Benjamin Lauderdale

Greater love hath no man than to ask for an R function to check a matrix for rows with identical elements, however permuted.

My understanding is that HMRC will not let an employer do this.

That said, there is still the timing problem: you can’t pay people until they have secured right to work.

Relocation Scheme Support and benefits for staff relocating.

These fees are not eligible to be reimbursed as expenses, they have to be paid (post-tax) by the applicant. UCL gives people relocating from abroad a taxable relocation supplement to cover this: www.ucl.ac.uk/human-resour...

Vive La France!

Analogously, if you want research assistants to code some data for you from some texts that purports to measure some feature of the texts, you don’t get to cite that someone once successfully used RAs to code data for them, you have to validate it for your application.

This issue goes way back, I wrote a blog post in 2018 about this with respect to Brexit, referencing TARP in 2008. Markets are not suited to provide informative political signals about the merits of policy: benjaminlauderdale.net/blog/archive...

Is it the year of Linux on the desktop yet?

When we had this visit many years ago they asked us “isn’t this a great form?” and it is a miracle that neither of us said anything impolite. Maybe you are supposed to frame it?

Political scientists are obligated to repost news of elections decided by single votes.

I'm sorry, but I don't make the rules.

Well, not the whole show, the Dambusters Dog Prize section will also be pretty substantial.

Operation Petticoat - Wikipedia

I will once again suggest en.m.wikipedia.org/wiki/Operati... (if you want to have that segment of the show become the entire show).

Had just tapped through to make this joke when I saw Andy’s reply.

🚨 New paper out at @bjpols.bsky.social with @benlauderdale.bsky.social & @chriswratil.bsky.social!

We ask: How does populist rhetoric affect the electoral fortunes of candidates? And is it successful in changing the issue preferences of voters?

Three key findings in the 🧵 below:

My solution to this constrained optimisation problem is to focus on generating lots of true and novel trivialities. Maybe some day they will add up to something? Maybe not. Other strategies are available.

My hot take is that saying things that are simultaneously true, novel, and also non-trivial is incredibly difficult, regardless of whether you are a social scientist or a journalist.

The Gaijin Cookbook by Ivan Orkin and Chris Yang

I would recommend this cookbook, for the Okonomiyaki as well as much else

An almost surely fictitious anecdote about the 1952 US presidential election has a supporter saying “Every thinking person in America will be voting for you.”, to which Adlai Stevenson replied, “I’m afraid that won’t do—I need a majority.” Apply mutatis mutandis.

I am sure that in a few hundred years it will read as darkly funny that what finally brought down the Trump administration was that there were no turkeys for Thanksgiving.

You can pick up some Manischewitz after you land.

The point remains that the combination of text ambiguity and visualising only three of the six potential states of flips is what is generating confusion here, and similarly in many Monty Hall presentations.

Ack, messed up my own example, predictably. To get it to be 1/2 you need to specify the statement such that there are three relevant states. Eg without the flip/randomization of which side is up. I.e “a random pancake is presented, with equal probability, of the three shown, orientation as shown)

I actually think this version is *more* misleading than the classic Monty Hall problem, because at least there we can understand why he doesn't ever reveal the good prize. Why would you go out of your way to present the burnt side of the pancake?

But the “trick” here is simply to under-specify the likelihood and then say “aha! I meant the other one! Gotcha!” Not really a statistical point. The statistical point is that the likelihood function is an important part of determining the right answer!

The version I put above gets you 1/2 as the correct answer. If instead it is “I select one of the three burnt faces to show, with equal probability”, then it is 1/3.

Depending on how you write that sentence, you can change the answer between 1/2 and 1/3. A lot of the confusion here, and in the MH problem, comes from people making different assumptions about the process that generates the observable data.

I think it would be a better formulated problem if instead of “I serve a pancake burnt side up” it read “I select a pancake randomly with equal probability, and flip it to show one of the two sides with equal probability.” As with the Monty Hall problem, the likelihood function is under specified.

One can certainly quibble about the REF results -> funding mapping, but a system of periodically assessing the production of the thing you want, and then providing resources proportionately to support ongoing production until the next assessment, is not a system that lacks ex post assessment.

Whether it was spent effectively gets evaluated by the next REF!

This is a very interesting question. However I fear that if I weigh in here, it might undermine my claim as your co-author that I am so terribly busy that I could not possibly contribute to tidying up these various publisher requests.