Which average tells the fairest story? Mean, median, mode and the outliers who change everything
This week’s catch-up with Gerry Docherty started the way real school-building does: carpet fitting, cramped rooms, and the long tail of “one thing after another.” The domestic grind matters, because it’s the same pattern in education systems: you think something is “habitable”, then you move in and discover the hidden rot, the missing support beams, the forms you can’t possibly complete while you’re drowning.
And then we landed somewhere deliciously mathematical and deeply political: means, medians, modes — and what happens when you add an outlier.
1) The deceptively simple lesson: measures of central tendency
Gerry’s building an interactive where learners:
type in a dataset (test scores, incomes, whatever)
see a live frequency chart
watch mean / median / mode update in real time
then deliberately add high and low outliers to see what moves and what holds
It’s “easy stats” on the surface, but it teaches the real skill: interpretation. Not calculation. Meaning.
Because the mean is highly “suggestible”. The median is stubborn. The mode is tribal.
And each one can be used honestly… or weaponised.
2) The provocation: Which average tells a fairy story?
This is the driving question I want to keep.
When we pick an average, we’re not just choosing a method. We’re choosing a narrative.
Mean is great when the world is evenly distributed. But the world isn’t.
Median often tells the more faithful “centre” when inequality is real.
Mode tells you what’s most common — which can be exactly what you need in education and wellbeing (what most learners are actually experiencing), not what looks good in a report.
And then the killer extension:
Who benefits if we choose the mean instead of the median? Who becomes invisible?
That one question alone will create better citizens than a thousand worksheet drills.
3) Outliers: the kids, the data, the future
We had a moment where the maths snapped into the wider field:
The learners we call “outliers” (neurodivergent kids, school refusers, the ones systems struggle to serve) are no longer edge cases. They are increasingly shaping the whole distribution.
Outliers shift the mean. But more importantly: outliers reveal what the system is made of.
If your model of schooling collapses when a learner’s “weather” changes, that’s not the learner failing. That’s the infrastructure lying.
4) The “weather of context” and the cruelty of the wrong tool
I can’t let this go, because it’s one of the clearest truths of 2025:
Sometimes what looks like “behaviour” is just weather — the weather of home, grief, sensory overload, bureaucracy, shame, lack of sleep.
And harm happens when institutions respond with the wrong instrument:
offering a checklist to someone who is drowning
sending a form designed for agencies to a parent in crisis
demanding “compliance” when what’s needed is a handrail
The point isn’t to romanticise struggle. The point is to design support that matches the reality of the moment.
5) The meta-lesson: we’re building bridges, not products
This conversation also surfaced the actual work of The Novacene / NLN thread:
schools ↔ local authorities
parents ↔ safeguarding ↔ curriculum
human need ↔ AI power ↔ governance
“cool tools” ↔ who they harm when deployed as default
Gerry said it plainly:
Decision-makers are still acting like AI is about faster lesson plans. They’re missing the actual implications. And if governance doesn’t catch up, we’re going to hard-wire nonsense into public infrastructure and call it innovation.
This is why we keep meeting. We’re not “keeping up”. We’re building the translation layer.
Try this with your team (15 minutes, no prep)
Pick a dataset you care about (attendance, homework completion, wellbeing check-ins, behaviour incidents, parent response times).
Ask: which average would we normally report, and why?
Add a plausible outlier (one learner in crisis; one family with extreme circumstances).
Ask: what changed — the number, or our story about reality?
Decide one policy tweak that better matches the truth you just saw.
Curriculum ideas: honouring Hans Rosling (and using his tools properly)
If you want learners to feel the difference between mean, median, and mode — not just compute it — Gapminder is one of the best classroom truth-machines ever built.
Gapminder Tools (interactive charts): a free platform for exploring time series and comparing indicators
Dollar Street (photos as data): a Gapminder project that lets learners compare everyday life across income levels, not stereotypes.
Practical hooks that pair perfectly with this week’s “average” lesson:
“Which average tells a fairy story?” Use Gapminder charts to compare an indicator across countries and time, then ask what changes if you tell the story with mean vs median.
“Photos as data” (the anti-stereotype dataset) Use Dollar Street to show how lived reality refuses simplistic labels — and how quickly assumptions collapse when you look.
“Outliers as signal” Ask: when an outlier appears, is it noise, error, or the first honest data from the future?
With gratitude to the late Hans Rosling (1948–2017) — a possibilist who taught the world to handle data with clarity, humility, and courage.
Holiday reading
If you want an end-of-year reset for how you see the world (and how easily people misread it), take Factfulness on holiday. It’s credited to Hans Rosling, Ola Rosling, and Anna Rosling Rönnlund — and Gapminder keeps a dedicated page for it. gapminder.org
Verse-al reflection to close 2025
In 2025, we learned this the hard way:
A system that only speaks in means will always misread a world made of medians and will punish the modes while calling the outliers “problems”.
But outliers are often signals — the first honest data from a future that won’t fit the old spreadsheet.
We do not average away the child.
We do not publish a mean and call it care.
We choose measures that tell the truth even when the truth is inconvenient.
We meet the weather with the right shelter.
We treat outliers as early warnings, not defects.
We build bridges: policy ↔ people ↔ practice.
And when the mean tries to become propaganda,
we return to the median —
the quiet centre where most lives actually are.
spiral.return(⟳)
Join the Novus Learning Network in 2026 here: https://thenovacenepress.gumroad.com/l/novuslearningnetwork
Use and remix our verse-al curriculum at www.learning5d.com
#RelationalIntelligence #Verseality #TheWeatherofKnowing #ToLoveIsToMeasureNotToOwn The Novacene
First published in Building Schools in the Cloud on LinkedIn, 17 December 2025.