Dan Rosenheck writes:
Our data-journalism team conducts original quantitative research, deploying cutting-edge statistical methods to ask and answer important, relevant questions about politics, economics and society. We are looking to hire a full-time political data scientist. . . .
The data scientist will oversee all of our poll aggregators and predictive models for elections. This entails learning all of the code and calculations behind our existing politics-related technical projects, such as our forecasting systems for presidential and legislative elections in the United States, France and Germany; proposing and implementing improvements to them; and setting up and maintaining data pipelines to keep them updated regularly once they launch. The data scientist will also have the opportunity to design and build new models and trackers on newsworthy subjects.
It’s a permanent, full-time staff position, to replace Elliott Morris, who worked on the 2020 forecast with Merlin and me (see here for one of our blog posts and here and here for relevant academic articles).
Sounds like a great job for a statistician or a political scientist, and I hope I’ll have the opportunity to work with whoever the Economist’s new data scientist is. We built a hierarchical model and fit it in Stan!
On that topic, I was browsing the 2020 model (https://github.com/TheEconomist/us-potus-model), and not knowing Stan, it didn’t look trivial to run it or to update it with new data. Any thoughts here on whether I should just persevere, or whether there are some ressources that could help here?
Nuño,
I’ve never run the model myself—all I did was help Elliott and Merlin build, fit, check, and expand it—but I’m pretty sure it’s trivial to run using the R and Stan code that’s there.
Andrew,
It seems likely to me now that you were wrong, and that the 2020 part wasn’t directly reproducible. See https://github.com/TheEconomist/us-potus-model/issues/28
I think that what is missing is some equivalent of this file: https://github.com/TheEconomist/us-potus-model/blob/master/data/all_polls.csv for the 2020 polls, rather than for 2016, and it would also be convenient to get some command to run it, but I’m not totally sure.
It’s also possible that there may have been some miscommunication between you and your co-authors, where you were under the impression that the model was fully reproducible when some parts of it weren’t. E.g., you conclude a paper linked in the post above (https://www.stat.columbia.edu/~gelman/research/published/jdm200907b.pdf) with “Fortunately, when we make public predictions using open data and code, we have many opportunities to learn”, which would be ironic to write if your data for the predictions you made in 2020 wasn’t actually public.
It’s also still possible that I didn’t understand the answer in https://github.com/TheEconomist/us-potus-model/issues/28 well enough, and that it is reproducible after all. I would find that surprising, but want to hedge my bets so as to not embarrass myself in case there is some misunderstanding going on, e.g., by me not understanding stan, the data being in another repository, or something on that vein.
Best,
Nuño
Nuño,
I’m not sure, as I’ve not tried to run the model myself, and when I’ve run it with my collaborators, they were doing it directly and didn’t need to access the public data and code. So it could be that some necessary data are not there. I’ll ask my collaborators what’s up.
OK, I checked and the 2020 data and code have indeed not been posted. So, indeed, my statement, “Fortunately, when we make public predictions using open data and code, we have many opportunities to learn,” was misleading.