Trump can win, but Hillary will win (not!)

UPDATE, The Morning After:

After laying out all the reasons why Trump could win (for months and months), I blatantly ignored that evidence and confidently predicted (below) a decisive Clinton victory.  The power of conventional wisdom and the ‘echo chamber’ was never greater than the past week in US election politics, only to be overcome by the voters who ultimately decide.  For a matter of minutes, each voter is in charge – in the privacy of the voting booth. Each voter is equal – a single mother in Michigan or retiree in Pennsylvania has the same weight as a Hollywood celebrity or Wall Street trader.  And the voters have proved, again, that they are very much in charge.

ORIGINAL POST:

Can Trump win?  That’s the question on everyone’s mind.

Yes he can – he has a pathway.  But I’m betting that Hillary Clinton will be the 45th President of the United States and it won’t be that close.  In fact, I have put my money where my mouth is by betting $5 through BC Lottery Corporation’s online election pool (expires at 4pm Tuesday).

bill-hillary-clinton.jpg

45th and 42nd Presidents of the USA

First, a few starting points to consider when watching the results:

  1. It takes 270 electoral votes to win.  Just because a candidate wins the popular vote doesn’t mean they win the electoral college.  Clinton gaining a higher popular vote in Texas or running up the margin in California is meaningless in terms of electoral votes.  She needs to win states.
  2. There has been a lot of early voting in places like Florida, where early turnout was much higher than 2012 and mostly before the FBI bombshell.  That mitigates late-campaign swings to some extent.
  3. The electoral map is always in a state of flux.  In 1960, the GOP won California and Washington and the Democrats won Texas and most of the South.  This election, we will see some states switch allegiances (in both directions) compared to recent elections.
  4. No candidate in recent memory has been as much of a disruptor as Donald Trump.  He is using social media as blunt-force trauma compared to Hillary Clinton’s better-resourced, data-driven approach.  Trump has ‘macro-targeted’ and his winning scenario is moving non-university degree white voters en masse.
  5. How many times have we been surprised lately?  Justin Trudeau’s majority, NDP in Alberta, Jeremy Corbyn as UK Labour leader (twice), David Cameron’s majority then Brexit, the rise of Bernie, and the rise of Trump.  The people will make up their own mind, thank you very much.  Many voters simply don’t cooperate with polls.  Will ‘cranky won’t says’ make the difference?  That would be good for Trump.

The best available information

Obama’s campaign manager Jim Messina says the Democrats have run 63,000 simulations every night since Obama’s first run for president.  The data available to the Democrats and the GOP is the product of hundreds of millions, if not, billions of dollars of investment.  The public polls may be indicative but, obviously, not wholly reliable.  This is why we mere mortals often get surprised.

Let’s take a look at the work of those trying to figure this out.

> Nate Silver 538 “Odds in HRC’s favour”

Nate Silver’s 538 website has closely tracked public polls.  He puts the odds at 71.9% Clinton, who he predicts will win about 302 electoral votes.  The New York Times ‘Upshot’ has Clinton’s odds at 84%.

In Silver’s winding road to victory graphic, Clinton crosses 270 in New Hampshire and pads the margin with Nevada, North Carolina, Florida, and the Maine 2nd district.

Screen Shot 2016-11-07 at 10.53.08 PM.png

> Real Clear Politics “Uncomfortably Close”

Real Clear Politics has Clinton at 203, Trump at 164, and Toss-Ups at 171.  When pushed into a “No Toss Ups” map, RCP has the margin at an incredibly close 272-266.

Huh?  Isn’t Clinton supposed to be further ahead?  RCP has Trump edging Clinton in Ohio, Florida, North Carolina, Iowa, and Arizona.  New Hampshire is in RCP’s Clinton column but has been flipping and flopping all week like a halibut sun bathing on a Boston Whaler.

Screen Shot 2016-11-07 at 10.59.53 PM.png

Here is the most recent State polling data on Real Clear Politics:

Screen Shot 2016-11-07 at 11.02.26 PM.png

> USC/LA Times Poll “The Outlier”

This nightly tracking poll (via online panel) has been a consistent outlier for months.   If Trump wins, they are geniuses – they have been about 4-5 points to Trump’s favour consistently compared to most pollsters.  This poll does provide a view of campaign momentum.  The RNC convention (7/25), subsequent self-induced Trump collapse (8/12), Clinton health scare (9/17), Billy Bush tape (10/17), and post FBI surge (today).

Screen Shot 2016-11-07 at 11.06.46 PM.png

 

Trump’s pathway

Building on my blog post last week (“Can Hillary lose? Not easily“), here are my revised prognostications going into Tuesday night.

The pathway for Trump to win 270 electoral college votes is not easy.  It would look something like this:

  1. Win all of Romney’s states (206).  Right now, he is forecasted to do that but has been vulnerable in North Carolina (15) and Arizona (11).  He seems to be pulling away in Arizona but NC is a toss up.  Utah is another wildcard where independent Evan McMullin has been in shouting distance of Trump.
  2. Consolidate consistent leads in Obama states (24).  Trump has been leading for a while in two states where Obama triumphed in 2012 – Ohio (18) and Iowa (6).  Now he’s up to 230 total votes with steps #1 and #2.
  3. Win Florida (29).  It would be very, very hard for Trump to win the White House without this state.  The polls are close.  Running total: 259.
  4. Find (11) votes from the following: New Hampshire (4), Maine 2nd district (1), and Nevada (6).  That’s 270 right there in Steps 1-4.  This is very similar to the RCP map above that has Trump at 266 – it’s just missing New Hampshire.
  5. Hail Mary scenario – If Trump’s carpet bombing of previously considered safe Democrat states succeeds, it changes the calculation: Pennsylvania (20) and Colorado (9) could add to or replace Florida’s 29 votes; Michigan (16) or Wisconsin (11) would replace or add to the smaller states in #4 above.  This would be white voters (college education or less) turning out “big time”.  This scenario is a tall order, indeed.

My prediction

I have unreliable data like the rest of you.  So this comes down to a gut feeling. Trump will not win all of the Romney states.  I believe he will lose North Carolina due to my perception of Clinton’s organizational advantage.  I’m shaky on that prediction, but I’m going with it.

Further, I believe Clinton will win Florida due to early voting and organization.  Nevada should also be in Clinton’s column.

Therefore, Trump has 191 Romney votes, plus gains in Ohio (18) and Iowa (6), and I will throw in New Hampshire (4) for a total of 219 votes to Clinton’s 319.  My sense is that the FBI-induced fever that plagued Clinton over the past week broke over the weekend.  Her campaign’s inherent strengths and Trump’s weakness with non-white voters will be a deciding factor in close races.  It will take an uprising in states where there is a higher proportion of white voters to elect Trump, IMHO. I’m betting the surprise on election night will be the size of Hillary Clinton’s margin of electoral votes, not a Trump win.

On Election night, channel flip over to Global TV’s BC1 news channel.  I will be speaking to results with Global’s Keith Baldrey throughout the evening.

The Rosedeer Prediction Map:

Screen Shot 2016-11-08 at 12.01.14 AM.png

 

 

Trumping Clinton: 7 days of momentum

USC is running a rolling track poll where they interview 300-400 people a day (online) right through to Election Day.  This is a serious poll with serious methodology.  The numbers shown daily represent seven days of tracking. Each day, the daily results from 7 days ago drop off and the current day is added, making it a rolling track.  This smooths results and shows more of a trendline rather than sudden shifts.  So, if there is a big move, it might not become fully apparent for several days.

For the past 7 days, Donald Trump’s support has increased to, now, a 7 point lead.  This includes several days now of the Democratic National Convention.  Trump certainly had an RNC  Convention bounce but yet to see a Dem bounce.

Chart 1: Election forecast (n=2150)

Screen Shot 2016-07-28 at 10.58.20 AM.png

Some Democratic pundits have cautioned against “bedwetting”.  Yes, it’s July.  There’s a long way to go.

The issue the Democrats have to confront, however, is that Trump can win.  There has been a lot of commentary about how it’s impossible for Trump to win because of lack of support among Hispanics, Blacks, women, etc.  However, he is crushing it with whites and males.

 

Here is a breakdown of the numbers to show how Trump is rising:

Chart 2: Predicted Winner

While Hillary Clinton is still seen as the likely winner by 49% to 45%, that gap has narrowed from 13 points to 4 points in the past 17 days.  More Americans are believing in the possibility of a Trump presidency – will that help or hurt Trump?

Screen Shot 2016-07-28 at 9.31.21 AM.png

Chart 3: Intention to Vote

Trump and Clinton supporters are virtually tied when it comes to whether they intend to vote.  They have leapfrogged on this.  Trump’s turnout numbers are likely helped because he has strong support among older voters; Clinton’s turnout numbers are likely helped because Trump is highly polarizing and antagonizing.

Screen Shot 2016-07-28 at 9.33.17 AM.png

Chart 4: Seniors 

Trump has a big lead (55% to 38%), and seniors typically vote at a higher rate.  Trump leads 18-34s too, right now.

Screen Shot 2016-07-28 at 9.36.31 AM.png

Chart 5: Whites

Trump leads white Americans 57% to 31%.  African-American voters are 81% to 4% for Clinton.  Hispanics, though, are reported at 50% to 37% for Clinton.  This is where one might wonder if the poll has a large enough, or representative, sample of Hispanic voters.  Or maybe that’s reality – are gender and age are ‘trumping’ race among Hispanics?

Screen Shot 2016-07-28 at 9.38.08 AM.png

Chart 6: Men

Trump leads Clinton by 17 points among men (53% – 36%) while Clinton has a two-point lead among women.

Screen Shot 2016-07-28 at 10.47.19 AM.png

What does it all mean?

Trump can win.  If you can rack up a 7 point lead, you can obviously win.  Even if this poll is inaccurate, other polls are showing Trump is leading.  Even though Hillary has a small lead in Ohio, Trump has a small lead in Florida.

The challenge for Democrats is to approach the race for what it is – a very unconventional campaign.  Trump is attracting voters who are very anti-establishment including alienated Democrats.  How many more examples do we need to see – Rob Ford, Brexit, Jeremy Corbyn, Bernie Sanders, and Trump himself – to understand that there is a very large constituency for those who tap into the vein of frustration, resentment, and anxiety?

This rise in Trump support may be short-term.  It may be illusory.  It may be overstated.  But it proves that Clinton is no shoo-in.   The presidential campaign has been very unkind to her personal popularity and favourables.  Bernie Sanders did a lot to soften her support and drive votes away.  She has gone from a plus 10% to minus 17% in two years.  At her peak back in 2008, she had 69% favourable rating.

Chart 7: Hillary Clinton’s favourables over past two years.

Screen Shot 2016-07-28 at 10.54.39 AM.png

 

So, the first thing Democrats have to face is that they have a problem.  Now, deal with it.  If the DNC Convention does not move the dial, then it’s time for Plan B, whatever that is.

 

Interpretations of the Calgary-Greenway By-election

Yesterday’s provincial by-election in Calgary-Greenway has contradictory interpretations.

fork

Interpretation of Calgary-Greenway is not straight-forward

It was a four-way race ranging from 27.7% for the winning PC candidate to 20.2% for the fourth-place NDP.  It’s rare to see four candidates place above 20%.  I could not find one such example in the 2015 federal election, unless you count Nanaimo-Ladysmith where the four-way race had the 4th place Greens at 19.7%.

In Calgary-Greenway, when only 7.5% separates 1st and 4th, it’s hard to see it as Earth-shaking.  Nevertheless, the PCs won and a win is a win.  Therefore, interpretation #1 is that the PCs are alive, that they must still be reckoned with, and the NDP’s relegation to fourth is a sign of their demise.

Interpretation #2 is that the Centre-Left (NDP/Liberal) has made major gains in this riding since 2012 and further reduced the PC-Wildrose combined vote from 2015.  In 2012, the NDP-Liberal vote in this riding was a combined 15.5%; in yesterday’s by-election it was 42.8%.  It rose from 36.2% in 2015 and given that that was solely the NDP vote, one can see how the NDP benefit from no Liberal in the race.

Table 1: Popular vote of parties in 2012 General Election (GE), 2015 GE, and 2016 by-electionScreen Shot 2016-03-23 at 10.42.06 AM.png

As Table 1 shows, the NDP was actually five times higher than its 2012 vote and the Liberals have doubled from 2012.  There is a lot of talk about PC-Wildrose cooperation, but the centre-left should probably be viewed in the same way.  Not that there is an imminent merger, but there is a competition for like-minded voters.  The Liberals bothered to show up to the by-election after missing the 2015 GE and their impact was significant.  That may be in part a result of local candidate influence, but I’m not sure how many saw the Liberals competing to win the seat.

Table 2: Raw vote of parties in 2012 GE, 2015 GE, and 2016 by-electionScreen Shot 2016-03-23 at 10.42.15 AM.png

The 2012 GE and 2016 by-election are interesting comparisons because the overall number of voters is very similar.  It shows the overall reduction in votes for the PCs and Wildrose (centre-right) and the significant increase for the NDP and Liberals (centre-left).  Let me nail this point a little harder in Chart 1 below:

Chart 1: Combined popular vote of PC-WRP (blue) and NDP-Lib (red)Screen Shot 2016-03-23 at 11.02.12 AM.png

Unlike the recent BC by-elections in Coquitlam-Burke Mountain and Vancouver-Mt. Pleasant, turnout in Calgary-Greenway as a percentage of the previous election was relatively high, and as noted, about the same as the 2012 GE.  The by-election turnout was about two-thirds of the 2015 GE turnout while the BC by-elections were about 40% of the previous GE.  That indicates a higher interest and engagement in the outcome and its possible impact on the next election.

If I was a PC or Wildrose strategist, I would interpret this result with some nervousness.  The pool of centre-left voters in this by-election was almost evenly split.  The voter pool that existed in 2015 massively went toward Rachel Notley’s NDP.  This is basically the Justin Trudeau/federal NDP vote bloc.  The Stephen Harper vote bloc was much larger provincially, but is also split.  A consolidated centre-left offering (whereas those voters group behind one strong alternative) appears still able to defeat a split PC/Wildrose offering.

In reality, it is more complicated than described above.  Voters move around between parties with more fluidity – a Liberal may never consider NDP and a PC may never consider voting Wildrose, and vice versa.  But the by-election does show that the situation has become more, not less murkier as a result of Tuesday’s outcome.  It’s too early, much too early, to write off the Alberta NDP.

Finally, it must be noted that – probably for the first time in Canada – there was a competitive four-way race between four South Asian candidates.  This may well have created a dynamic that disrupted prevailing provincial political currents.  I’m not close to the ground so I defer to others and invite comments.  Regardless, the fact remains that the market share for the Centre-Left in this riding has increased sharply since 2012.

 

 

 

 

 

 

 

 

 

A deeper dive on BC By-election turnout

My post on by-election turnout earlier this week was picked up in the Vancouver Sun and Globe & Mail, proving that there’s a lot of interest in apathy.

In my earlier post, I stopped at 2004 in terms of comparing turnout to past BC by-elections.  Because I couldn’t help myself, I have now gone back to 1981 to see how engaged BC voters have been when it comes to mid-term voting.  If you’re not a political junkie, avert your eyes.  This is really geeky stuff.

obit-bill-bennett

Bill Bennett was successful in electing a government member in the 1981 Kamloops by-election.  It hadn’t happened since 1966 and it hasn’t happened since, unless the candidate’s name was Christy Clark.

This week, turnout in the two by-elections was less than 22% of eligible voters.  Put another way, the turnout was only 39% and 41% of the voters who voted in the 2013 general election.  In Coquitlam-Burke Mountain, less than 8,000 voted in the by-election compared to about 20,000 in 2013.My review of 15 by-elections between 1981 and 1999 shows a much higher turnout by comparison.

Only one by-election had less than half the voters of the previous general election – that was in Vancouver-Quilchena in 1994 when Gordon Campbell was elected to the Legislature for the first time.  He won in a cakewalk so arguably voters didn’t see a lot of reason to vote.

Table 1: Turnout in by-elections compared to preceding General Election.  Asterisks (**) indicate a two-member seat.  Until 1991, BC had some riding that had two members with voters having the choice to vote for two MLAs on their ballot.

Screen Shot 2016-02-04 at 9.38.09 PM.png

The other 14 by-elections saw a turnout of at least 63% of the previous general election.  Most of these races had a compelling storyline.

1981 Kamloops – The Socreds fought to hold their seat at a time when the margin in the 57(!) seat BC Legislature was razor-thin.  1984 Okanagan North – the then-unpopular Socreds succumbed to the NDP in a strong Socred seat.  The Socreds regained the seat in 1986.  Both of these by-elections had a turnout rate of about 80% of the previous election.

During the Vander Zalm years, there were six by-elections created by the departure of four Socreds and two NDPers.  The NDP took all six by-elections.  Turnout was high relative to current-day by-elections.  At the height of Vander Zalm’s unpopularity, almost as many people voted in by-elections in the Cariboo and Oak Bay as voted in the preceding general election.  They showed up with their pitchforks!

Turnout was at 65% to 70% of the previous election in three Fraser Valley by-elections between 1994 and 1997.  While the governing NDP did very poorly, they were not really at issue in these by-elections.  It was a brawl-for-it-all for the Free Enterprise mantle between the BC Liberals and the Socreds (1994), and the BC Liberals and Reform (1995 and 1997).  There was a lot at stake and voters, for the most part, turned out.

The Parksville-Qualicum 1998 by-election was more of a BC Liberal-NDP fight with Judith Reid eclipsing former (and future) MLA Leonard Krog.  This by-election was a verdict on Glen Clark at the time and had a turnout that was double of this week’s Coquitlam by-election.  Pitchfork time again.

In 1999, it was another BC Liberal-Reform fight when Bill Vander Zalm mounted his political comeback.  The BC Liberals rallied to win 2:1, with about twice the turnout of this week’s Coquitlam by-election.

The 1981-99 era had a lot of by-elections with high stakes and high turnout.  Even the low stakes by-elections had higher turnouts.  Why is it different now?  Perhaps voters are harder to get in a distracted world.  Media sources are more fragmented.  Voters are not herding toward BCTV (sorry Keith) like they used to, nor are they all opening the door in the morning to retrieve the morning Sun and Province like days gone by.  Fragmentation aside, the mainstream media seemed to great the by-elections with a collective yawn this time.  With less media interest, there is inevitably less chatter about politics at coffee-time and over the backyard fence and, ultimately, less turnout.

This all serves to reinforce my view in the earlier post that, unlike the Zalmy 1980s and Parksville in 1998, this week’s by-elections were not pitchfork-carrying affairs.  Low turnout indicates a lack of urgency and lack of consequence.  The turnout, and margin of loss, in Coquitlam-Burke Mountain was lower than in the neighbouring Port Moody-Coquitlam 2012 by-election.  The BC Liberals regained that seat, so there’s no reason to suggest they can’t do the same next year.  We’ll see.

 

 

 

Turnout goes underground in BC by-elections

Tuesday’s by-elections were remarkable for one key factor – rampant apathy.  It was fitting that the by-elections took place on Groundhog Day – the voters were underground.

apathyCompared to the turnout in the preceding general election, both Coquitlam-Burke Mountain and Vancouver Mount Pleasant had the lowest turnouts of any BC by-election since 2004.

In both ridings, less than half that voted in 2013 showed up to vote in the by-election.  In Coquitlam it was 39%, in Mt. Pleasant 41%.  That’s not 39% turnout, that’s 39% of the turnout of the last election so the overall turnout was barely 20%.

Compare that to turnouts in other by-elections, again, comparing to the previous election:

  • Vancouver Fairview (2008):  44%
  • Vancouver Burrard (2008): 49%
  • Port Moody-Coquitlam (2012): 60%
  • Vancouver Pt. Grey (2011): 70%
  • Surrey-Panorama (2004): 77%
  • Westside-Kelowna (2013): 77%
  • Chilliwack-Hope (2012): 85%

Voters, especially BC Liberal voters, just didn’t seem motivated to vote.  Why?  There was not a lot of media attention and there was not a lot at stake – the government is not going to fall.  It is not unlike the 2008 by-elections where there was not a lot driving the public debate and thus turnout was low.  I lived through the 2012 by-elections, which were critical in terms of clarifying the free enterprise option for the 2013 election.  As poorly as the BC Liberals fared relative to the preceding election, they fared well enough to survive another day and helped to ultimately resolve matters with many potential BC Conservative voters.

Say, what do Jenn McGinn, Gwen O’Mahoney, and Joe Trasolini have in common?  They were NDP MLAs elected in by-elections one year prior to a general election.  In all three cases, they were defeated in the general election.  When turnout returned to its usual levels, BC Liberals prevailed at the polls.  That’s not to assume it will happen in 2017 in Coquitlam-Burke Mountain, but it will be a much different fight with many more voters voting.

Let’s take a closer look at the by-election results in Coquitlam:

In 2013, the BC Liberals had 49.9% of the vote and the NDP 37.4% – a 12.5% spread.

In the by-election, The NDP won 3,562 votes (46%).  In 2013, their candidate received 7,315 votes, so the by-election candidate scored about half as many votes.

Of course, the BC Liberal dropped further, from 9,766 votes to 2,936 votes (38% of the popular vote).

Compared to 2012 by-election in the neighbouring Port Moody-Coquitlam riding, the BC Liberals did much better.

In 2012, the BC Liberals dropped from 52% to 30% while the NDP gained from 40% to 54%.  The BC Libs dropped 22% and the NDP gained 14%.  In this week’s by-election, the swing was less – the BC Libs dropped 11% and the NDP gained 9%.

As mentioned, the turnout in Mt. Pleasant was also very low with the NDP losing market share dropping about 5 points while the BC Liberals dropped further by about 8 points.  The Greens gained at both party’s expense.

What does it all mean?  By-elections are a tough place to turn out voters and BC’s history shows that Opposition parties usually prevail.  The last time a government MLA won a by-election – who was not the Premier of BC – was 1981 in Kamloops.

When stacked against the by-elections of 2012, the government fared much better in Coquitlam-Burke Mountain.

Congratulations to the NDP and the newly elected MLAs.  A win is a win, no matter how you slice it.  See you in 2017.

 

 

 

Conservatives held their vote but lost market share

Did the Conservatives crash?

No. They had 96% of the votes they received when they won a majority in 2011.  Had the number of voters stayed the same, they would have received 36% of the popular vote.

However, the market expanded.  An additional 2.7 million Canadians voted, pushing voter turnout from 61.1% to an estimated 68.5%.

Chart: Total votes by party from 2006 to 2015

2006 2015 votes

The blue line is remarkably consistent over four elections, ranging from a low of 5.2 million (2008 minority CPC government) to 5.8 million (2011 majority CPC).  The total number of CPC votes in 2015 was higher than minority governments in 2006 and 2008.

The absolute number of votes for the Greens has declined and levelled off.  The Bloc continues its decline in overall voters.  More seats this time, but that was thanks to a four-way fight in Quebec.

The NDP can take some solace that they had the second most votes in the history of their party, almost a million more than Jack Layton’s campaigns in 2006 and 2008, but over a million less than the breakthrough 2011 campaign.

Then there’s the Liberals.  After successive declines, 2015 blew the roof off of their support eclipsing the steady Conservatives.

On the surface, the 250% increase in Liberal support can be attributed to stealing market share from the NDP but moreso in terms of increased voter turnout overall.

The Chinese community and the federal election: did anyone ask what they think?

Co-authored by Tung Chan 陳志動, former CEO of S.U.C.C.E.S.S. 

It would be nice to know what Chinese-Canadians are thinking about the federal election.  It would also be nice if they had been being asked… properly.

In Metro Vancouver, over 430,000 Chinese-Canadians make up 19% of the region’s population, which is a conservative estimate since this is based on the 2011 census.  Across BC, over 1 in 9 are Chinese.

The concentration is higher in areas like Richmond, Vancouver, and Burnaby.

The impact of the Chinese-Canadian vote on a significant number of federal ridings is undeniable.

Riding Cantonese Mandarin Chinese NOS* Taiwanese Total
Richmond Centre 16.77% 12.22% 14.61% 0.71% 44.31%
Vancouver Kingsway 18.32% 3.22% 10.89% 0.07% 32.50%
Vancouver South 17.50% 3.63% 10.82% 0.28% 32.23%
Steveston-Richmond East 13.10% 8.22% 8.14% 0.21% 29.67%
Burnaby South 7.49% 10.84% 9.28% 0.68% 28.29%
Vancouver Granville 8.07% 7.60% 8.32% 0.67% 24.66%
Vancouver East 11.65% 1.65% 6.56% 0.05% 19.91%
Burnaby-North Van Seymour 7.62% 4.65% 5.83% 0.26% 18.36%
Vancouver Quadra 4.29% 6.89% 6.43% 0.42% 18.03%
Coquitlam-Port Coquitlam 4.70% 3.86% 4.04% 0.13% 12.73%
New West – Burnaby 3.29% 4.37% 3.80% 0.23% 11.69%

*Not otherwise specified (NOS)

Every day, new opinion polls are being reported by the media.  These polls only tell part of the story, because we have no way of knowing if they are talking to a proportionate share of Chinese-Canadians.

Why?

  • Telephone surveys – there is no indication that media polls are being conducted in-language for those that speak comfortably in Cantonese and Mandarin, but not English.
  • Online surveys – those that are not confident in English are not likely to participate in an online panel.
  • IVR surveys – automated messages for media polls are almost always in English.

A tell-tale that multicultural communities, such as Chinese-Canadians, are not being properly represented is that polls are not weighted to reflect these communities.  In other words, if 25% of a riding has a Chinese language as their mother tongue, the poll should have a sample of 25% Chinese.  This isn’t happening.

Yes, some Chinese Canadians will be participating in these surveys but, it is likely below their share of the electorate, for the reasons listed above.

A recent example are riding polls released by organizations like Lead Now and the Dogwood Initiative.  They did not release a breakdown of ethnicities in riding surveys conducted in Burnaby-North Vancouver Seymour (18% Chinese) or Vancouver South (32% Chinese).  If they have under-reported the Chinese Canadian voters in those ridings, they may well be providing voters with a misleading portrait.

We would love to be proven wrong, but it is clear to us that media polls (usually polling that is provided free of charge to news outlets or released into social media) has a cultural bias.  It simply costs more to do it right.

It is true that the voter turnout rate for Chinese-Canadian voters can be lower than BC average.  The provincial riding with the highest population of Chinese Canadians, Richmond Centre (49.88% of population), also had the lowest turnout rate (43.65%) suggesting a lower turnout rate from that community.

Voter participation increases as proficiency in English increases and as length of residency in Canada increases.  This is intuitive – as newcomers become more integrated into their community, they tend to participate more.  Even with lower voter turnout, the impact of Chinese Canadian voters cannot be ignored.

Though in the recent transit plebiscite, the voter turnout rate in Richmond was almost the same as the region-wide average.  Low turnout among Chinese voters may in fact be overstated.

So, does this even matter?

Chinese-Canadian voters, on the whole, tend to have different values than other groups.  The results of the 2010 HST referendum show this.  Only 25 of 85 provincial ridings supported the HST, with the strongest BC Liberal seats being among those that provided the most support.  Yet, BC Liberal strongholds in Richmond and South Vancouver voted overwhelmingly against the HST.  It was a major swing compared to other BC Liberal ridings with lower Chinese populations. Chinese Canadians surely made a critical difference; the HST had taken a beating in Chinese media and at the retail politics level.

The following table shows 7 BC Liberal ridings based on proportion of Chinese Canadian population (mother tongue).  While 24 of 49 BC Liberal constituencies voted in favour of the HST, only 1 in 7 of the ridings with the highest Chinese population supported the HST.  The exception being the seat of the Finance Minister.  The 2010 pro-HST vote and BC Liberal 2009 election vote were almost identical on a BC-wide basis.  But in these 7 ridings, all with a Chinese Canadian population of over 20%, the pro-HST vote runs behind the BCL vote significantly, with the highest Chinese ridings having the highest discrepancy.  Even Quilchena, which had a pro-HST vote of over 60%, ran behind its BC Liberal vote.

BC Liberal-held ridings Chinese % (mother tongue) Riding Pro-HST vote 2009 BCL vote Diff: HST-BCL
Richmond Centre 49.96% 36.23% 61.51% -25.28%
Richmond East 37.88% 34.42% 58.73% -24.31%
Vancouver Langara 35.41% 38.35% 58.87% -20.52%
Vancouver Fraserview 32.15% 33.99% 49.29% -15.30%
Richmond Steveston 31.88% 44.81% 60.78% -15.97%
Vancouver-Quilchena 26.98% 62.40% 70.22% -7.82%
Burnaby North 22.68% 39.66% 48.19% -8.53%
British Columbia 8.20% 45.27% 45.82% -0.0055

We want to make it clear that the issue we are raising is not solely a Chinese Canadian issue.  This is a South Asian issue, a Filipino issue, a Korean issue, a Persian issue.  For example, the 45 % of residents in the riding of Surrey-Newton say that their language at home is not one of Canada’s official languages, with the largest group speaking Punjabi.  Metro Vancouver has changed and will continue to do so.

In this election, the smart political parties are tracking opinion carefully so that they know what is actually going on.

Media outlets and any organization conducting research should be no different as when they fail to account for large segments of the population, they are ignoring them at their own peril.

Postscript:

A 2009 survey of Cantonese and Mandarin speakers for S.U.C.C.E.S.S. by Innovative Research Group  provided interesting insights into newspaper reading habits.  Even among those Chinese-Canadians fluent in English, Chinese media sources were preferred.

Why Conservatives have hope

Conservatives who pay attention to media polls are praying for a turnout advantage based on older voters and firmly committed supporters.  And supporters of other stripes will want to head into the weekend with their eyes wide open.  I’m sure complacency is not an issue for anyone.

I’m not sure this praying cat is Conservative, but who doesn’t like a praying cat?

Today’s Angus Reid Institute (ARI) – hint, older people more likely to vote:

Screen Shot 2015-10-16 at 2.11.29 PM

Then there’s this:

Screen Shot 2015-10-16 at 2.25.51 PM

ARI looks specifically at likely voters, shrinking Liberal lead from 35-31 to 34-33:

Screen Shot 2015-10-16 at 2.14.37 PM

Today’s EKOS (65 and over):

Screen Shot 2015-10-16 at 2.12.36 PM

Ekos has overall race at 34 Lib, 33 CPC.

Is there a Liberal surge?  Nanos has six point Liberal lead and Liberal strength among seniors.  Mainstreet is predicting a Liberal majority.  Innovative also has a big red spread.

Ekos and ARI provide counter-evidence to suggest the CPC are far from dead and buried.  Not in majority territory (I don’t think we have a David Cameron surprise here) but still in the hunt for a plurality.

Pick your poison.

Reasons to second guess polls

The latest polls give plenty of fodder to suggest that there is a Liberal surge overtaking the race.  There are a number of supportable points on this:

  • Nightly poll tracking by Nanos has been reinforced by Innovative and Ekos.  The narrative is that the Liberals are pulling away from the Conservatives with the NDP far behind.
  • The “not ready” line of attack has been embraced by the Liberals; they have played it differently than “just visiting” and “not a leader” attacks from previous campaigns.
  • The heat of the anti-Harper passion is far beyond that what has been seen in previous elections.  Various poll metrics (eg. “time for a change”) suggest voters are ready for a new government.
  • Advance voting turnout is high which could mean higher overall turnout favouring the Opposition.

What could possibly go wrong betting on a horse race?

Yet, there is that gnawing feeling that there could be another polling surprise just around the bend.

Look at the UK election last May.  Screen Shot 2015-10-12 at 11.08.00 PMThe intensity of British media coverage and polling exceeded that of the Canadian election yet no one saw a Conservative majority coming.  Even famed predictor Nate Silver blew it badly.  When BBC forecasted a majority moments after polls closed – based on results from exit polls – pundits were absolutely gobsmacked.  Not only that, two of the party leaders were caught with their pants down by their ankles and resigned by morning.  The prime minister (“Bluedini”) was likely as surprised but had the winning strategy on his side.  (Exit polls were based on interviews with voters immediately after they voted, not pre-voting surveys)

Yes, there is the litany of Canadian surprises too.  Mainly favouring incumbents – Christy Clark, Alison Redford, Greg Selinger, Dalton McGuinty, and Kathleen Wynne to name some plus Jean Charest who just missed re-election when pollsters had him in third.  Remember Doug Ford?  The final polls in the Toronto mayor’s race had him dead and buried but he only lost by 6.5%.  Overall, it’s certainly not a sterling track record.  I’m speaking about media polls here.  Sure, some parties have got it wrong too, but clearly some (the winners) are getting it right.

But they called it in Alberta, right?  A quick check from the 308 poll aggregator site shows that most pollsters (not all) overstated NDP support and understated PC support.  It didn’t matter since the NDP won handily.  But in a close election, some of the pollsters were off by a considerable margin when you look at the NDP-PC difference.  The poll aggregator had that gap at 20 points (it was 13%).  That’s a big difference in a close election and an error of similar magnitude in this election would lead to a different outcome.

Screen Shot 2015-10-12 at 10.23.41 PM

Three things I’m watching

  1.  The engagement level of voters

Last week, I wrote about Greg Lyle’s extensive research.  He updated his research this week with a research deck of over 100 pages.  There is enough data here to keep Wikileaks busy for a month.  Overall, his survey reported a 35-30-24 race (Lib – CPC – NDP).

Screen Shot 2015-10-12 at 10.30.49 PMGreg has built a respondent profile based on consistency versus ambivalence.  If the respondents have a consistent pattern to their responses, they are at the ‘consistent’ end of the scale.  If they answer “don’t know” to a lot of questions, they are on the ‘ambivalent’ end of the scale.  Conflicted respondents are in the middle.

About 15% of respondents are ambivalent.  Most will not vote.The ‘perfectly consistent’ are primed to vote.

This week, Greg posted the vote results by each of these clusters.  Bearing in mind the Liberals had an overall five-point lead on the CPC, here’s how that broke down among the consistency segmentation:

Screen Shot 2015-10-12 at 10.30.25 PM

The race is tied among the ‘perfectly consistent’.  I would throw out the 15% of ambivalent respondents, which shaves a fraction off the Liberal lead.  Among the 18% of ‘conflicted’ voters, the Liberals have a substantial lead over CPC, which appears to be based on NDP switchers (hence the fact they are ‘conflicted’). These respondents are a lot more likely to vote than ‘ambivalent’ but less likely than consistent voters.

Therefore, the Liberals have more work to do to mobilize this voter group in order to realize a five-point win.

         2.  Inconsistencies between pollsters in age and gender

There has traditionally been a gender split with the Conservatives doing better among men and the Liberals doing better among women.  Ekos, which had the Conservatives ahead two points, shows the Conservatives leading among women.

Screen Shot 2015-10-12 at 10.46.40 PM

Meanwhile, Nanos in the field at the same time, has the Liberals with an 11.6% lead among women.  That’s a pretty big gap between pollsters.

nanos female
Is Ekos overestimating CPC support through a blip in female support, or is Nanos overestimating Liberal support among males (below).  Nanos has shown a consistent Liberal uptick.

Screen Shot 2015-10-12 at 10.44.51 PM

Oh, and did you see the Ekos age split?  48-30 for CPC among seniors.  Nanos has the Liberals with a 1.5% lead among plus 60 voters.  Again, another big difference.  Who’s right?  I dunno.

        3.  Regional races

There is considerable variance in the regional horse race numbers, especially in Ontario and Quebec.

In Ontario, it is either a death-duel between the Blues and the Reds OR it’s the Reds walking away.

In Quebec, it is either a four-way collision or the NDP retain enough support to win the majority of seats, keeping it in national contention.  When you see numbers like this, the party above 30% can harvest a lot of seats.  The difference in Quebec between 27% and 32% could be 40 seats.  The latest poll findings are inconclusive, to say the least.

308 Leger Nanos IRG Forum ARI
NDP 30% 28% 33% 29% 25% 31%
LIB 27% 28% 29% 26% 29% 24%
CPC 19% 20% 14% 19% 22% 17%
BLOC 21% 23% 23% 22% 21% 27%

Finally, there are the usual warnings:

  • 10-second Tories (in BC, we called them 10-second Socreds).  Voters that decide in the final analysis to hold their nose and vote for the incumbent.
  • “Cranky won’t says” – the 6-8% of voters who won’t and don’t cooperate on surveys are a statistical wildcard.
  • The final weekend – voters have a sixth sense that is not entirely detectable.  The Redford win in 2012 manifested from an unease about Wildrose in the final week.  A combination of not-likely voters, strategic voters, and strange bedfellows changed the game.
  • Advance and special votes – Perhaps up to 20% of votes are already in the bag.  Who do they favour? Will the winner on October 19th lose the overall election?

I’m not forecasting anything here – only caution.  It is clear that the Liberals have won the campaign thus far.  They started with low expectations and have exceeded them, and have to this point eclipsed the NDP in the ‘primary’ that established which party had the best chance to defeat the Conservatives.

Having said that, if I was a Liberal strategist, I would be tempering my grassroots’ naturally-occurring public-poll-based-optimism with Eeyore-like gloom and insist they are still running from behind.  I wouldn’t want to be like the gobsmacked Brit who couldn’t read the tea leaves – even at tea time.  If I was a CPC strategist, I wouldn’t assume the poll numbers will necessarily improve – it is going to be a gruelling week but a plurality is very much possible if they have a strong finish, particularly with the likeliest of voters.  If I was an NDP strategist, I would move mountains to move vote in Quebec.  If they lose Quebec, all is lost.

Ultimately, the great thing about campaigns is that it’s up to the voters.  The strategic voting organizations, the media outlets, and the pundits are not inside the voting booth.  It’s between the voters and the names on the ballot.  And that is the greatest variable – voters just damn well choose who they want to, sometimes with surprising results.

Wakey wakey. This election may be baked by Monday.

Wakey wakey.  It’s almost 10/09… time to vote.  Advance polls open October 9-12.

More and more voters are setting their alarm clocks to vote in advance polls.

In BC’s 2013 election, advance voting rose to over 20% of all votes cast.  A sharp rise over five successive elections when only 5.74% voted in advance (1996).

Screen Shot 2015-10-07 at 9.31.03 PM

In Alberta, advance voting jumped from 14% to 16% from 2012 to 2015.  Why are people voting earlier?  Convenience? Increased efforts by political parties to lock up the votes?

It’s kind of a thing with older people.  Elections Canada also reports a rising trend of advance voting between 2004-2011.   Those keenest to vote early are age groups that are the likeliest to vote.  Federally, 5% of 18-24s voted early compared to 17% of 65-74s.  After that morning coffee at Tim Horton’s, why not round up the pals and go vote?

Screen Shot 2015-10-07 at 9.30.42 PM

So, older people are keen to vote early and they are keen to vote often.  My estimate – back of envelope – is that over three-fifths of advance voters are beyond age 55.

An interesting strategic issue is at play.  If advance voters are disproportionately older – even moreso than overall turnout stats – then the voting population on Election Day (October 19) will be more balanced by age.

The voters heading to the advance voting stations this weekend – once they vote – they are done.   The advertising and media that bombards them in next 24-96 hours will have done its work – one way or another – and those votes will be locked up.  Maybe it will be 20% of the electorate, like BC, but whatever the amount, it will be well over a million voters.

Of the remaining voters, younger voters will be more important proportionally, so will messaging be tweaked?  We’ll have to see.

Political parties are more sophisticated than ever in mobilizing voters.  Social media will play an increasing role this time.  Canadian political parties are learning from mobilization techniques that have proven highly successful south of the 49th where early voting is an even bigger thing.

The recent polls are unclear who’s really winning this election.  Is it a clear Conservative lead as some suggest or a tight Liberal lead as others purport?  And what about Quebec voters – will they show up en masse for advance polls or wait and see?  The Conservatives will look to put a stranglehold on the race by encouraging the Grey March to the polls.  The Liberals will try compete among seniors in the advance poll, thereby undermining what should be a Conservative edge.  Either way, the votes cast by turkey time will be a huge advantage for the party with momentum heading into a long, political weekend.