For model calibration (esp via logistic regression), does anyone know of a statistical investigation of the properties of the resulting calibrated predictions?

IOW, if we use predictions from one model as inputs to another model, do we know the probability distribution of the final predictions?

I don't think many of my stats folks are here, but FYI - I am registered for this year's Joint Statistical Meetings! Hope to see a bunch of friends there

#Digitalization#Statistics: "We are inclined to assume that digital technologies have suddenly revolutionized everything – including our relationships, our forms of work and leisure, and even our democracies – in just a few years. Armin Nassehi puts forward a new theory of digital society that turns this assumption on its head. Rather than treating digital technologies as an independent causal force that is transforming social life, he asks: what problem does digitalization solve?

When we pose the question in this way, we can see, argues Nassehi, that digitalization helps societies to deal with and reduce complexity by using coded numbers to process information. We can also see that modern societies had a digital structure long before computer technologies were developed – already in the nineteenth century, for example, statistical pattern recognition technologies were being used in functionally differentiated societies in order to recognize, monitor and control forms of human behaviour. Digital technologies were so successful in such a short period of time and were able to penetrate so many areas of society so quickly precisely because of a pre-existing sensitivity that prepared modern societies for digital development.

This highly original book lays the foundations for a theory of the digital society that will be of value to everyone interested in the growing presence of digital technologies in our lives."

So I'm probably going to be nerd sniped into developing a Jupyter notebook to examine the question of how well are mid income families 2 adults and 2 kids doing relative to how well their parents were doing 30 years earlier. I'm going to use a dirichlet prior over the weights on a 5 item CPI based expense index. The missing part is paired nominal earnings of people and their parents... Anyone know a dataset #statistics#data#economics@economics@a.gup.pe

The average parking lot in the US has 38.4% of cars with people staring at their phones for 30 minutes or more. 42.7% of those are scrolling through suggested TikTok videos. 23.8% of the cars leave the engine running.

Particularly egregious misuse of stats from the Guardian: "Although more than a third of the women in the study had been sexually inactive during the past month, fewer than half expressed dissatisfaction with their sex lives."

Sooo... 1/3 inactive, >1/3 dissatisfied. And yet the article is trying to suggest it's at the other end of the scale by framing it as "fewer than half" and behaving as if it's surprising?

If you take the population and divide by the rate of housing starts per year, you get a quantity in dimensions of time and units of years. This quantity roughly speaking is related to the "longevity of a dwelling" you need to have in order for the housing per person that's available not to decline. So if real longevity of houses is more or less a constant, then when this graph is high housing availability is declining, and when it's low it's growing... There's a reason millennials feel cheated

Dagnab it, I am constantly wishing I had more text in my messages and forgetting to tag stuff in my first post. This message is just to tag @economics@a.gup.pe and some hash tags #economics#housing#data#statistics

This discussion is about housing longevity and the adequate production rate of housing starts to keep housing from becoming scarce. There's a graph in the first post that shows very interesting dynamics.

Happy birthday to founder of modern nursing, social reformer, statistician, data visualization innovator & writer Florence Nightingale (1820 – 1910)!
Nightingale earned the nickname "The Lady with the Lamp" during the Crimean War, from a phrase used by The Times, describing her as a “ministering angel” making her solitary rounds of the hospital at night with “a little lamp in her hand”. 🧵1/n
#linocut#printmaking#sciart#womenInSTEM#datavis#nursing#statistics#mathart#MastoArt

English social reformer, statistician and the founder of modern nursing Florence Nightingale was born #OTD in 1820.

Nightingale became famous for her work as a nurse during the Crimean War (1853–1856). Beyond her work in the Crimean War, Nightingale was a prolific writer and statistician. She used statistical methods to analyze and present data on healthcare and public health, making significant contributions to the field of medical statistics.

"Randomized trials cannot address all causal questions of importance in medicine and health policy and may have limited generalizability; thus, investigators may need to use observational studies as a source of evidence to address causal questions. The challenge, then, is to balance the importance of addressing the causal questions for which observational studies are needed with caution regarding the reliance on strong assumptions to support causal conclusions."

"Many of us out here doing applied science have to entirely self-teach and un-learn poor statistics and poor methods training."

So true.

I see recent graduates with the same faulty NHST-based statistical education that I received decades ago. It's disappointing how poorly education has kept up with new and better statistical methods.

In #QuantumFieldTheory, scattering amplitudes can be computed as sums of (very many) #FeynmanIntegral s. They contribute differently much, with most integrals contributing near the average (scaled to 1.0 in the plots), but a "long tail" of integrals that are larger by a significant factor.
We looked at patterns in these distributions, and one particularly striking one is that if instead of the Feynman integral P itself, you consider 1 divided by root of P, the distribution is almost Gaussian! To my knowledge, this is the first time anything like this has been observed. We only looked at one quantum field theory, the "phi^4 theory in 4 dimensions". It would be interesting to see if this is coincidence for this particular theory and class of Feynman integrals, or if it persists universally.
More background and relevant papers at https://paulbalduf.com/research/statistics-periods/ #quantum#physics#statistics

Here's the logical structure of what you will be taught in terms of #statistics as a masters student in pretty much any #science field.

If MY DATA is a sample from two random number generators of PARTICULAR TYPE, and MY TEST has a small p value then MY FAVORITE EXPLANATION FOR THE DIFFERENCES IS TRUE.

This is, quite simply, a logical fallacy. The first thing wrong is that your data IS NOT a sample from a random number generator of that particular type. So we can ignore the rest logically.

Today I am writing on the AIC functions available in my hashtag#R hashtag#Package TidyDensity.

There are many of them, with many more on the way. Some of them are a little temperamental but not to worry it will all be addressed.

My approach is different then that of fitdistrplus which is an amazing package. I am trying to forgo the necessity of supplying a start list where it may at times be required.

Here it is people. A PhD student describing details of what they've come to realize is the completely scientifically bankrupt methodologies their high-powered successful, well funded lab PI demands the lab members do. Everything this person says is basically commonplace in todays labs #science#openscience#statistics#bayesian

Want a simple form of #MCMC analysis in #R well, I got you covered.

My #R#Package TidyDensity has a function called tidy_mcmc_sampling() that is pretty straight forward. It takes a raw vector and performs the calculation you give it over a default of 2k samples.

A five-star rating for Everything is Predictable: How Bayes' Remarkable Theorem Explains the World by Tom Chivers, from Brian Clegg at Popular Science Books.