Tessera
Folgen

Juraj Medzihorsky

@medzihorsky.bsky.social

Assistant Professor in Social Data Science at Durham U · Project Manager at V-Dem · Previously at LSE Methodology and U of Gothenburg · CEU PhD

1718 Posts1743 Follower628 Folgt
Juraj Medzihorsky@medzihorsky.bsky.social · 10h ago

The four longest-serving non-royal state leaders rule over some of the youngest populations Born during the rule of … 🇨🇲 91% Mr Biya (93) 1975- 🇬🇶 84% Mr Obiang (84) 1979- 🇺🇬 85% Mr Museveni (82) 1986- 🇪🇷 76% Mr Isaias (80) 1991- 2025 Electoral Democracy Index (V-Dem) 🇨🇲 0.26 🇬🇶 0.17 🇺🇬 0.25 🇪🇷 0.07

A chart showing four population pyramids in a 2by2 grid. These are (left-to-right, top-to-bottom) Cameroon, Equatorial Guinea, Uganda, Eritrea on 1 Jan 2026 according to the median projections of UN WPP 2024. Each pyramid highlights the population born during the rule of the state leader.
Juraj Medzihorsky@medzihorsky.bsky.social · 13h ago

Would like to see how the current batch of LLMs performs when it comes to null hypothesis significance testing: • Do they still follow the null ritual (and would thus fail a test like this one used by Gigerenzer et al 2004)? • If so, to what extent it depends on the context, esp. the discipline?

Screenshot of the following text from Mindless Statistics by Gigerenzer et al 2004 Rituals call for cognitive illusions. Their function is to make the final product, a significant result, appear highly informative, and thereby justify the ritual. Try to answer the following question (Oakes, 1986; Haller and Krauss, 2002): Suppose you have a treatment that you suspect may alter performance on a certain task. You compare the means of your control and experimental groups (say 20 subjects in each sample). Further, suppose you use a simple independent means t-test and your re- sult is significant (t = 2.7, d.f. = 18, p = 0.01). Please mark each of the statements below as “true” or “false.” “False” means that the statement does not follow logically from the above premises. Also note that several or none of the statements may be correct. 1.You have absolutely disproved the null hypothesis (that is, there is no difference between the population means). [] true/false [] 2. You have found the probability of the null hypothesis being true. [] true/false [] 3. You have absolutely proved your experimental hypothesis (that there is a difference between the population means). [] true/false [] 4. You can deduce the probability of the experimental hypothesis being true. [] true/false [] 5.You know, if you decide to reject the null hypothesis, the probability that you are making the wrong decision. [] true/false [] 6. You have a reliable experimental finding in the sense that if, hypothetically, the experiment were repeated a great number of times, you would obtain a significant result on 99% of occasions. [] true/false [] From G. Gigerenzer / The Journal of Socio-Economics 33 (2004) 587–606
𝙅𝙖𝙘🍩𝙗 𝙉𝙮𝙧𝙪𝙥@jacobnyrup.bsky.social · yesterday

It is impressive/terrifying how much Claude has improved its quant skills over the last year. It used to make tons of weird decisions; not accounting properly for time, not accounting for autocorrelation, coming up with nonsensical robustness tests etc. Now most of what it does seems reasonable.

Juraj Medzihorsky@medzihorsky.bsky.social · 2d ago

Datavi(be)z Anno MMXXVI “… like a face drawn in sand at the edge of the sea.” I suppose the first half of September did indeed pass faster.

Raphael Wimmer@raphaelwimmer.bsky.social · 2d ago

Some evidence supporting the assumption that an image generation tool was used (or someone drew the graph 'by hand'): the vertical lines are not evenly spaced - 15 days in October seem to last longer than 15 days in September. Also, the y-axis range for 2500 - 3000 is smaller than the others.

Juraj Medzihorsky@medzihorsky.bsky.social · 2d ago

Works also without assuming that the polls are unbiased. Just set it up within a dynamic model that includes house effects. Here's a plot from this setup with Czech pollsters in 2017.

Bild
Leonardo Carella@leonardocarella.bsky.social · 2d ago

So my new approach inverts the question. My measure of herding answers the question 'given these polls, assuming they are unbiased, what average sample size would produce this much variance?' I call this the "Implied Sample Size" measure: we want it to be *lower* than the (average) reported N.

Juraj Medzihorsky@medzihorsky.bsky.social · 5d ago

Holborn and St Pancras, avoiding the percentbrain ·2024· ·2026· Lab 18,884 12,292 Grn 4,030 8,928 Con 2,776 1,922 Ref 2,371 1,612 Res 692 LD 2,236 924 AF 7,312 Lab+Grn+LD 25,150 22,144 Con+Ref+Res 5,147 4,226

Juraj Medzihorsky@medzihorsky.bsky.social · last wk.

Václav Havel would have turned 90 yesterday. There's something epic about the fact that Czechoslovakia's first and last presidents were both such striking characters, envisioning and achieving goals that others struggled to imagine.

A black and white photo of Vaclav Havel in the back seat of a convertible somewhere in Capri, Italy in 1990.
Juraj Medzihorsky@medzihorsky.bsky.social · 2w ago

Smer and Republika make another step closer to each other: Smer MEP Ms Laššáková steps down as their candidate for the Prešov governor, making Republika's MEP Mr Mazurek the de facto candidate of the government coalition, even though Republika has no MPs and cannot be its part.

Laššáková nebude kandidovať za županku. Šance Mazureka sú teraz dramaticky vyššie, tvrdí Republika | Aktuality.sk

Judita Laššáková sa podľa Smeru rozhodla odstúpiť v záujme „zreálniť voľby“ župana.

aktuality.sk

Juraj Medzihorsky@medzihorsky.bsky.social · 2w ago

Mr Musk's favorite Slovakian party Republika (ESN) with a bit of a 🎷 pattern in the polls. Since the '23 elections, the range has increased in steps. In the past few months it reaches from 8% (Aug Infostat, quota F2F) to 16% (Sept NMS, online panel). Latest AKO (phone and online mix) has it at 10%.

Pooled opinion polls for the Slovakian party Republika [ESN].