And then there were two: Vaughan headed to a provincial byelection

The retirement of Ontario Liberal veteran Greg Sorbara forces a by-election in the provincial riding of Vaughan. The seat will likely be filled this fall, the same day as another by-election, in Kitchener-Waterloo.

With Dalton McGuinty’s government (tantalizingly or alarmingly, depending on your point of view) close to a majority, any by-election will be closely watched.

Bearing that in mind, here is an interactive map breaking down how Vaughan voted in the 2011 election. The top layer shows the plurality winner by poll; the next five show four parties in isolation, and turnout.


October 6, 2011: How Vaughan voted

Use the scrolldown menu to see parties in isolation

Heat-mapping baby-making

There was an infographic meme going around in mid-May centred on a heat map Matt Stiles published on his blog showing the most common birthdays in the United States over a period of 25-odd years. Then Andy Kriebel used the same data to create an interactive version in Tableau. (There’s a helpful discussion in the comments.)

A calendar-based heat map is a useful alternative to more traditional line graphs for showing change over time (it’s more original, it looks more interesting), so I decided to make one – by way of learning how to do it in Tableau – using our conception data.

(We crunched birth date data from B.C., Alberta, Ontario and Nova Scotia, deducting 40 weeks, to show an apparent spike in conceptions in the period roughly around Christmas.)

Here’s how it looks:

Searchable interactive: 93 years of Ontario baby names

When we asked the Ontario vital statistics people for historical baby name data going back as far as electronic records were available, I thought we’d turn up 20 years or so – the equivalent Alberta data is available from 1990. So it was a wonderful surprise when a spreadsheet popped up in my e-mail with totals for boys’ and girls’ names recorded in the province going back to 1917. The result is a pair of searchable interactives showing the ups and downs of thousands of Ontario baby names covering nearly a century.

Conceptually, it owes a lot to the excellent interactives Chad Skelton produced for the Vancouver Sun a few years ago, which show British Columbia name trends from 1960 to 2010. (It was clear that there was nothing to be gained from our doing this exercise for BC.)

Here are some highlights:

Names come and go, but sometimes they reappear. The Ontario data shows that several girls’ names of the early 20th century, more or less disused for decades, are now being revived: Lillian, Hazel, Violet and Ruby.)

Edwards peaked in 1936, the year that the ill-fated Edward VIII was crowned – he would abdicate before the year was out, succeeded by his younger brother George VI, a better-respected king. (In Ontario, Georges kept declining after the coronation and also during the war, which seems somehow unfair.)
Charleses and Elizabeths both peaked in 1953, the year Elizabeth II was crowned. Elizabeth has another spike after the 1959 royal tour to Canada – a marathon one by modern standards, at 45 days.
Marilyns peaked in 1956. It’s hard to tell whether the main influence was actress Marilyn Monroe, then at the height of her career, or swimmer Marilyn Bell, whose arrival on the Canadian shore of Lake Ontario two years earlier had attracted hundreds of thousands of people.
Linda, a juggernaut name of the postwar baby boom, peaked in 1947 – Lindas had been rocketing upward in popularity as the Second World War progressed – and declined just as suddenly. Other classic postwar names are Donald (second peak in 1947) Barbara (peaks 1947), Sharon (peaks 1947) and Carol (peaks 1946).
Josephs, Marys and Maries peak in the late 50s and early 60s, then have a sharp decline apparently connected to Vatican II, the reform process that caused many changes in the Catholic church in that period. (Michael peaks in 1963, but with a gentler decline in following years.)

Here’s how it looks with a sample of names from various eras:


Here’s how it looked on Global Toronto’s newscast on Friday:

and on the morning show today:

Crime has a schedule: Time-based visualization of 911 calls

We’ve had a monstrous data set (over half a million lines) around for a very long time (since some point last summer) showing the time but not date of a year’s worth of Toronto police dispatch calls by category – ‘assault in progress and so forth’. The reason we never did anything with it had to do with not seeing how to visualize it in the way you see below. (The concept owed a lot to this graphic).

Recently, we figured out how to create what we needed in Tableau (see below) and what follows is a unique portrait of the daily cycles of Toronto’s dark side. Click on the image to see the graphic/story package:





jQuery and the stray firearms list

Our list of misclassified handguns and automatic weapons likely to be deleted along with the long gun registry is now at 406.

The story lends itself very well to a jQuery data table, an embeddable, searchable, sortable sort of spreadsheet I’ve been wanting to use since seeing it demonstrated at the NICAR conference. There was, as always, a learning curve – troubleshooting interactive tools on deadline (while also trying to report a story conventionally) isn’t anybody’s favourite experience, certainly not mine, but it worked out in the end.

– The first had to do with a quirk of the code in which the last item in the table must not end in a comma, as all the others do. Neglecting this breaks the table in a way that is only visible in IE – in other words, if you tested the thing in Firefox or Chrome at a low-stress point in the day, and didn’t see a problem then, there’s potential for a nasty surprise down the road at a high-stress point of the day. (NICAR-L was helpful solving this problem, I think within 15 minutes.)

– The second had to do with figuring out how to massage the complete table below 920px/w for display in our CMS. This turned out to be a problem with a very low-tech solution – column width is determined by the longest continuous string of characters in the column. A few instances of HARRINGTON&RICHARDSON instead of HARRINGTON & RICHARDSON will push a complex table too far out to the right. The solution, which does not really seem like a digital-age thing, involves holding a straight-edge (a pencil will do) up to the end of the line and scanning down for the long string that breaks the table. This looks ridiculous but works, like some other things.