to switch to anything new since they already know how to use tools that let them get their work done.
- sanderjd 9y ago
  This sounds like a generational thing. I suspect there are up-and-comers who are hungry to be better programmers, and are interested in the abundant better tools. So maybe the problem to solve isn't so much getting people to switch tools as aiding education and discovery for people who don't yet know what tools to use. Edit to add: I know a couple folks with backgrounds in statistics. One of them taught a data visualization class using R during his postdoc and is now teaching data science in Python for a code school. The other works for a big established company that does lots of data analysis, and still uses one of the old-school tools you listed, but is always very interested when we chat about the newer tools that are out there. So I think it's a mistake to think most folks are stuck in their ways.

workthrowaway27 9y ago

Over the holidays I met my brother's GF's granddad, who is an economist. He said one of the biggest problems he sees in the social sciences is that the tools they have for data analysis (he specifically mentioned Stata, SAS, and SPSS) make it hard to keep track of what transforms you've made to the data, so people get results and then can't reproduce them because they don't remember what sequence of operations they've performed. Related to this, he said most social science people aren't good programmers which hurts their ability to understand their data. So I think you're right, although I suspect it'll be hard to get people to switch to anything new since they already know how to use tools that let them get their work done.

sanderjd 9y ago
This sounds like a generational thing. I suspect there are up-and-comers who are hungry to be better programmers, and are interested in the abundant better tools. So maybe the problem to solve isn't so much getting people to switch tools as aiding education and discovery for people who don't yet know what tools to use. Edit to add: I know a couple folks with backgrounds in statistics. One of them taught a data visualization class using R during his postdoc and is now teaching data science in Python for a code school. The other works for a big established company that does lots of data analysis, and still uses one of the old-school tools you listed, but is always very interested when we chat about the newer tools that are out there. So I think it's a mistake to think most folks are stuck in their ways.