Cenpy 1.0.0

Apr 8, 2019 - 1 minutes
I’m really excited to announce that my longstanding package to work the with US Census Bureau API, cenpy, has gotten some long-needed love and attention. The new method of working with the data is really slick (if I do say so myself). Check it out in this gist! If you like it, hate it, or want to improve it, hit me up over on the project or on my twitter. Read more ...

Crunchtober

Oct 24, 2018 - 4 minutes
One thing I find so difficult to accept is crunch time. Not that I can’t cope or even succeed in crunch, per se. But, rather, I’m beginning to find it very peculiar that folks respond to difficult times through the faith in their ephemerality… as if this specific difficult phase is going to pass is somehow heartening. Sometimes, when my job is at its worst, I hear this a lot. Read more ...

Who's Fixed?

Sep 13, 2018 - 3 minutes
When comparing a multilevel model to a fixed-level model, it’s important to consider how things are parameterized. For instance, let’s say you’re conducting comparisons between a no-pooling model and a partial pooling variance components model. In this case, we have: $$y= \Delta u + \epsilon$$ as the specification, where $\Delta$ is the dummy variable matrix, $y$ is the outcome of interest, and $\epsilon$ is the usual homoeskedastic error term for the responses. Read more ...