Jo Wood

@jwolondon

Professor of Visual Analytics who does datavis, visualization storytelling and natural travel. Did I mention visualization?

393 Folgt3.336 Follower773 Posts

YouTube YouTube Bluesky

alt

Ist das dein Profil?Übernimm es — dann gehört dir jwolondon.dirty.social.Profil übernehmen

Jo Wood@jwolondon· Jul 6

London Data Week #LDW starts today with an interesting and packed programme of free events (but book in advance): londondataweek.org/ldw26-events/ I will be giving a talk on Friday about revealing voting biases with 📊 following the recent local elections. londondataweek.org/ldw26-events...

BallotMap for London showing name-ordering bias in ranked votes

A grid of 32 squares arranged in an approximate diamond shape representing the 32 boroughs of London.

Each square is a London borough. Within each, the candidates for the main parties are shown as coloured lines (Greens: green, Labour: red, LibDems: orange, Conservatives: blue, Reform: cyan, Aspire: pink). They are ordered in rows where the top row is the alphabetically first candidate within their party and ward, the second row alphabetically second, and the bottom row alphabetically last. Darkest lines are those where the candidate received most votes within their party; medium lines where they received the middle number of votes within their party and lightest lines where they received the fewest.

Only those in ballots where they were one of three party candidates shown.

This allows us to see where people have split their vote between parties, choosing just one or two from the slate of three in a given party.

A darker lines in the top row compared to the bottom row of each square indicates voting perference for candidates with names closer to start of alphabet.

We see darker lines in the top row for every borough in London. Some boroughs with particularly strong name bias (strong contrast between top and bottom rows) include Enfield (Enf), Havering (Hvg), Bromley (Brm) and Merton (Mrt). Slightly weaker effects can be seen in Richmond (Rch) and Tower Hamlets (Tow) where party preference exerts a strong influence.

Voting data: Democracy Club, 21st May, 2026.

Jo Wood@jwolondon· May 27

These two were juxtaposed in my timeline. Yours might be from the Ladybird Book of the Future "Bicycle riders use a miniature personal computer to watch a live bicycle race."

Screenshot of two BlueSky posts with similar images. The first from Helen Day shows a page from the Ladybird Book "The Story of Football, 1965, painted by Jack Matthew). It shows a 1960s family in a kitchen with father and son at the table completing the Pools. Behind them a mother attends to a daughter. Both conforming to stereotypical gender roles.

Below is the post by Ian Walker showing 6 people looking at a laptop on a table with coloured mugs. All look like they might be cyclists, 4 wearing green cotton cycling caps.

Jo Wood@jwolondon· May 13

Looking forward to reading this. I agree that the questions we ask can be quite revealing. I did this one based on Google queries in 2020 which reveals much about how the UK reacted in a year dominated by Covid (see alt text; questions best read from top to bottom in date order through the year).

A stacked line chart of white lines on a black background in the style of Joy Division's famous Unknown Pleasures album cover design.

Comprises 44 lines each corresponding to most popular 'how / when / what / who' questions to Google over the period of a year. Lines arranged in time order based on the peak time in which a given question was asked. Read from top to bottom this reveals the most pressing questions changing over the 12 months of a year in which the Covid-19 pandemic arose. The questions reading from top to bottom (and therefore broadly from January to December 2020) are:

When does Love Island start?
Who killed Graham?
What does woke mean?
Who dies in Eastenders?
Who is Mabel's mum?
What time is Fury vs Wilder?
How did coronavirus start?
How many cases of coronavirus in UK?
How many people have died from coronavirus?
How to get tested for coronavirus
How to treat coronavirus
What are the symptoms of coronavirus?
What is a pandemic?
Who are key workers?
How to boost immune system
How to make hand sanitizer
How to self isolate
What is coronavirus?
What time is Boris Johnson's speech today?
How to cut your own hair
How to make bread
What does furlough mean?
Who is Carole Baskin?
How to cut men's hair
When will lockdown end?
What is VE day?
When will schools reopen?
How to make a face mask
When does furlough end?
Who was George Floyd?
When will hairdressers open?
When will pubs reopen?
When will gyms reopen?
What does WAP mean?
When does Eat Out to Help Out end?
Who did Zara cheat on Sam with?
Who is Joe Biden?
Who is the president of America?
Who won the election?
When does lockdown start?
When will the new president be announced?
Where does vanilla flavouring come from?
How to cook eel
What tier am I in?

Jo Wood@jwolondon· May 12

I didn't want to pollute this wonderful thread from Moose, so have kept this one separate, but it seems apposite today. Amazing how Prime Minister technology has developed. Our current Prime Ministerial average term of office (left) and an original 1980's term of office, the size of a brick!

A sketchy bar chart showing two bars. A short red bar on the left representing a value of 1.23 years average term of office for Prime Ministers in the 2020s (so far). Above it is the label "2020s". To its right a larger blue bar representing the value 11.57 years for the average term of office for the Prime Minister in the 1980s (who shall not be named). Within it is the label "1980s". The vertical axis labels to the left of both bars show 2, 4, 6, 8, 10 and 12 years.

Jo Wood@jwolondon· May 11

Now all election results for London are in, here is a 📊 BallotMap showing significant name-ordering bias in voting. See annotations / alt-text for details. The map shows us where people choose to split their vote between parties choosing from top of ballot paper, and where party loyalty dominates.

BallotMap for London showing name-ordering bias in voting behaviour.

Each large square is a London borough. Within each, the candidates for the main parties are shown as coloured small squares (Greens: green, Labour: red, LibDems: orange, Conservatives: blue, Reform: cyan, Aspire: pink). They are ordered in rows where the top row is the alphabetically first name within their party and ward, the second row alphabetically second, and the bottom row alphabetically last. Darker cells are those where the candidate was elected to office.

Only those in ballots where they were one of three party candidates shown. Elected candidates indicated in darker colours.

This allows us to see where people have split their vote between parties, choosing just one or two from the slate of three in a given party.

A ◥ shape in a party's distribution indicates voting perference for candidates with names closer to start of alphabet.

Boroughs with split votes showing strongest bias include Haringey (Hgy), Ealing (Elg), Hounsalow (Hns), Wansworth (Wns) and Southwark (Swr). Boroughs with the lest name bias effect are those with strong party preference such as the Liberal Democrats in Richmond (Rch) and Kingston (Kng); Aspire in Tower Hamlets (Tow); Reform in Havering (Hvg); and all parties in Hillingdon and in Bromley.

Voting data: Democracy Club, 11th May, 2026, downloaded at 18:00

Jo Wood@jwolondon· May 10

Interestingly, those London boroughs where Labour held out better than you expected, show evidence of people more likely to split their vote across parties. This is shown by comparing alphabetically first, second and third candidates for each party (see alt text for details). ◥ shape: split vote.

BallotMap showing name-ordering bias in voting behaviour.

Each large square is a borough. Within each, the candidates for the five main parties (blue: Conservative, red: Labour etc.) are ordered in rows where the top row is the alphabetically first name within their party and ward, the second row alphabetically second, and the bottom row alphabetically last.

Only those in ballots where they were one of three party candidates shown. Elected candidates indicated in darker colours.

This allows us to see where people have split their vote between parties, choosing just one or two from the slate of three in a given party.

A ◥ shape in distribution indicates voting perference for candidates with names closer to start of alphabet.

Some boroughs have no recorded results yet (Brent, Redbridge, Tower Hamlets, Newham); Hillingdon only partial result.

Voting data: Democracy Club, 10th May, 2026, downloaded at 15:30

Jo Wood@jwolondon· May 8

If your name wasn't near the front of alphabet you had less chance of being elected yesterday. 📊 For wards with 3 candidates standing per party, we can compare alphabetical name ordering with vote ordering by party If no name bias we'd expect all squares to have the same number of candidates. But…

A set of 6 'bias matrices' arranged in two rows of 3. Each matrix comprises a 3 by 3 grid along 2 axes: Horizontal, rank of votes received within party (1st, 2nd, 3rd); vertical, alphabetical rank of candidate name within the party they are standing for (1st corresponding to names near the top of the alphabet, 3rd near the bottom of the alphabet).

The 6 matrices show respecictively:

Top-left (greys) - counts for all parties with 3 candidates submitted;
Top-centre (blues) - counts of Conservative candidates where 3 Conservative candidates were standing.
Top-right (reds) - counts of Labour candidates where 3 Labour candidates were standing.
Bottom-left (greens) - counts of Green Party candidates where 3 Green candidates were standing.
Bottom-centre (cyans) - counts of Reform UK candidates where 3 Reform candidates were standing.
Bottom-right (oranges) - counts of Conservative candidates where 3 Conservative candidates were standing.

All grids show a strong name bias with higher numbers of candidates both first in terms of party votes and first alphabetically; higher numbers of candidates 2nd and 2nd; 3rd and 3rd.

The effect is strongest for Labour and the Greens and weakest (but still dominant) for the LibDems and Reform.

Jo Wood@jwolondon· May 7

Almost a radar plot, but back in 2015 I designed a tea towel in support of Steve Abraham's world record attempt at the furthest cycled in a year (75,065 miles). Each tomato shows Steve's distance from the village of Marsh Gibbon measured at 15 minute intervals over a 24 hour period. #neverDeleted

A pale cream tea-towel showing a grid of 59 tomato-like symbols in red with a green cross-shaped stalk at the centre. Under each tomato symbol is the date "Jan 1st"..."Feb 28th" written in a 1970's "Goodies" font. At the top is the title "It's Gibbon Time!". Each tomato symbol shows the distance ultra-cyclist Steve Abraham was from the English village of Marsh Gibbon during his world record attempt a the furthest cycled in a year. Distances were measured at 15 minute intervals over 24 hours, producing a a slightly irregular radial outline for each tomato.

The visual style is in reference to the 1970s comedy sketch show "The Goodies". Two of their most famous contributions involved squeezy tomato sauce dispensers ("Bunfight at the O.K. Tea Rooms") and their song "Funky Gibbon".

Close up of the same tea-towel with a red teapot and mug and tin of Red Label tea.

Jo Wood@jwolondon· Apr 16

In response to some of the queries and suggestions for some of the patterns, I've broken down the London data by borough. It shows clearly that more car-centric boroughs are where the correlation between deprivation and no-car access is the most direct.

A grid map of the 32 boroughs of London + City of London, arranged in their approximate geographic location. Each of the 33 squares shows a scatterplot using the same layout as previous scatterplots: Low deprivation on the left, higher deprivation on the right, fewer no-car households toward the bottom of each square, more no-car households towards the top. Dots are coloured by their beta value (direction of local correlation) with most boroughs showing a positive correlation and green dots. A few show clusters of brown dots indicating a negative correlation such as Tower Hamlets and Newham, centre-right, Southwark, centre-bottom. Brown negative correlations tend to be toward the top of each square suggesting they only occur when no-car access levels are high. The strongest positive correlations in green such as in Enfield at the top and Bexley in the bottom-right tend to be associated with lower no-car boroughs.

Jo Wood@jwolondon· Apr 15

It is in London where there are some localities where less deprivation is associated with lower car access (orange) and we might want to ask ourselves why this is the case. Perhaps car-rejection is more discretionary, with fewer incentives to own a private vehicle. [7/8]

Map of London using the same orange-green colour scheme as previous image but shown as a continuous surface across London. Borough boundaries are shown as black lines. Most of the map is green, varying in lightness, indicating broad positive correlation between deprivation and no-car households. But there are around 15 local clusters of orange dots indicating a negative correlation: less deprived LSOAs are the ones with fewer no-car households. The largest cluster of orange is around the City of London and East London extending from the Isle of Dogs northward through Tower Hamlets to the Lea valley.

Jo Wood@jwolondon· Apr 15

Geographically Weighted Regression (GWR) allows us to compare no-car access with deprivation more locally, showing the correlation is positive for most of the country (green), but with small pockets showing a negative correlation (orange). [6/8]

Map of England and Wales as a grey background with dots overlaid on England for each of the 34000 LSOAs. The majority of dots are shades of green indicating a positive correlation between deprivation and no-car households. As would be expected in a map of LSOAs, dots are concentrated in centres of population. There are a few small pockets of orange dots most notably in central London.

Jo Wood@jwolondon· Apr 15

If we split the comparison between deprivation and no-car households by region, London stands out showing a greater spread of car access across deprivation levels. This suggests other geographic factors are influential such as urban density and transport policy. Let's explore this further... [5/8]

A grid of 3 by 3 scatterplots representing the regions of England in approximate geographic position. Top row is 'North West' (green), 'Yorkshire and The Humber' (brown), 'North East' (blue-green). Middle row is 'West Midlands' (light blue), 'East Midlands' (dark blue), 'East of England' (yellow). Bottom row is 'South West' (purple), 'London' (red) and 'South East' (pink). Each scatterplot has a form similar to the first scatterplot in this thread showing a curved positive correlation from bottom-left to top-right.

Jo Wood@jwolondon· Apr 15

If we examine each of the 34,000 'LSOAs' of England — local areas of c.800 households each — and compare deprivation with proportion of households with no access to a car, we see a clear pattern: Areas with more deprivation tend to have more households with no car access. [3/8]

Scatterplot showing deprivation ranking from left to right (less deprived to the left and more deprived to the right), and percentage of households with no access to a car from top-to-bottom. The vertical access has labels ranging from 5% to 85% in 5 percentage point increments. There is a black vertical line through the centre of the plot indicating average deprivation and a horizontal line about a quarter of the way up the chart indicating the mean no-car access of 23%. There are 34000 red-brown dots forming the scatterplot show a positive correlation from bottom-left to top-right. The densest region of points is in the lower part of the plot, rising from left to right between around 10% to 50% non-car access. There is some spread of dots above the main line of correlation, especially towards the right of the plot.

Jo Wood@jwolondon· Apr 15

As @ianwalker.bsky.social mentions in the video, richer → more driving. The census records the number of cars/vans every household has access to, which we can split by deprivation 'decile'. Consistently, the more deprived an area is, the less household access to private vehicles. [2/8]

0 vertically stacked bars arranged from left to right. Each bar is divided into 4, ranging from pale orange at the bottom to dark orange at the top. The proportion of each bar that is in the darker shades decreases from left to right. The bars are labelled at the bottom with 'less deprived' on the left 'deciles' in the middle with arrows pointing left and right, and 'more deprived on the right'. The four shades of orange a labelled on the right with the darkest orange representing 'three or more cars', the medium orange labelled 'two cars', the lighter orange 'one car' and the lightest orange 'no car access'. At the very bottom is the attribution "Source: ONS 2021 Census (English LSOAs only)".

Jo Wood@jwolondon· Mar 22

I love the dissonance that results from two opposing uses of quotation marks. On the one hand, they emphasise by saying someone else says this too, so must be true. On the other, scare quotes do the opposite — it's only an unreliable claim from someone else. See also 1980's Seabrook crisps.

A picture of an empty packet of Seabrook smokey bacon crisps. The labelling on the packet design reads, with quotation marks included:

"More" - than a "snack"
"Value" 30 grams "e"
"Crinkle cut"
Seabrook sea salt smokey bacon flavour potato crisps

Jo Wood@jwolondon· Feb 26

Shadowed somewhat by the more egregious fabrication of "Tactical Choice" by Labour in the Gorton and Denton by-election was this chart crime. Although credit to them, they did at least provide some handy grid squares for checking that non-zero axis problem.

Photo a leaflet provided by the Labour Party (Photo credit: Green Party) on the eve of the Gorton and Denton by-election. It shows a bar chart with 4 hours representing the predicted proportion of the vote for 4 parties: 32% Reform in teal; 31% Labour in red; 28% Green in dark grey; and 9% others also in dark grey. The heights of the bars show a larger gap between Labour's 31% and Green's 28% than we would expect if height was linearly scaled with value. It suggests the baseline of the scale is a value greater than 0. In the background of the bar chart area is a regular grid square pattern of grey lines.

The text in the top right of the bar chart region reads "Source: Britain Predicts" with small print below it reading "(not including polling funded by Green Party donors)".

To the right of the bar chart is text reading "Your polling day checklist. Remember to bring your ID to vote; Find your polling station at iwillvote.org.uk; Polls are open from 7am to 10pm"

Jo Wood@jwolondon· Feb 26

Mostly harmless. Great #datavis work from @theo.io bluesky-map.theo.io?live=1

A 'star map' of the BlueSky universe. Hundreds of thousands of glowing yellow and green dots against a black background give the impression of clusters of galaxies in an astronomical image. Some clusters are labelled including (in anti-clockwise order from the top): Frisian Cultural Delta; German Antifascist Expanse; Web Developer Reef; Spanish Literary Peninsula; Turkish Progressive Basin; Brazilian Youth Constellation; British Socialist Highlands; Life Sciences Supercluster; Game Development Galaxy; Resistance Plateau'; Furry Fandom Plains; Black Diaspora Nebula; Dutch Green Lowlands; Finnish Intelligentsia Ridge; K-Pop Indonesian Shoals; French Progressive Archipelago; VTuber Straits; and Japanese Tech Nebula; Towards the left of the image where clusters are smaller but more frequent is my Bluesky handle @jwolondon shown in pink.

A slightly zoomed in portion of the star map of the BlueSky universe. Hundreds of thousands of glowing green dots against a black background give the impression of clusters of galaxies in an astronomical image. Most clusters are circular or elliptical in shape and many are labelled with the topics/themes associated with the clusters of BlueSky posters. Towards the centre of the image is my Bluesky handle @jwolondon shown in pink.

A further zoomed in portion of the star map of the BlueSky universe. A few hundred individual green dots against a black background give the impression of star map. There are 5 main dispersed clusters of dots, 4 of which are labelled: Complex Systems Science Web; R Programming Tidyverse; Geospatial Professional Atlas and Data Visualization Studio. Above and to the left of the centre of the image close to the Data Visualization Studio cluster is my Bluesky handle @jwolondon shown in pink.

Jo Wood@jwolondon· Feb 20

#duboischallenge2026 Week 2. Original and @observablehq.com Plot recreation Hachures and watercolor effects via SVG styling of Plot marks: observablehq.com/@jwolondon/h... (subtle) hand-drawn edge style via sketchy plot: observablehq.com/@jwolondon/m... Typography: www.vocaltype.co/buy-du-bois

Original Plate from W.E.B DuBois's Paris 1900 Expo collection. Title at the top reads (in capital letters) "Relative negro population of the states of the United States" in black against a faded pale brown background. Below in the centre of the page is choropleth map of the conterminous US states colored mostly in light grey watercolor representing low black populations with other states towards the south and east in browns, reds and yellows representing higher categories of population counts. Below the map is a legend of the 10 categories arranged in two columns ranking population categories from under 10,000 (light grey) to 750000 and over (black).

Reproduction of the W.E.B DuBois's map. Title at the top reads (in capital letters) "Relative negro population of the states of the United States" in black against a faded pale brown background. Below in the centre of the page is choropleth map of the conterminous US states colored mostly in light grey watercolor representing low black populations with other states towards the south and east in browns, reds and yellows representing higher categories of population counts. Below the map is a legend of the 10 categories arranged in two columns ranking population categories from under 10,000 (light grey) to 750000 and over (black).

Jo Wood@jwolondon· Feb 15

I do think one of the most significant developments in computing in the last decade (apart from ~that~ one) has been the computational notebook. Browsing through my @observablehq.com notebooks, they capture my visual and computational thinking more effectively than anything that preceded them.

A grid of 140 small images representing the 'thumbnails' of Observable notebooks. Below each is a small block of text (largely unreadable at this scale) with the title of each notebook.

Jo Wood@jwolondon· Jan 1

I feel obliged to mark the occasion (even if somewhat temporarily).

Photograph of a panettone from above against a wooden table with a quarter of it eaten.

Jo Wood@jwolondon· Dec 18

Interested in a fully-funded #datavis PhD with training? Applications for our 2nd cohort of PhD students at Diverse-CDT open! We're keen to diversify the sector, encouraging anyone with enthusiasm for using 📊 to make change, even if you've not considered PhD study before: diverse-cdt.ac.uk

Photo of students and staff from our first Diverse CDT cohort working through a group data visualization exercise together around a table in our dedicated visualization lab.

Photo of three students from our first cohort of Diverse-CDT PhD students reading and discussing visual thinking in our dedicated visualization workspace.

Sketch of a 'skills map' design co-created by Diverse-CDT PhD students during their first 'immersion week' at the start of their PhD journey. Sketch courtesy of Ana Marques, one of the students. It comprise a series of coloured dots, each representing a skill held by one or more of the students. Dots are clustered by skill type with some labelled for context. Dots are connected with lines in a tree-like structure with 'Diverse CDT' at the centre emphasising relationship between skills and skill types.

A photo of our first cohort of Diverse CDT PhD students who embarked on their PhD journey together in October 2025. Students are located in our dedicated Visualization learning space, posing for the camera with arms folded.

Jo Wood@jwolondon· Dec 17

Such a lovely day yesterday. Sandwiched between days of rain.

Photograph of sunset over Ambleside. Dry stone wall in the foreground on a hill overlooking the roofs of the southern edge of Ambleside. On the horizon an orange glow from the setting sun behind low lying fells. The sky blends from orange on the horizon to a grey blue at the top of the phot.

Jo Wood@jwolondon· Nov 17

Taken today (17th November 2025). "New" doing some heavy lifting.

Photograph of a road sign against a blue sky with light clouds and the corner of brick terrace in Islington, North London. The road sign shows a standard 'No through road' "T" sign above another sign reading "New road layout for social distancing" in white text on a red background.

Jo Wood@jwolondon· Oct 19

Do you want a bit more sketchiness in your #datavis? Here's my @observablehq.com Plot plugin that turns an Plot specification into a sketchy version with minimal extra work: observablehq.com/@jwolondon/m... Builds on the work of Gordon Tu and rough.js

A grid of 24 data visualizations all rendered in a sketchy style. From left to right, top to bottom: An area chart in red and black; a vector arrow in red; a grid with axis labels; three red dots; an empty frame; a world map; a bar chart; a Bollinger chart; three box plots; a grid; a hex bin density plot in red and blue; a red line with one maximum and one minimum point of inflexion; a contour plat in rainbow colours; three elliptical regions in yellow, blue and red; a dodge plot showing two peaks; a scatterplot of blue points with red linear regression line; 4 lines intersecting at a central point; a normally distributed histogram; a normally distributed histogram represented as red vertical lines; vertical tick marks in red looking like a bar code; a tree diagram with 8 nodes; a vector arrow; a voronoi diagram with yellow, blue and red regions; a waffle chart showing 3 categories as vertical stacked squares in blue, orange and red.

Jo Wood@jwolondon· Oct 15

Possibly something happening in your browser/OS? Here is the same block of text as seen on my Mac / Safari. The x-height is pretty consistent. Note also the larger gap below the dots on the i.

Small section of text in the Quiosco font. The text reads

"Trump's decisive Elect
more sharply defined st"

with the left and right words truncated in this enlarged sample.

Jo Wood@jwolondon· Sep 27

Channels

A complex knotted path of curved red lines but structured around a meandering channel and apparent contours each with a different texture of curved lines.

Jo Wood@jwolondon· Sep 27

Fields

A complex knotted path of curved red lines but structured in seven rows of unequal height, each with a different texture of curved lines.

Jo Wood@jwolondon· Sep 27

And another thing...

A complex knotted path of curved red lines that show a grid-like structure.

Produced by a tessellation of hex Truchet tiles with arcs and bezier curves that intersect with each hex edge's midpoint. Rendered in Observable Plot with a custom Truchet mark.

Jo Wood@jwolondon· Jul 29

* Where shall we go for a walking holiday this year? - How about the Alps? * No, way too scary. I don't have a head for heights. - OK, what about the gentle grassy domes of the Auvergne? * Sounds great. Day two of our walking holiday...

A black-and-white photograph of a dramatic mountain landscape around Puy De Sancy, Auvergne. Jagged peaks rise steeply, their dark, textured surfaces contrasting with the low-hanging clouds that partially obscure the summit. Near the center top, a solitary building (cable car lift) sits above a steep cliff.

Jo Wood@jwolondon· Jul 26

Currently in Lyon for some #eurogames and in preparation for some Auvergne walking next week. And Lyon is certainly good preparation, being a city of many many steps. Here's just the first set from the 400+ up to our apartment.