JoNova
A science presenter, writer, speaker & former TV host; author of The Skeptic's Handbook (over 200,000 copies distributed & available in 15 languages).
Jo appreciates your support to help her keep doing what she does. This blog is funded by donations. Thanks!
Follow Jo's Tweets To report "lost" comments or defamatory and offensive remarks, email the moderators at: support.jonova AT proton.me
Statistics
Phew it was hot in the UK today, tropical heat, not dry.
Record wise it was not as dramatic as predicted.
The hottest area was a squat blob about 100 mile diameter with its bottom sitting on the Isle of Weight.
Several places broke the June record, Gosport highest reported so far at 36.1C, so half a degree in it.
My town had its warmest night 21C on record for any month, and warmest day today 34.9C, both by a fair margin.
Tomorrow is expected to be at least as hot, but the hottest area will move further west. But 39/40C looks less likely now.
Europe sizzled records in the mid 40s, massive part of Western Europe 35C or more.
00
That happen with hot air coming directly from Northern Africa.
00
And hereâs me thinking all the hot air emanated from Brussels!
00
I find I learn so much from this site and others.
Another site I particularly enjoy is XKCD.com where the creator blends comics with real information. I recommend it for your daily dose of uplifting humour.
Yesterday the comic used the term P-Hacking. I had not come across it, so, as I do, I went searching. It is probably a term a lot of Joâs readers already know.
P-hacking (also known as data dredging or data fishing) is the practice of manipulating data or statistical analyses until a study yields a “statistically significant” result, usually represented by a p-value of less than 0.05. It undermines scientific integrity by artificially creating false positives.
In statistics, a p-value measures the probability that your results happened purely by random chance. The standard cutoff for significance is p < 0.05. P-hacking occurs when researchers try "trial and error" tactics until they cross this magic threshold, rather than objectively testing a pre-written hypothesis.
Common tactics include:
⢠Selective Reporting: Running dozens of different tests, but only publishing the one or two that yielded significant results.
⢠p-Harking: Changing your initial hypothesis after looking at the data to match whatever random trend happened to be significant.
⢠Data Exclusion: Throwing away data points or "outliers" until the remaining data proves your desired point.
⢠Selective Subgrouping: Slicing a larger dataset into smaller demographics (e.g., separating by age, gender, or location) until one group shows a significant effect.
⢠Stopping Data Collection Early: Repeatedly checking for significance while the experiment is running and stopping the exact moment p < 0.05 is reached.
So I asked google:
How P-Hacking Manifests in Climate Research
Climate science relies heavily on massive, multi-variable, and highly complex observational datasets. This complexity grants researchers high "researcher degrees of freedom," allowing them to manipulate results consciously or unconsciously in several ways:
⢠Selective Time Windows: A researcher might test various time subsets (e.g., 1970â2010 vs. 1975â2015) until the correlation between CO2 spikes and a specific extreme weather metric becomes statistically significant.
⢠Model Specification and Regressors: In economic climate researchâsuch as studies mapping country-level CO2 driversâthere is no unified empirical framework. Researchers can cycle through endless combinations of control variables (covariates) until their preferred variable yields a low p-value.
⢠Proxy Calibration and Selection: Reconstructing historical global temperatures relies on climate "proxies" like tree rings or ice cores. P-hacking can occur if data is selectively standardized against specific calibration periods rather than the entire dataset, artificially weighting certain data points.
⢠Data Dropping and Outliers: Researchers might systematically exclude certain measurement stations or specific regional anomalies under the guise of "cleaning data," stopping only when the remaining data aligns with the desired hypothesis.
So, for me, it was an illuminating read. However it quickly morphed into the Wikipedia and YouTube default to brainwash the uninformed.
You know the story (consensus and 105% of experts agree blah blah). See what I did there, I P-Hacked my own research to come up with an impossible number, /s
After warning us about the dangers of manipulating data through P-Hacking, the last 3 points downplay its significance and tell us that science is complex, and you cannot possibly understand it so believe in the results you are fed. After all, itâs scientific and can be found on our controlled websites where we tell you what to believe.
Impact on Scientific Consensus
It is critical to separate the statistical noise of individual studies from the broader foundational physics of climate change.
1. Robustness Against Individual Bias: Climate science draws on vast, corroborating evidence from entirely different disciplines, including basic physics, oceanography, satellite telemetry, and paleoclimatology. A p-hacked anomaly in a single localized study cannot undermine this multi-disciplinary framework.
2. Meta-Analyses and Truth-Filtering: Methodological overviews published in sources like PLOS Biology indicate that while p-hacking is a widespread issue across overall science, its effect is mathematically weak relative to true effect sizes. It fails to alter the overall conclusions of rigorous climate meta-analyses.
3. The Overwhelming Consensus: According to Wikipedia's track of scientific consensus, between 98.7% and 100% of publishing climate experts agree that human activity is the primary driver of modern climate change.
Finally, some of the events you can celebrate or commemorate today, June 25th
Global Beatles Day
National Leon Day ("Noel" spelled backward). the exact midway point to Christmas
The Battle of the Little Bighorn ("Custer's Last Stand
The Korean War began when North Korean forces invaded South Korea
Anne Frank's The Diary of a Young Girl was officially published
1998: Microsoft officially released the Windows 98 operating system
2009: The date is widely remembered for the tragic passing of music icon Michael Jackson.
Enjoy your Thursdayđ
00
Good to know … I guess.
“3. The Overwhelming Consensus: According to Wikipedia’s track of scientific consensus, between 98.7% and 100% of publishing climate experts agree that human activity is the primary driver of modern climate change.”
So it’s looking like the new ‘human’ driver of Climate Change will be generating enough power by any means necessary to develop artificial ‘human’ intelligence.
(We need an irony referee up in here.)
The great consensus didn’t mean squat.
Not much disappoints like a human consensus.
That’s because Climate Change ‘science’ was never science at all and was entirely politics.
Politics being money.
The only thing overwhelmed was the Renewal Energy/Net Zero bubble because a new bubble is now inflating and will inevitably deflate like this one.
Though we will still suffer continuing blows to the head from the Consensus cudgel.
And we will have no AI Wonderland just as we have no Renewable Clean Energy Wonderland.
00
00
Such anti-energy policies of the LNP are why they can’t be trusted on energy, or much else really.
It’s One Nation or no nation.
20