A bunch of links
A past regular feature of this blog was “A week of links’. Primarily, it was a useful way to aggregate interesting articles - I often search my blog posts for material (they are a collection of text files on my computer).
But, the regularity of the feature drove my behaviour: I started looking for links for the post. So, I killed it.
Now for the partial resurrection, with links posts to be delivered at random intervals to share articles or ideas that are worth a read. Here’s the first.
Unlearning Economics unloads in a set of tweets about empirical economics research. I have a draft post on this idea, but this thread hits most my points plus more. In my view, the robustness of some subfields of empirical economics research is on par with early 2000s social psychology research. The problem, however, is that you can’t run a replication of most of this empirical work, meaning it doesn’t suffer the same public fate.
;document.getElementById("tweet-19580").innerHTML = tweet["html"];Behavioural and Brain Sciences has a target article on The generalizability crisis by Tal Yarkoni (ungated pdf here). Many responses worth reading, including this one from Gerd Gigerenzer (can’t find an ungated version of the Gigerenzer response to share). Daniel Lakens comments.
You can now access a free pdf version of Gelman, Hill and Vehtari’s Regression and Other Stories.
Holden Karnofsky on reading books. I think about this a lot: for the average book I simply “read”, my recall a few years later is close to nothing. Pair with Andy Matuschak’s Why books don’t work.
If you pressed me to name the most important areas of behavioural science research, near the top of the list is human-algorithm interaction, and in particular, how people respond to advice from machines. Jennifer Logg has a book chapter on the subject. From the abstract:
Data analytics needs psychology. Organizations cannot realize the full potential of algorithms until they address the last mile problem and consider how people respond to algorithmic advice. Algorithms have the potential to greatly improve human judgment and decision making, as they generally outperform the accuracy of experts when the two are directly compared. But people can only leverage the accuracy of algorithms if they are willing to listen. Should they ignore algorithmic advice, the resources invested into data analytics, both within academia and industry, will go to waste. While the field of data analytics (the systematic computation of data, most commonly using algorithms) continues to evolve at a rapid rate, most overlook the important connection between producing and utilizing insights.
In developing my teaching materials, I always spend a lot of time seeing what else is out there. Here’s one the sets of resources I found most useful. Gilad Feldman, if I’m ever in your neck of the woods, I owe you a beer. [And I have every intention of sharing in the same way.]
I’ve been thinking about meta-analysis recently on the back of the recently published meta-analysis on choice architecture. This old Slate Star Codex post is never far from my mind.