Last Sunday, Orbán's regime collapsed: not in a few cities, but everywhere. New piece using election and admin data from Hungary. The gist of it is that Tisza won broadly, turnout increased across the board, and local economic conditions barely mattered. blogs.lse.ac.uk/europpblog/2...
ONS reports both AWE and real time PAYE estimates as gross, as well as the ASHE survey (included here after inflation adjustment with correct labels)
These are normalised and would be as similar as it gets, but as always…
bsky.app/profile/ding...
Uk and US real median incomes over time, normalised to year 2000
An interesting bit, levels aside
Just kidding. It was an enormous waste of time and I’m so happy no one has to go through that anymore.
Friday after market close
When a reviewer mentions your 'valiant effort' it's never a good thing
It shouldn’t, but it often does. It’s hard to think of a good study that asks “how does x affect y” and answers “between +50% and -50%.” It could indicate a lack of power, but it could also indicate the approach needs a rethink.
Precise nulls are not that hard to publish, but quite hard to do.
I think Im struggling to square it with study design/setting. I’m not saying the tests are actually independent, but I think “arbitrarily many tests” requires some strong assumptions about how they are related. Unless it’s not about the “true” distribution and is about people fudging on the margin
then this is more an implication of phacking, à la Broduer Cook and Hayes? Or just stating that we care too little about magnitude? If an RCT shows a noisy null it rarely teaches us anything, but a noisy signal in one direction is generally quite useful.
But to Noah’s point a lot of those studies belong in that file drawer. If you take them out, the only thing to be learned is that “executing this bad idea was a bad idea”. One could argue this isnt always the case, but my graveyard is full of ideas that are pure noise and I’d rather not share
To put simply: the UK takes DSGE models more seriously than Kydland and Prescott do
I see a lot of papers pretend to do the second by including all the controls and no identification, thus making them less useful at being the first.
I think most would agree with this, but this is much harder to do than what can be published now. Doing the harder and better work only gets marginally more credit, and the journals (including JHF!) hardly let you say it’s better.
If this is the case, why do those papers tend have kitchen sink models that “control” for lots of things that make it hard to know the incidence and distribution? I think the crux of the issue is that big regressions are easy, isolating effects is extremely difficult
Very little upside to this for anyone without ironclad tenure, unfortunately.
Hi! I’m Mary and I’m on the #EconJobMarket this year.
Extreme heat doesn’t just affect students, it affects the people teaching them.
JMP 🧵:
I think people know, but there is nothing stopping them from being paid more. Stanford gives six figures to all residents now.
What would you say your ratio is on papers you submit?
The brilliant Kate Ho has passed away.
She was an amazing economist and a genuinely kind human being. Her work shaped how we think about healthcare markets and her generosity touched everyone lucky enough to know her.
The profession and the world are poorer without her.
-overground
-Thameslink
-home building (compare to nyc or Bay Area)
-cycleways quadrupled
-air quality
A tough 10+ years sure, but slow moderate successes can be hard to appreciate
Every other month on here there is a discussion from top people how they ignore replies and block frequently, and every month in between there is a discussion about why the site has not grown. These discussions somehow never connect.
A "methods primer" article in the journal "BMJ Medicine", titled "Factors associated with: problems of using exploratory multivariable regression to identify causal risk factors"
We wrote an article explaining why you shouldn't put several variables into a regression model and report which are statistically significant - even as exploratory research. bmjmedicine.bmj.com/content/4/1/.... How did we do?
Academia rewards originality, as long as it looks exactly like what came before.
Honest question, why is polling taken so seriously in the UK? What has yougov done to have such a stranglehold on everyone?
Maybe it would have been *even more* much much worse if not for your work
Possibly, but I do think it’s funny how every month or two this sentiment is brought up, and in alternating months there’s a somehow disconnected discourse about why this site is not growing
Just FYI the link appears broken
