Experiment design with Stephen Bailey

[Chang Lee]

In my opinion, social science researchers are great people to learn statistics in a practical way from, so I put together a coffee chat with Stephen where he showed me how he set up experiments in educational neuroscience and different ways to make sure the the experiments produce meaningful results.  

We discussed this tweet on reproducible research. Apparently, not all researchers have enough funding to complete step 1: collect very large dataset. Most of the time one has to work on a limited dataset and rely on common sense/experience/statistical knowledge to convince themselves that they are making ethical data decisions in their research. Knowing that others struggle with this problem is a good makes me feel better because I often feel insecure about the ways I split data into training, validation, and test sets. We also talked a little bit about time-series signals that showed up in neuroscience. Thanks Stephen! 

Chang