5 points to victory; 6 points to majority

Introduction:

I am not a big fan of seat projections at the best of times. With the changes between 2008 and 2012, I think any attempt at a projection for the 2012 election based more on science than intuition was wishful thinking.

That said, given the eleventh hour collapse of the Wildrose vote, I wondered what the result would have been, if Wildrose support did not collapse. What if it were higher by a few points or by several points? As I already had regions separated out in my results spreadsheet, I put together a a reverse projection model based on percentage point swings. For the sake of brevity, the methodology can be found at the end of this post.

It goes without saying there are many assumptions one makes when doing projections, and there are many limitations of such an exercise. Engaging in post-hoc conjecture is purely just-for-fun, not for publication in Electoral Studies.


Five points to a Wildrose victory

Question: What magnitude of a shift would have been needed to change the outcome of the election?

Answer: A five point increase to the Wildrose’s share of the popular vote in every region at the PCs’ expense.

Table 1: Popular vote by region (+5 pts WR)

Region PC WR Lib NDP
Province 39.2% 39.6% 10.0% 9.9%
Edmonton CMA 38.7% 27.3% 13.8% 17.9%
Calgary CMA 40.4% 43.4% 10.9% 4.6%
Outside CMAs 38.5% 47.3% 5.4% 7.6%

Table 2: Seat distribution (+5 pts WR)

Region PC WR Lib NDP Total
Edmonton (inner-city) 5 0 1 5 11
Edmonton (suburbs) 7 0 1 0 8
Edmonton (metro) 6 1 (+1) 0 0 7
Calgary (inner city) 3 7 (+6) 2 0 12
Calgary (suburbs) 8 4 (+3) 1 0 13
Calgary (metro) 0 3 0 0 3
Northern AB (Fort McMurray) 0 2 (+2) 0 0 2
Northern AB (Grande Pairie) 1 1 (+1) 0 0 2
Northern AB (rural) 2 5 (+4) 0 0 7
Central AB (Red Deer) 0 2 (+2) 0 0 2
Central AB (west, excl. RD) 2 4 (+1) 0 0 6
Central AB (east) 2 3 (+1) 0 0 5
Southern AB (Lethbridge) 0 2 (+2) 0 0 2
Southern AB (Medicine Hat) 0 1 0 0 1
Southern AB (rural) 0 6 (+1) 0 0 6
Province 36 (-25) 41 (+24) 5 5 (+1) 87

Notable swing ridings:

  • Stony Plain swings from PC to NDP (only Edmonton CMA WR pickup)
  • Edmonton-Gold Bar swings from PC to NDP (only non WR pickup)

Regardless of how or why the Wildrose’s numbers imploded on the eve of E-Day, this projection shows that, had they received five percentage points more of the popular vote across all regions, they would have won a minority government.

Of course, there are limits to the caveat of ‘ceteris paribus’, and the model cannot account for factors such as local GOTV effectiveness or special circumstances in individual ridings. Still, five points is not a huge gap, and the resulting popular vote numbers look very similar to numbers we were seeing during the campaign.


Six points to a Wildrose majority

Question: What magnitude of a shift would have been needed to produce a Wildrose majority?

Answer: A six point increase in the Wildrose’s share of popular vote in every region at the PCs’ expense.

Table 3: Popular vote by region (+6 pts WR)

Region PC WR Lib NDP
Province 38.2% 40.6% 10.0% 9.9%
Edmonton CMA 37.7% 28.3% 13.8% 17.9%
Calgary CMA 39.4% 44.4% 10.9% 4.6%
Outside CMAs 37.5% 48.4% 5.4% 7.6%

Table 4: Seat distribution (+6 pts WR)

Region PC WR Lib NDP Total
Edmonton (inner-city) 4 (-2) 1 (+1) 1 5 (+1) 11
Edmonton (suburbs) 7 0 1 0 8
Edmonton (metro) 6 (-1) 1 (+1) 0 0 7
Calgary (inner city) 3 (-6) 7 (+6) 2 0 12
Calgary (suburbs) 5 (-6) 7 (+6) 1 0 13
Calgary (metro) 0 3 0 0 3
Northern AB (Fort McMurray) 0 (-2) 2 (+2) 0 0 2
Northern AB (Grande Pairie) 1 (-1) 1 (+1) 0 0 2
Northern AB (rural) 2 (-4) 5 (+4) 0 0 7
Central AB (Red Deer) 0 (-2) 2 (+2) 0 0 2
Central AB (west, excl. RD) 2 (-1) 4 (+1) 0 0 6
Central AB (east) 2 (-1) 3 (+1) 0 0 5
Southern AB (Lethbridge) 0 (-1) 2 (+2) 0 0 2
Southern AB (Medicine Hat) 0 1 0 0 1
Southern AB (rural) 0 (-1) 6 (+1) 0 0 6
Province 32 (-29) 45 (+28) 5 5 (+1) 87

Notable swing ridings:

  • Edmonton-Mill Woods swings from PC to WR (only City of Edmonton WR pickup)

Pickups that push WR from minority to majority:

  • 3 seats in suburban Calgary (Hawkwood, Lougheed, Northern Hills)
  • 1 seat in inner-city Edmonton (Mill Woods)

One of the most notable things about the distribution of the popular vote is that small shifts can create big changes. It is not reported here, but a four point increase in Wildrose support across the province would still return a PC majority, albeit a significantly reduced one (45 PC, 33 WR, others unchanged). Because of regional variances, drastically different results can occur, in terms of seats won and lost, even though changes in the overall province-wide popular vote are comparatively small.

Another thing I noticed, but is unreported here, is that the current distribution of the popular vote across regions predisposes Alberta to majority governments. Alberta has a history of electing majority governments, and as the model shows, it does not take a lot to shift the outcome from a PC majority to a Wildrose one. I tried a few more scenarios in the model, and the likelihood of a minority government seems to increase alongside increased regional divisions in partisan support. This is a casual observation; not the result of systematically testing different scenarios, so I could be wrong.

My use of the four- and five-point swings is illustrative and meant to show how relatively small shifts can produce relatively large changes. Time permitting, I may run and write up the results of a few other different scenarios.


Methodology:

  • Regional divisions are defined as in previous posts (separate divisions for inner-city, suburban, and metro Calgary/Edmonton; separate divisions for urban and rural areas outside of the two CMAs)
  • Changes in the popular vote are calculated using percentage points, with region-wide swings being applied uniformly to each riding. (i.e. If the Wildrose is up by five points in inner-city Edmonton, five points are added to their percentage of the popular vote in each inner-city Edmonton riding.)
  • All changes are zero-sum (i.e. if a party gains five points, there must be a total of exactly five points lost by one or more parties).
  • This is a closed system and only ‘reallocates’ the choices of existing voters; it cannot account for an increase or decrease of turnout. It also assumes that all other variables remain constant (e.g. star candidates, strength of local campaigns, strength of central campaign, existence of third-party campaigns, etc.)
Advertisements

About jbsantos
Polling, politics, PR and outdoor pursuits.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: