The hottest Statistics Substack posts right now

And their main takeaways
Category
Top Technology Topics

Despite Biden's claims, Gaza health ministry death toll is accurate, say peer-reviewed scientific studies

Geopolitical Economy Report • 637 implied HN points • 20 Dec 23
🌍 World Politics Statistics
  1. Peer-reviewed scientific studies confirm the accuracy of the Gaza health ministry's death toll statistics after criticism from US President Biden.
  2. The Gaza health ministry has a history of reporting reliable figures, crucial for international organizations' use in understanding the situation.
  3. Experts from Johns Hopkins University and the London School of Hygiene & Tropical Medicine found no evidence of inflated mortality reporting and confirmed the validity of the data provided by the Palestinian MoH.

What I'm reading (April 2024 edition)

Alberto Cairo's The Art of Insight • 199 implied HN points • 27 Apr 24
📚 Literature Statistics
  1. Statistics and probability have a complex history that affects many sciences today. It's important to understand that probability is more about uncertainty than just measuring how often something happens.
  2. Books like 'Normality' explore how the idea of normal has been used to marginalize certain groups of people. The meanings of normal have changed over time and can be harmful.
  3. The connection between different thinkers and ideas can help us understand reality better. Books like 'The Rigor of Angels' look at these links and ask important questions about what we truly know.

I am offering mentoring

DYNOMIGHT INTERNET NEWSLETTER • 1062 implied HN points • 16 Jan 25
🚌 Education Statistics
  1. A mentoring program is being offered to help people achieve their goals. It's open to anyone, regardless of age or career stage.
  2. The mentor will have three 30-minute video sessions with chosen participants, focused on helping them with their specific challenges.
  3. This program is free and based on an application process to select three people that the mentor believes they can assist most effectively.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

How Silver Bulletin calculates our polling averages

Silver Bulletin • 28 implied HN points • 22 Jan 26
🇺🇸 U.S. Politics Statistics
  1. They include almost every professional poll but exclude known fake surveys, hobbyist/DIY polls, polls that use MRP-style smoothing, and polls with leading questions, while internal or campaign polls are allowed if they meet standards.
  2. Each poll is weighted by the pollster’s rating, sample size (with diminishing returns), and recency, and the model caps a firm’s influence so one pollster can’t flood the average; the final averages are produced with local polynomial regression tuned to avoid over- or under-smoothing.
  3. The averages are adjusted for persistent "house effects" through an iterative process (with a small partisan prior applied to explicitly partisan polls), and the generic ballot is translated into state benchmarks using a partisan-lean score combined with a state-specific "elasticity" that measures how swingy each state is.

Did the Thunder get too good, too fast?

Silver Bulletin • 399 implied HN points • 28 Jun 25
🎾 Sports Statistics
  1. The Oklahoma City Thunder became a top team in the NBA really quickly, winning their first championship just three years after being one of the worst teams. This fast rise is unusual in sports.
  2. Despite their success, many fans and analysts are skeptical about the Thunder's achievements because they have not faced well-known competitors in the playoffs, which affects how their championship is viewed.
  3. The Thunder are very young and have a lot of potential for future success, but their small market and style don't generate the same excitement compared to other teams with star players or bigger fan bases.

The Chicago Cubs Really Were Hurt by Playing Day Games

Something to Consider • 139 implied HN points • 09 May 24
🎾 Sports Statistics
  1. The Chicago Cubs had many daytime games which may have made them tired in the second half of the season. This could explain why they didn't perform as well later in the year.
  2. The team only started playing night games in 1988, much later than other teams, which might have also hurt their performance.
  3. Even today, the Cubs have fewer night games compared to other teams, and this could still affect their chances of winning.

Issues with Terminology and Statistics in Weight Science - Part 1

Weight and Healthcare • 459 implied HN points • 13 Dec 23
🏥 Health & Wellness Statistics
  1. The weight loss industry manipulates terminology to market weight loss as a treatment for obesity, leading to misconceptions and ineffective interventions.
  2. The term 'weight-related conditions' is often used inaccurately to imply causation, ignoring confounding variables like weight stigma and healthcare disparities.
  3. The concept of 'sustained weight loss' is sometimes misrepresented by the weight loss industry to imply success, when in reality, it often refers to temporary weight loss followed by regain.

Three Common Statistics Snafus in Weight Science

Weight and Healthcare • 718 implied HN points • 19 Apr 23
🏥 Health & Wellness Statistics
  1. Repeated attempts at intentional weight loss can have decreasing odds of success, and weight cycling can lead to significant harm.
  2. Just because a study result is statistically significant doesn't necessarily mean the effect is important or impactful.
  3. Understanding the actual increase in risk percentage versus the absolute risk percentage is crucial in healthcare decision-making.

Machine learning never cheats but it may play flawed games

Mindful Modeler • 259 implied HN points • 27 Feb 24
🕹 Technology Statistics
  1. Machine learning models may use shortcuts or exploit quirks in data, but it's important to consider them as playing the game according to the rules set by the data.
  2. Detecting flaws in prediction games is crucial, as models can unintentionally learn and act on misleading information from the data.
  3. Designing prediction games effectively requires a deep understanding of the data-generating process, tools like sampling theory, design of experiments, and a statistical mindset can be valuable in shaping prediction tasks.

Nathan Yau's friendly voice

Alberto Cairo's The Art of Insight • 99 implied HN points • 29 May 24
🎨 Art & Illustration Statistics
  1. Nathan Yau is known for making data visualization fun and approachable, both in his blog and his book, 'Visualize This'.
  2. The second edition of 'Visualize This' offers updated examples and tools, making it more cohesive than the first edition.
  3. Reading Yau's work feels like getting hands-on help from an experienced designer, which makes learning enjoyable.

Self-indulgent anniversary post

Logging the World • 518 implied HN points • 04 Nov 23
📰 News Statistics
  1. The author reflects on their first year on Substack, the experience of a post going viral, and their content on COVID and other topics.
  2. The post discusses the author's favorite non-COVID topics including a clever idea, an education policy, and the joys of walking.
  3. The article highlights the impact of a post on Dominic Cummings boosting views, emphasizes the unpredictable nature of virality, and teases future discussions on the UK COVID Inquiry.

Workshop announcement: Causal I

Scott's Substack • 334 implied HN points • 12 Jan 24
🚌 Education Statistics
  1. Workshop announcement for Causal Inference I starting on February 3rd.
  2. Key topics covered in the workshop include potential outcomes and selection bias.
  3. The importance of letting go of anger, bitterness, and seeking human connection in New Year's resolutions.

America's Peacetime Conundrum

Max Meyer Blog • 569 implied HN points • 10 Sep 23
🏥 Health Politics Statistics
  1. In 2022, there were no US soldier combat deaths but over 300 soldier suicides.
  2. The number of veteran suicides has been consistently higher than the civilian rate, with rates increasing over time.
  3. Efforts have been made to address veteran suicide, including anti-suicide legislation and national strategies, but the impact is still being observed.

The 9.3x Factor; Data Isn't Hate Speech

~~datahazard~~ • 550 implied HN points • 12 May 23
🇺🇸 U.S. Politics Statistics
  1. A Black person is 9.3x more likely to murder a White than a White person is to murder a Black.
  2. Comparing murder rates between different population groups can lead to misleading conclusions.
  3. It's important to consider more meaningful rates, like the 'Stereotype Rate', when analyzing murder statistics.

The Dunning-Kruger Effect: Cognitive Bias or Statistical Artefact?

UX Psychology • 218 implied HN points • 16 Feb 24
🔬 Science Statistics
  1. The Dunning-Kruger effect explains how people with lower ability often overestimate their competence, while highly skilled individuals tend to underestimate their capabilities.
  2. The disconnect between competence and metacompetence judgments is seen not only in psychology but also in areas like aviation, memory, and medical skills. It can impact product development and UX teams by highlighting the challenge of relying on feedback from less competent individuals.
  3. Debate continues on whether the Dunning-Kruger effect is purely a statistical artefact or a psychological phenomenon. Regardless, it prompts the need for critical self-awareness, peer review, and continuous growth to mitigate its impact in fields like UX.

Matt Goodwin's train nightmare

Odds and Ends of History • 268 implied HN points • 30 Jun 25
🇺🇸 U.S. Politics Statistics
  1. Getting good phone service on trains is tough. There's a new government plan to fix this problem.
  2. Some official economic data might be messed up, which makes it hard to know how the country is really doing.
  3. Growth is important for spreading wealth, and there's a push to make a specific philosophy part of the government's plans.

Data Science Weekly - Issue 536

Data Science Weekly Newsletter • 179 implied HN points • 01 Mar 24
🕹 Technology Statistics
  1. The DSPy framework makes working with large language models easier by focusing on programming instead of complex prompting techniques. This helps reduce errors and improves usability.
  2. A new sequence model approach shows better performance than traditional Transformers, especially for long data sequences. It also works faster, making it a promising development in the field.
  3. Learning resources like online courses and free books on deep learning and causal ML can help deepen understanding of data science. They provide structured material that is great for both beginners and advanced learners.

The most neglected problem in Britain?

Comment is Freed • 45 implied HN points • 08 Dec 25
📰 News Statistics
  1. More than a quarter of women have been sexually assaulted since age 16, and about a third of those incidents are rapes or attempted rapes.
  2. Sexual offences haven’t fallen over the last 20 years and receive surprisingly little public or political attention, even though other crimes have decreased significantly.
  3. There are proven, evidence-backed reforms — for example improved policing and investigation models like Operation Soteria — that could cut offences in the next few years, so policy should focus on these alongside longer-term cultural change.

BA.2.86 growth rates

Logging the World • 418 implied HN points • 23 Aug 23
🔬 Science Statistics
  1. New COVID variant BA.2.86 has mutations that suggest fast growth, but estimating its growth rate is tricky.
  2. Statisticians use models and likelihood functions to estimate parameters like growth rates, but uncertainty exists in the estimates.
  3. The work of statistician C.R. Rao, like the Fisher information, shows fundamental limits to parameter estimation and the role of geometry in statistics.

Momentum

Market Sentiment • 412 implied HN points • 25 Jun 23
💰 Finance Statistics
  1. Momentum investing is based on the idea that winners will keep winning and losers will keep losing.
  2. In a story about finding a lost submarine, leveraging the collective wisdom of diverse individuals led to success.
  3. A simple momentum investing strategy has shown consistent, outperforming results over years and different market conditions.

Ratings Are Broken

Zwischenzug • 412 implied HN points • 29 Jul 23
🎾 Sports Statistics
  1. Lower rated players are winning more often than expected.
  2. New underrated players entering the rating pool are causing downward pressure on ratings.
  3. Proposed solutions include a one-time rating boost for players under 2000 and changes in rating calculations.

Off-Ball Screens and Three-Pointers

Hoop Vision • 196 implied HN points • 30 Jan 24
🎾 Sports Statistics
  1. Offensive scheme in basketball is more complex than just looking at screen and cut data.
  2. Virginia's offense is unique with a high volume of off-ball screens but low three-point attempts.
  3. The effectiveness and impact of Virginia's Blocker-Mover offense on generating threes is a point of interest for analysis.

Stuff I found interesting in December

Samstack • 1537 implied HN points • 31 Dec 23
🔬 Science Statistics
  1. Be cautious of assuming correlation implies causation, as the sign can be opposite of the true effect
  2. Income inequality in America may not have risen much since the 1960s, contrary to popular belief
  3. Anti-immigration voters often consider the issue more important than pro-immigration voters, impacting public perception

A new chapter on generalization

Mindful Modeler • 99 implied HN points • 16 Apr 24
🔬 Science Statistics
  1. Many COVID-19 classification models based on X-ray images during the pandemic were found to be ineffective due to various issues like overfitting and bias.
  2. Generalization in machine learning goes beyond just low test errors and involves understanding real-world complexities and data-generating processes.
  3. Generalization of insights from machine learning models to real-world phenomena and populations is a challenging process that requires careful consideration and assumptions.

Exaggerating China's military spending, St. Louis Fed breaks all statistical rules with misleading graph

Geopolitical Economy Report • 438 implied HN points • 24 Jan 23
🌍 World Politics Statistics
  1. The Federal Reserve Bank of St. Louis published a misleading graph suggesting China spends more on military than the US, but in reality, the US military budget is three times larger.
  2. The graph's depiction of military expenditure by different countries was misleading due to the use of separate axes for China and the US, creating a false impression.
  3. China's military spending, when looked at accurately, is much smaller than the US when considering GDP percentage, per capita measurement, and even in comparison to other top spenders.

Natural selection: what's the effect size?

Wyclif's Dust • 2414 implied HN points • 07 Apr 23
🔬 Science Statistics
  1. Many polygenic scores are significantly correlated with the number of children a person has, indicating a link between genetics and fertility.
  2. The effect sizes of these correlations can be influenced by the accuracy of the polygenic scores, with noise potentially underestimating the true effects.
  3. Improving polygenic scores and considering the impact of rare genetic variants are important for better understanding the relationship between genetics, fertility, and education.

Stray Thoughts... - You mean Cerberus?

THE BATFLIP • 314 implied HN points • 29 Sep 23
🎾 Sports Statistics
  1. Chris Bassitt reached the 200 inning mark with a stellar performance for the Blue Jays, a rare accomplishment in today's game.
  2. Pitchers going deep into games for 200+ innings is becoming increasingly rare due to injury prevention, expanded bullpens, and awareness of third-time-through-the-order penalty.
  3. Brandon Belt's performance, Rangers' bullpen struggles, and playoff scenarios are important factors to watch for Blue Jays' postseason hopes.

Defending machine learning in a room full of old-school statisticians

Mindful Modeler • 299 implied HN points • 27 Jun 23
🕹 Technology Statistics
  1. Be mindful of your modeling mindset and be open to exploring other modeling cultures beyond your current beliefs.
  2. Recognize that differences in modeling mindsets are deeply rooted in culture and background, influencing how individuals approach statistical modeling.
  3. Interpretability remains a significant concern for modelers, especially in the context of machine learning advancements, although progress has been made in providing tools for better understanding models.

The Social Network (Part 2)

Logging the World • 279 implied HN points • 13 Apr 23
🔬 Science Statistics
  1. Real social networks exhibit more complex behaviors than simple mathematical models can capture.
  2. The structure of social media follower counts differs significantly from the Erdős–Rényi network model, with some users having exponentially more followers than others.
  3. Recent network models like the Barabási-Albert model better represent the dynamics of online social networks like Twitter, where heavy-tailed distributions of follower counts emerge.