How'd they vote?

As usual, it is taking Elections Newfoundland and Labrador an unusual amount of time, compared to its peers in other Canadian jurisdictions, and given the small number of data points involved, to publish the final poll-by-poll results of last November's provincial election.

So here they are, in cleverly colour-coded map format. Enjoy.

The new battlemap

Because, for some reason, the most open and transparent government EVAR can't bring itself to do any such thing, nor even make the underlying data available for public download, here's a zoomable map of the 36 new electoral district in Newfoundland. (Since the four Labrador districts are unchanged, they aren't mapped here.)

Zoom in to see detail at the local level in your community or neighbourhood.

Apologies for the lack of district labels.

Indecision desk

In three provinces or territories, coloured here in red, a tie vote in a provincial or territorial election is decided by a random draw, or, in the case of PEI, a coin toss.

 

In Ontario and New Brunswick (blue), the returning officer casts a deciding vote.

 

In the rest of the country, and federally (green), the tie result is voided and a new election is held.

 

The numbers are references to the relevant provision of the appropriate electoral legislation.

 



By popular demand

By popular demand, here are maps overlaying the proposed boundaries released today by the Electoral Boundaries Commission, on top of the current boundaries, colour-coded by the 2011 general election results. (Changes since 2011 as a result of by-elections and floor-crossings are not reflected.) Existing boundaries are the thin white lines; the proposed boundaries are the thicker black ones.

Click on any map to enlarge it.

Here is the general overview of Newfoundland (the four Labrador districts are unchanged, and so omitted for the purposes of this exercise):

The greater Corner Brook - Bay of Islands region:

The northern Avalon:

And the St. John's metro region:

Thirty-eight (II)

Alan Hall has gone back to the drawing board with some slightly changed assumptions, and drawn up another 38-district electoral map.

Hello, radio listeners.

Thirty-eight

Via electoral cartographer-psephologist extraordinaire Alan Hall, here's a putative example of what a 38-district provincial electoral map could look like.


[Edited to fix embedded map.]

The map

General elections results of 2011 on the left.

 

Current standings (including provisional November 25, 2014, by-election wins) on the right.

 



 

Canada is big

Some rainy Saturday figgerin'.

This is the current 308-riding electoral map of Canada (urban insets excluded.)

The five ridings in the darkest shade of blue constitute more than 50% of Canada's landmass.

The next five, from Labrador in the east to Skeena–Bulkley Valley in the west, bring the total up to 69%.

Adding the next five, in palest blue, which include all ridings of more than 100,000 square kilometres of area, gets you to 79% of Canada.

On the other extreme, the 147 geographically smallest ridings, in urban and suburban areas, contain just over half the country's voting population, but only 0.1% of the area. That's roughly equal to the size of the current riding of Okanagan–Shushwap in the B.C. Interior, highlighted here in yellow.

Confederation, cartographed

A little first for this corner: an interactive map.

This map shows the results of the second-round referendum of July 22, 1948, in which the people of Newfoundland and Labrador chose between Confederation and Responsible Government. This was only the third time that the people of Labrador had the right to vote; the first two being to choose a delegate to the National Convention in 1946, and the second, the first round of the referendum in June 1948.

Each symbol shows a community location and the result on July 22nd. The two green squares are the only two places in Labrador which voted for Responsible Government. The squares in various shades of red voted for Confederation, with the Confederate side taking between 50 and 99% of the vote. The four shades, in ascending depth of colour, represent 12.5% brackets. The yellow stars are communities where 100% of the vote was to join Canada.

Click on a symbol to see the community name (as it was given in the referendum return) and detailed numerical result.


Through some historical accident or another, the poll-by-poll results of the referendum in the electoral district of Labrador survived. In most of Newfoundland, they did not. The figures shown here are drawn from the table published in the December 1984 edition of Them Days. Any errors or omissions are theirs, not mine.

If a community symbol is not in the right location, leave a correction notice in the comments.

Endorsemap (II)

(Updated from here.)

Now that he's no longer Speaker, Ross Wiseman's piece of turf is now shaded a Paul Davis shade of green. And the Hon. Member for Whocares—I'mouttahere, Ray Hunter, is finally on the map for Davis' neighbour and leadership rival Steve Kent.

Outgoing Interim Interim Premier Marshall is the only remaining neutral grey, unless Interim Speaker Wade Verge walks back his Davis endorsement before Saturday. (Click to enlarge.)

Extinction level events

Earlier today, CRA came out with its quarterly report from its August fieldwork.

(If you are a Newfoundland and Labrador Progressive Conservative, this is a good time to close your eyes, stick your thumbs in your ears, and hum loudly.)

Here's what the notional seat forecast looks like, based on today's CRA numbers, if they were the popular-vote numbers in a provincial general election. (Click to enlarge.)

Dark red indicate Liberal holds, paler reds are projected Liberal pickups, and oranges are notional NDP holds. The only small splotch of blue is Humber East, which is, presumably, about to be vacated by Interim Interim Premier Marshall after the PCs pick their new leader in mid-September.

The only other faint hope for the Tory cause, on those popular-support numbers, are the dozen "toss-up" seats on which two projection models can't quite come to agree on the notional winner. The PCs are notionally in contention in ten of those districts. However, on the simplest forecast model, the "uniform swing", the Tories would lose all ten.

(Private to NL PC caucus staff: do you still think the games your guys played with House of Assembly opposition caucus funding was a brilliant idea?)

The NDP would notionally retain three of its 2011 seats, and be in contention in two or three others. However, one of those three holds would be St. John's North, where the formerly-NDP incumbent, Dale Kirby, is now a member of the Liberal caucus. While the NDP is well off the highs it witnessed during its polling surge in 2012-2013, at 15% it is still actually at historically high levels in an NL provincial vote-intent question. Between 1998 and 2011, the Orange team had only ever cracked the 15% mark in a CRA (or any other) poll once, in May 2004, in the wake of widespread unrest in the public sector.

The usual seat-forecasting caveats apply: the notional model does not take into account changes in affiliation since the last election, namely:

• In St. John's South, former PC Tom Osborne, after an interlude in self-imposed exile as an Independent member, is now a Liberal.
• In Mount Pearl South, former PC Paul Lane has also donned a red uniform.
• In St. John's North,  former NDP MHA Dale Kirby will re-offer, this time as the newly-nominated Liberal candidate in the 2015 (14? 16?) general election.
• In The Straits and White Bay North, former NDP MHA Chris Mitchelmore is now the Liberal MHA and nominee in that district.
• In Virginia Waters, Liberal Cathy Bennett now holds the seat following the by-election to succeed former Premier Dunderdale.
• In Carbonear–Harbour Grace, Liberal Sam Slade carried in last fall's by-election what was in 2011 the second-Toriest district in the province.
• In Cartwright–L'anse au Clair, a purely academic change saw Liberal Lisa Dempster succeed Liberal Yvonne Jones.
• In St. George's–Stephenville East, Joan Shea's departure resulted in last week's by-election pickup by Liberal Scott Reid.

As with all poll projections, the overall seat-count result tends to be more accurate than the district-by-district count when the models are tested post-election. The errors in individual district forecasts tend to cancel one another out in the overall composition of the notionally-projected legislature.

Endorsemap

The PC leadership contest endorsements, cleverly colour-coded and mapped. (Click to enlarge.)

Districts shaded in green are where the sitting PC MHA has endorsed Paul Davis; brown are Steve Kent supporters; blue are Ottenheimer's Ottenhomeys. The latter class includes, until Friday, Charlene Johnson in Trinity–Bay de Verde.

Dark green and dark brown are Davis and Kent themselves. Pink and pale orange are held by opposition MHAs or MHA-elect, as the case may be.

The three grey districts are where the PC incumbent has not made any public endorsement, or is officially neutral. The Interim-Interim Premier, and the Speaker of the House, you can understand.

What's up with Ray Hunter?

Comparative contests

If you are still wondering why Progressive Conservative members and prominent operatives, having spent the latter half of 2013 ridiculing the provincial Liberals' leadership process, are now rushing to adopt it — for the 2016 leadership race — wonder no longer.

This map expresses the turnout at the PC leadership delegate selection meetings (DSM) held over the summer as a percentage of the total vote in the Liberals' "virtual" leadership vote in each district. (Click to enlarge.)

In only three districts — Kilbride, Lewisporte, and Lake Melville — did the PC leadership turnout exceed half that of the Liberal leadership vote. In twenty-six districts, more than half the province, the PC leadership turnout was 15% or less that of the Liberal leadership turnout in the same district. This "enthusiasm gap" is especially pronounced outside metro St. John's and "around the [Conception] Bay". However, in both Tory-held districts where the sitting MHA is a leadership candidate — Paul Davis' Topsail and Steve Kent's Mount Pearl North — the PC leadership turnout was less than 10% that of the Liberal vote last year.

The low turnout in Davis and Kent's back yards may be attributable, in part, to the fact that they were among the first districts to hold DSMs. The other metro-area district with particularly poor turnout, St. John's Centre, also held its DSM during the first week of meetings.

There is some evidence that either enthusiasm among PC supporters, or ability of the camps to get their supporters out to the bingo halls, or both, increased, if modestly, as the campaign went on. This is what happens when you sort the DSMs chronologically, and group them by week (starting on Mondays) of the campaign period. Where multiple PC district associations held their DSMs on the same day, the districts are sub-sorted alphabetically. Again, the PC leadership turnout is expressed as a percentage of the comparable figure for the 2013 Liberal leadership contest. (Click to enlarge.)

And here's the PC turnout expressed as a percentage of all eligible voters in the district (per the 2011 election return), which yields a clearer upward trend as the campaign progressed:

Expressed this way, the highest PC leadership turnout was in Lake Melville, at 3.4%.

The lowest turnout in last year's Liberal leadership? Lewisporte. 4.0%.

H/T to @TelegramJames for compiling the PC DSM statistics:

My spreadsheet of PC leadership participation is complete. Average of 80.5 voters per district. https://t.co/e65XxyJ1oC #nlpoli
— James McLeod (@TelegramJames) August 15, 2014

Panic on the streets of Gambo

CRA got scooped on today's release of its NLpoli fieldwork for the month of May.

Here is what the notional-projection electoral map would look like in a general election where today's CRA vote-intent figures reflected actual popular support across the province. Dark colours indicate holds. Light red indicates notional Liberal pickups, while light blue indicates narrow PC holds. Notionally NDP seats (see below) are yellowy-orange. Grey indicates districts where the forecast models are in disagreement about the notional outcome. (Click to enlarge.)

The Liberals would take a minimum of 29 notional seats, and would be in contention in another nine being "tossup" districts where the seat-forecasting models are in disagreement with one another. Even without winning a single tossup seat, this is a solid majority of the House of Assembly's 48 chairs.

For the incumbent PCs, increasingly torn between pollyannatastic denial and desperation, if their Twitter feeds are anything to go on, only five districts would remain in their notional win column, though they would be in contention in the nine previously-mentioned "tossups". But even if all the tossup seats tilted Tory, the governing party would be reduced to an opposition of fourteen members, with only two MHAs west of Harbour Main... and that, assuming that the PC candidate who succeeds Tom Marshall in Humber East holds the seat for the Blue Team. If the Tories fail to win even half the tossup seats, the opposition Liberals would win a commanding majority government of thirty-something members.

The NDP would notionally hold five seats, the same number, though distributed differently, that they won on election night in 2011. However, one of those is St. John's North, where the incumbent MHA, having one for the orange team, just this week won re-nomination as a relatively newly-minted Liberal.

In fact — and this is a major caveat — across the board, the notional model does not take into account changes in affiliation since the last election. These districts are marked with asterixes. There have been quite a few:

• In St. John's South, former PC Tom Osborne, after an interlude in self-imposed exile as an Independent member, is now a Liberal.
• In Mount Pearl South, former PC Paul Lane has also donned a red uniform.
• In St. John's North, as previously noted, former NDP MHA Dale Kirby will re-offer, this time as the newly-nominated Liberal candidate in the 2015 (14? 16?) general election.
• In The Straits and White Bay North, former NDP MHA Chris Mitchelmore is a contestant in next week's Liberal nomination in that district.
• In Virginia Waters, Liberal Cathy Bennett now holds the seat following the by-election to succeed former Premier Dunderdale.
• In Carbonear–Harbour Grace, Liberal Sam Slade carried in last fall's by-election what was in 2011 the second-Toriest district in the province.
• In Cartwright–L'anse au Clair, a purely academic change saw Liberal Lisa Dempster succeed Liberal Yvonne Jones.

Finally, just this week, Joan Shea resigned cabinet and the legislature, leaving a vacancy in the west coast district of St. George's–Stephenville East. This district has been a notional Liberal pickup seat in forecasts derived from public polls going back nearly 18 months, to early 2013. A by-election will have to be called here by August 1st, and held by the end of that month. The rumour mill is rife with delicious speculation about other impending vacancies from the government side of the aisle as well.

As with all poll projections, the overall seat-count result tends to be more accurate than the district-by-district count when the models are tested post-election. The errors in individual district forecasts tend to cancel one another out in the overall composition of the notionally-projected legislature.

The lay of the land – February 2014

CRA is out today with the first poll conducted after the dust has settled on Kathy Dunderdale’s resignation. Here is what the notional-projection electoral map would look like in a general election where the CRA vote-intent figures reflected actual popular support.

Dark colours indicate holds. Light red indicates notional Liberal pickups, while light blue indicates narrow PC holds. Grey indicates districts where the forecast models are in disagreement about the notional outcome. (Click to enlarge.)

The Liberals would take 28 seats – up from 25 in the other most recent poll by Abacus – including St. John’s West. Another nine seats would be tossups in which the Liberal candidate would be one of the contenders.

The incumbent PCs would retain at least eight seats, and would be contenders in eight of the previously-mentioned tossups. If all of the cookies crumbled their way, the PCs would have a remote chance of retaining up to 16 seats and official opposition status.

The NDP would notionally hold on to three seats... but one of those is St. John’s North, where things have changed. Note the important caveat that the forecast models do not take into account changes in party affiliation of the incumbent (St. John's North, St. John's South, Mount Pearl, Straits and White Bay North), nor a seat which has changed hands through a by-election (Carbonear–Harbour Grace). These districts are marked with an asterisk.

As with all poll projections, the overall seat-count result tends to be more accurate than the district-by-district count when the models are tested post-election. The errors in individual district forecasts tend to cancel one another out.

Interestingly, the notional “tossup” seats, those in which the projection models are in disagreement, are heavily concentrated in and around St. John’s. These include the newly-vacated Virginia Waters, where a recent IVR poll showed a close contest between the Liberals and PCs as the by-election campaign looms.

Poll projections in the latter half of 2013 also suggested a notionally close or “tossup” race in Carbonear–Harbour Grace, which in a real-world by-election test was carried by the Liberal candidate.

Paul Oram is right

The PC's are back.

In second place.

Per this morning's Abacus poll for VOCM, conducted in the days following Kathy Dunderdale's resignation, here's what the notional electoral map would look like. The Liberal seat count would be at least 25, including one notional win in St. John's, with another eight potential pickups, not including seats which the caucus has already picked up through floor-crossing and a by-election.

The incumbent PCs would struggle to retain a caucus of more than ten, though they would save the furniture in their traditional northeast Avalon chateau-fort.

The NDP, despite the Troubles, and despite being well off their record-breaking polling highs of 2012 and 2013, would notionally hold onto a caucus of at least three.

Dark colours indicate holds. Light red indicates notional Liberal pickups, while light blue indicates narrow PC holds. Grey indicates districts where the forecast models are in disagreement about the notional outcome.

* The models do not take into account changes in party affiliation of the incumbent (St. John's North, St. John's South, Mount Pearl, Straits and White Bay North), nor a seat which has changed hands in a post-2011 by-election (Carbonear–Harbour Grace). It also does not account for the impact of likely pre-election vacancies (Virginia Waters) or incumbents who choose not to re-offer (Humber East).

About that corner being turned

Here’s the notional district-by-district map* based on today’s CRA vote-intent figures. For the current opposition parties, dark colours indicate holds and pale colours are pickups. For the incumbent outgoing PCs, dark blue is a hold, while paler blue is a hold by less than a notional 10% margin of victory. Light grey indicates a district where the two projection models are in disagreement about the notional winner. The projection models do not take into account changes in affiliation of incumbent MHAs, including the impact of the NDP Big Snit and the Osborg migration from the blue to red teams. (Click to enlarge.)

The Liberal caucus would notionally jump to at least 29 members, with another nine districts too close to call — and all of those, potentially in the Liberal column — for a total of up to 38 seats. Between one and seven districts in St. John's and the suburbs are potential Liberal pickups. In fact, most of the “too close to call” districts are on the northeast Avalon. In Labrador, and off the Avalon, the Liberal party would be expected to sweep nearly every seat. The only notional PC hold, Humber East, is held by incumbent Tom Marshall, who has already made it clear that this is his last term in office.

Without bearing in mind the impact of the Big Snit, the NDP could retain a caucus of five members, including Snit members. The decline in the NDP support figures, while large compared to their recent high polling numbers, is not nearly as large when compared to their 2011 actual result. A notinal loss in The Straits–White Bay North would be offset by a gain in Burin–Placentia West. (Where is that court case, anyhow?)

The more important figure bearing on the NDP fortune is the ongoing suffering of the incumbent outgoing PCs, who have retained barely half their 2011 popular support. The PCs would notionally hold from six to ten seats, depending on how the “too close” races split. And one of those notionally “too close” districts, Carbonear–Harbour Grace, just went Liberal in a real-world by-election test, with an outright majority of the vote and a margin of victory of just over eight percent.

* To be taken with a grain of salt – the overall seat totals in swing models are more accurate than district-by-district projects, as the errors in the latter tend to cancel one another out.

The message from Carbonear–Harbour Grace

Taking the swing from Tuesday's by-election in Carbonear–Harbour Grace, and applying it, for fun's sake, province-wide, yields a notional election result that looks like this (click to enlarge):

Pale colours indicate pickups for the opposition, or narrow holds for the outgoing PCs. There is only one in that latter category: Humber East, where the incumbent has already made it clear he's on his way to Pensionville.

The handful of gray districts are those where the forecast models disagree on the outcome, given the input assumptions. Of the ten tossup districts, only five are tossups where the PC party is still in play. In either case, the largest "universe" of PC seats, where one or the other model has them winning or in contention, is six districts.

Just six.

The forecast model does not take into account changes in affiliation (Osborne) or The Big Snit (the entire NDP and Dissident caucus). Recall that, despite The Big Snit, the NDP vote in the late by-election was essentially unchanged from the general election of 2011 in that district.

So, when does floor-crossing season begin, again?

Math is hard

Nearly 47% of Danny Dumaresque and the two Bennetts' supporters in the NL Liberal leadership vote, failed to mark the second and third preferences on their ballots which would have given them a say on the final count.

Many excuses have been offered, before voting was finished and since, as to why this would be so, but the basic reason seems to be: math is hard. A segment of voters seem to think, entirely wrongly, that marking a second and subsequent preference somehow hurts the chances of their first choice candidate. (It does not.)

Basic dumbosity such as this newspaper column doesn't help. Spot the glaring error.

Of 6539 ballots cast for the three runners-up, 3047 were incomplete after the first or second count. While mostly academic in the present instance, in a tighter leadership race, the incomplete ballot rate could help determine the eventual winner. For some reason, the incompleteness rate was especially high in Baie Verte and Burin–Placentia West.

Contrary to some privately-expressed concerns about the willingness of rural voters to complete their multiple choices, most of the districts with the lowest rates of ballot-incompleteness were in rural Newfoundland.

Play by play

With the full district-by-district Liberal leadership numbers now out, you can do fun things like this. (Click any map to enlarge it.)

First Count

Ball carried 28 districts on the first count, evenly divided between getting a majority of the vote and a plurality. Antle carried fifteen, and Cathy Bennett five, both with pluralities. Neither Dumaresque nor Jim Bennett carried any. Ball took a plurality of the vote in J. Bennett's own, and strongest district, St. Barbe, where J. Bennett was only able to muster 22.08 points. Dumaresque's strongest showing was 27.02 points in his native Cartwright–L'Anse au Clair.

Second Count

As to be expected with so few re-allocations up for grabs, the second count made little difference to the big picture. St. Barbe turned into a Ball majority, Trinity–Bay de Verde into an Antle majority, and St. John's North flipped from the Antle to the Ball column.

Third Count

With Cathy Bennett eliminated, and only two candidates remaining, every district became either a Ball or Antle majority. (There were no ties.) Bennett's re-allocations put Antle back on top in St. John's North, and also turned Cape St. Francis and Mount Pearl South into his column. Bennett's second choice votes gave Ball majorities in Terra Nova, Isles of Notre Dame, and Virginia Waters.

[Updated to correct colour key in third map.]