lingo.lol is one of the many independent Mastodon servers you can use to participate in the fediverse.
A place for linguists, philologists, and other lovers of languages.

Server stats:

64
active users

#statistical

0 posts0 participants0 posts today

At our #statistical consult, I helped a #master student with her meta-analysis in #R . She "admitted" she wrote all code with the help of #chatGPT . (of course, it's the only one, right?)

One hour in, I typed ?forest (the basic R command prompting documentation about the specific function she was using, forest), because I needed to know something about an function argument.

She was shocked to see the help file being displayed saying "Wow, where did you find all this information?"

When performing multiple imputation of missing data, it is essential to evaluate how the imputed values compare to the observed data.

The attached image, created with the bwplot() function, showcases how the distributions of observed and imputed values vary across different imputations for multiple variables.

I’ll be hosting an 8-week online workshop on Missing Data Imputation in R: statisticsglobe.com/online-wor

Continued thread

Krugman:
This is, by the way, [what] standard autocratic regimes are known for.

⭐️In some ways, among their first targets are #statistical #agencies because ➡️ they want the numbers to say what they want the numbers to say.

I’ve been at conferences in Asia where the Chinese government announces that the economy grew 5.3 percent.

And everyone at the conference asks not “why did the Chinese economy grow by 5.3 percent?” but “why did the Chinese government decide to say that it grew by 5.3 percent?”

⭐️The numbers are our political statements, not reality.

🆘And if I were a federal employee at the Bureau of Labor Statistics,
I would be extremely frightened;
quite quickly they’re going to be in the line of fire.

The last time around, back during the Obama years, when there was a lot of "inflation truthers" claiming that the inflation numbers were being manipulated to make it look like there was less inflation than there was.

❌Such accusations are always projections
—it’s what they would do, not what was actually happening.

We turned to various kinds of private sector independent measures of inflation, many of which were originally developed by economists in places like Argentina,
where manipulation of the data was standard so they developed their own ways to measure.

We’re going to be having to do that.

⚠️ My guess is by sometime next year, we’re going to be having to look at proxies for what’s actually happening to the economy,
possibly for what’s actually happening to crime,
because the official numbers are going to be corrupted.

I have #genomic variants called using #FreeBayes. I want to identify sex-linked regions of the genome. As a complement to visualization I’m considering doing some type of #statistical analysis, in part driven by wanting to learn more #stats. I’m less interested in allele differences (GWAS) and more in #heterozygosity, coverage depth, etc. Are there standard models to identify region differences? Could a #MixedModel be used for this?

Professor @melaniemitchell gives 2023 #MargaretBoden Lecture:

”the inability to form conceptual abstraction —and to make abstraction-driven analogies— is a primary source of brittleness and unreliability in #sota #AI

”I think the the problem that AI has faced at least up till recently is that it has relied on #statistical associations rather than forming concepts”

youtu.be/uEN_rOxKkag

#LLMs #cogsci #neurobuzz #neurodon #ml #systemsneuroscience #complexsystems #neuroai #dl
#PhilosophyOfAI

Continued thread

2/

"When #ComplexSystems, such as the overturning circulation, undergo critical transitions by changing a control parameter λ through a critical value λᶜ, a structural change in the #dynamics happens. The previously statistically stable state ceases to exist and the system moves to a different statistically stable state. The system undergoes a bifurcation [...] there are #EarlyWarning signals (EWSs), #statistical quantities, which also change before the tipping happens" [1]

Call for Submissions - Special Issue on Innovations in Mixed-Mode Surveys of the Journal of Survey Statistics and Methodology
Due: March 31, 2023

Topics include:

Innovations in implementation of mixed-mode #surveys
Development and application of #statistical methods of adjustments for mode effects
Evaluation of current imputation, weighting and estimation methods when applied to mixed-mode designs

#data #ResearchMethods

@surveystatisticsandmethodology
@sociology
@politicalscience

Continued thread

I have been thinking about a series from a #statistical point of view, working with people presenting information in audio form - on how they do this, what they do if they make a mistake.

Where should a #podcast address #errors - in either fact or interpretation?
If the source material is wrong, should the approach of correcting differ?

Given that non-traditional media are going to become even more popular, how do we ensure quality within what we consume?
I welcome examples of good practice.

Migrated to sciences.social and lost my #introduction, so:

I’m a #computational #social #scientist, primarily interested in #theoretical #models, #statistical models based on theory, and #history. Applications mostly in #international #conflict and international #order.

My wife Kristen and I are based in #Oslo, #Norway while I’m on sabbatical this year. We love to #travel and try new #cuisines, and I love to read to and play #math #games with our 6-year-old daughter.

As a new person here, part of the twitter exodus, I'll offer an #introduction. I'm a professor of #sociology at Boston College, and starting next year at University of British Columbia. Much of my #research focuses on the #human dimensions of #ClimateChange and the #ClimateCrisis. I work at the intersections of #EnvironmentalSociology, #PoliticalEconomy, and primarily use a range of #statistical methods. I co- #edit the journal #SociologyOfDevelopment, and I'm a chapter author for #NCA5.