How Maritimers (Especially Nova Scotians) Pick Political Leaders

I wanted more information about political leadership conventions in Nova Scotia and my googling kept leading me to the same door: that of retired Acadia University political science professor Ian Stewart.

Stewart and a colleague, University of Calgary political science professor David Stewart, have literally written the book on the subject: it’s called Conventional Choices? Maritime Leadership Politics, 1971–2003, and if you’re a political junkie like me, you will love it.

The book is based on post-leadership-convention surveys of participants in Progressive Conservative, Liberal and New Democratic Party leadership conventions in Nova Scotia, New Brunswick and Prince Edward Island between 1971 and 2003. The surveys, conducted by the Acadia Poli-Sci department, have produced:

…a rich data set spanning twenty-five conventions over thirty-two years…[T]he data set, while extensive, is not complete; data collection has been most comprehensive in Nova Scotia. In fact, from John Buchanan’s second ballot victory in 1971 through to 2003, only one of that province’s leadership contests (the 1997 Liberal gathering) is absent from the data set.

The book, published in 2007, does not cover post-2003 leadership conventions, but in the case of the Nova Scotia PCs, there have only been two: the 2010 convention at which Jamie Baillie (remember him?) won by acclamation and the 2006 convention from which Rodney MacDonald emerged victorious. This last leadership contest, I’m happy to report, was the subject of a separate paper by the same authors, entitled “Ahead to the Past: The Return of the Delegated Leadership Convention,” so it’s actually correct to say the data for PC leadership races stretches from 1971 to 2006.

I thought it would be interesting to evaluate CBRM Mayor Cecil Clarke’s chances based on what I’ve learned about leadership conventions. But first, let me tell you what I’ve learned about leadership conventions.


The Rules

The provincial Tories are borrowing the voting system used by the federal Conservatives in their 2017 leadership race.

The system is open ballot, meaning anyone who is a member in good standing of the party as of September 11 can cast a vote, either by mail or in person at the convention, which will be held in Halifax on October 26-27.

But it is also weighted by constituency — that is, each of the province’s 51 constituencies will be worth 100 points, those points will be awarded to candidates based on the percentage of the constituency vote they win, and the first candidate to win a majority of the points (2,551) will win the leadership.

Voters can mail in ballots that rank their candidates in order of preference or they can vote in multiple ballots at the convention.

Parties don’t choose voting systems by accident, so it’s worth taking a moment to consider why the PCs have opted for this particular system — and why they’ve decided to return to an open ballot (albeit one weighted by constituency) after having held a delegated convention in 2006.


Open Ballot

Former Nova Scotia Premier John Hamm

Former Nova Scotia Premier John Hamm

Open ballot or universal ballot or all-member votes are arguably the most “democratic” way to choose a leader, as each vote is given equal weight. The problem, according to the Professors Stewart, is that:

…a riding association with 1,000 members has 10 times the voting influence of one with only 100 members…Such a circumstance contains obvious pitfalls for any party which must subsequently compete in general elections in which every constituency is of equal importance.

The Stewarts suggest that in 2006, when they opted to return to a delegated convention (after having chosen John Hamm by universal ballot in 1995), the Tories were:

…clearly mindful of the recent leadership travails of the Nova Scotia Liberal Party. On three occasions since John Hamm had been elevated to the Conservative leadership, their Liberal opponents had employed the universal ballot to select their new head. In every instance, the influence of Cape Breton members had been disproportionately high [emphasis mine] and without exception, the Liberal Party’s experiences in subsequent general elections had been disappointing.

Just to jog your memories, those three Liberal contests had looked like this:

1997: Russell MacLellan
MacLellan replaced John Savage in ’97, won a minority government in ’98, lost a confidence vote in ’99 and then lost the subsequent elections. He resigned as leader in 2000.

2002: Danny Graham
Graham, who didn’t have a seat in the legislature at the time, won the leadership on the first ballot in 2002. A propos our discussion about open ballots, a  Chronicle-Herald post-convention report said:

Dennis James, a Truro lawyer who pulled out of the race when he saw most of the party’s new members were in industrial Cape Breton and figured he couldn’t win, said he was feeling “less tender” now that it’s over.

Graham won a seat in Halifax Citadel in the 2003 elections which saw his party finish third, behind the Tories and the NDP. He stepped down as leader in 2004, due to his wife’s illness.

2004: Francis MacKenzie
MacKenzie, who had come in second to Danny Graham in 2002, won the race to replace him in 2004, although he didn’t have a seat in the legislature. He attempted to win one in the 2006 provincial elections but was not successful. The Liberals were reduced to nine seats and MacKenzie resigned as leader shortly after the election. Michel Samson stepped in as interim leader.

Although we’re focusing on the provincial PCs, it’s surely worth noting that when the Liberals replaced MacKenzie in 2007, they held a delegated convention and elected a non-Cape Bretoner, Stephen McNeil, who then went on to win back-to-back majority governments.


The Neighborhood Effect

Although their 1995 open-ballot leadership contest had not produced a Cape Breton leader, and although most Tories (including outgoing Premier John Hamm) wanted another open-ballot contest, the party executive opted to return to a delegated convention in 2006. Ironically, the result was — a Cape Breton leader!

Yes, Inverness MLA Rodney MacDonald, who “had a grab-bag of ministerial responsibilities in the final Hamm administration: Tourism and Culture, Immigration, and Health Promotion.” MacDonald defeated Bill Black, a Halifax businessman who had never held political office, and Neil LeBlanc, the CAO of Argyle municipality, who had spent 14 years in the provincial legislature and served as Minister of Finance in the Hamm government.

The map shows the "neighborhoods" of each of the candidates (to date) for the provincial PC leadership. Elizabeth Smith-McCrossin is the MLA for Cumberland North, Tim Houston is the MLA for Pictou East, Cecil Clarke is mayor of the CBRM (and a resident of Sydney Mines, in the riding of Northside Westmount); John Lohr is MLA for Kings North and Julie Chaisson ran previously in Chester-St. Margarets.

The map shows the “neighborhoods” of each of the candidates (to date) for the provincial PC leadership. The three sitting MLAs (constituencies marked in royal blue) are Elizabeth Smith-McCrossin of Cumberland North, Tim Houston of Pictou East and John Lohr of Kings North. The unseated candidates (“neighborhoods” marked in lighter blue) are Cecil Clarke, mayor of the CBRM and resident of Northside Westmount; and Julie Chaisson, executive director of the Halifax Seaport Farmers’ Market, who ran previously in Chester-St. Margarets.

How did MacDonald do it?

Well, according to the Stewarts, it had a lot to do with what political scientists call the “neighborhood” effect: people’s tendency to vote for their neighbors — that is, for candidates who come from their region or riding.

Voters in Maritime leadership conventions are particularly susceptible to the neighborhood effect. Of the 75 candidates in the 25 leadership conventions studied by the Stewarts, all but 10 benefited from a “neighborhood boost.” (And when the authors looked more closely at those 10 candidates, it often emerged that they had been pitted against a rival from the same neighborhood.) That boost, on average, meant their home vote was 19 percentage points higher than their vote elsewhere. (Although it could be much higher — the highest recorded neighborhood boost in the data was that of Helen MacDonald, the former Cape Breton the Lakes MLA, who won the NDP leadership in 2000. Her home vote was 68 percentage points above her vote elsewhere.)

Although the PCs’ return to a delegated convention was intended to force candidates to seek backing from across the province, it didn’t eliminate “regional variations in delegate support.” Each of the three candidates in the 2006 PC leadership — MacDonald, Black and LeBlanc — received “disproportionate support from the delegates from their area” but:

The loyalty of Cape Breton is particularly noteworthy. MacDonald won the support of every delegate from his home county of Inverness and his support on the rest of the island was also strong. More than 80% of Cape Breton delegates opted for MacDonald on the first ballot.

That loyalty, say the authors:

…played a major role in MacDonald’s victory. Absent his Cape Breton vote, MacDonald would have finished third on first ballot resulting almost certainly in a LeBlanc victory.

That said, the authors also note that when delegates were asked, post-convention, to rank the candidates in order of preference, MacDonald was the second choice of both Black and LeBlanc voters. MacDonald beat Black on the second ballot and would very likely have beaten LeBlanc, had the ex-finance minister been his opponent. The Cape Bretoner, say the authors, was “the convention’s overwhelming choice.”

He became premier, called elections that same year and won a minority government. In 2009, after losing a confidence vote, he lost the subsequent provincial elections to the NDP under Darrell Dexter. MacDonald resigned as party leader in 2009 and as MLA later that same year.


Mixed system

The system the Tories will use in 2018, then, is an attempt to combine the best of both worlds: hold a universal ballot, so that every member has a chance to participate, but weight the vote by constituency to encourage candidates to build a broad base of support.

As PC leadership committee co-chair Chris D’Entremont, MLA for Argyle-Barrington, explained, the system:

 …creates an “equality of ridings” that is meant to give candidates an incentive to build support for the party in every part of Nova Scotia.

“Ultimately, we’re going to want to go out in those constituencies and build them. I mean, we need to have members across this province in order to make all of this work,” he told reporters.

If I’m reading the Stewarts’ analysis correctly, they would consider this system to be more democratic than delegate systems generally — and certainly more so than the delegate system employed by the Liberals in 2007. Party brass argued that “every member” was able to participate in that leadership race because every member was permitted to vote for the delegates, but as the Stewarts point out, those delegates, under the provincial Liberal system, “were not bound to vote for a particular candidate on the first ballot.”

In fact, the authors actually suggest that if the Liberals’ goal in 2007 was to reduce the influence of Cape Breton, they “could have used a universal ballot system that treated constituencies equally regardless of membership size.”

In other words, the system the Tories are using this time around.

Implications for Cecil Clarke:

I am not going to make predictions. (Cape Breton has psychics for that. I know because I saw one on the front page of the Cape Breton Post last week.)

That said, I don’t think it takes a psychic to predict that Clarke will benefit from the neighborhood effect — particularly if he is the only Cape Breton candidate in the race. (Which he may or may not be, as candidates have until August 13 to submit their nomination papers and a lot can happen in six months.)

But even if he enjoys Rodney MacDonald levels of support — say he wins 100 points in his own riding and 80 points in each of the other six Cape Breton ridings — that would give him only 580 points. He’d need another 1,971 to win, which would be (roughly) 45 points out of 100 in every other riding in the province.

So the big implication for Clarke — and therefore for the CBRM, which he claims he can continue to lead while campaigning for the PC leadership — is that he will have to spend a lot of time on the mainland between now and October 27. He said as much to the CBC following the announcement of the rules:

“I think it’s going to mean we’re going to have a very robust, exciting and engaged leadership [process],” said Clarke.



Here’s another interesting factor, given there are two women in this Tory leadership race (bringing the total number of women who have ever run for the Nova Scotia provincial Tory leadership to two): when there are women in the race, Maritime women tend to vote for them.

Alexa McDonough (CBC photo)

Alexa McDonough (CBC photo)

The Professors Stewart reach this conclusion very cautiously: given that of the 25 leadership contests they studied, only six featured female candidates, the evidence is obviously not overwhelming, but it’s there.

In all six contests involving women, the female candidates received a disproportionate share of the female vote. In three of these cases, however, the “sex cleavage” (which is what the authors insist on calling this tendency for voters to vote along gender lines but which I hope to avoid for the rest of this section) was statistically invalid. However, the Stewarts suggest this was because the prospects of the women in question were either considered hopeless (Deborah Kelly-Hawkes) or unassailable (Catherine Callbeck).

YearJurisdictionPartyCandidateVote % FemaleVote % MaleGapRatio of male to female voted/sigma d*
First ballot1980NSNDPAlexa McDonough87.176.310.81.142.06
1993PEILiberalCatherine Callbeck89.
1996NSNDPYvonne Atwell34.810.324.53.374.01
1997NBPCMargaret Ann Blaney11.
2000NSNDPHelen MacDonald34.929.
Maureen MacDonald 33.618.714.9
2002PEINDPDeborah Kelly-Hawkes4.304.31.800.79
Second ballot2000NSNDPHelen MacDonald34.229.54.7NA2.46
Maureen MacDonald35.626.98.7
Third ballot2000NSNDPHelen MacDonald66.046.519.51.163.49

*any d/sigma over 1.96 is statistically significant

In the three contests where the female aspirants were considered viable candidates and there was real competition for the leadership (and all three involve the Nova Scotia New Democrats), the support of women for women was clear although, as the authors explain, other factors were also at play:

Of course, factors other than the competitive context sensitized Nova Scotia New Democrats to the importance of gender at these three conventions. In 1980, for example, Alexa McDonough was bidding to become the first female party leader in Canada, and media coverage of the event unconsciously reflected this fact. In one account of the three candidates, Len Arsenault was characterized as a “teacher,” Buddy MacEachern as “outspoken,” and Alexa McDonough as “attractive.” Indeed, one columnist enthused that McDonough was “pretty enough to catch the eye of any man at 100 yards. Sixteen years later…New Democratic delegates were jolted by Yvonne Atwell’s allegation that she had been victimized during the campaign by “sexism.” And, at the 2000 convention, the gender conflicts were rendered explicit after the second ballot, when departing candidate Maureen MacDonald chose to crown a queen rather than a king, despite the fact that fellow MLA Kevin Deveaux had overtaken Helen MacDonald on the second ballot. Helen MacDonald secured the party leadership on the third ballot.

What, I wonder, might “sensitize” Progressive Conservatives to the “importance of gender” in 2018? The Weinstein affair? The #metoo movement? The fact that their last leader was hooked ignominiously offstage before he could take his final bow over a sexual harassment charge even PC members still know nothing about? I’m thinking, all of the above.

Flora MacDonald

Flora MacDonald

Implications for Cecil Clarke:

I guess it will depend on how many women vote in this race, but the presence (to date) of two women on the ballot is certainly interesting. Especially when you remember that although the provincial PCs haven’t previously had a female leadership candidate, the federal PCs produced the first female leadership candidate in Canadian history — Flora MacDonald — as well as the first female Canadian prime minister — Kim Campbell. (Let’s not dwell too much on her fate.)

Throw in the current zeitgeist and it doesn’t seem crazy to suggest the “gender factor” could play a role in this race.

Oh good grief. I can’t believe I just wrote that. I really shouldn’t indulge in this pseudo-punditry. I’d be better off making flat-out predictions. Let me try to be a little more direct: being a woman could be an advantage in this race.



Cecil Clarke has declared himself a Christian publicly, most famously during his hallway, pre-council-meeting prayer session to protest the Supreme Court’s ruling against in-Chambers, pre-council prayer.

Elizabeth Smith-McCrossin identifies herself as a “devoted Christian.”

John Lohr is a “moderator of the New Minas Baptist Church” where he has held various positions including “Sunday School teacher, deacon and secretary.”

Julie Chaisson’s Facebook bio for her 2017 campaign in Chester-St. Margarets notes that she has “invested in people through years of volunteering with the local church, community outreach organizations.”

Tim Houston, alone of the five candidates to date, makes no mention of religion in his bio.

The Stewarts use the term the “friendship” effect to describe support for a candidate based on religion or ethnicity. Religion, they say:

…has historically been important in Maritime politics…the Protestant/Catholic division in Nova Scotia was recognized by a Liberal party tradition of alternating their leaders — a tradition that persisted until the first universal ballot in 1992.

The data show “significant divisions on the basis of religion” in 12 of the 25 conventions studied. (Interestingly, the data also show that “religion does not seem to be as important” for the party which, in Nova Scotia, is currently led by a United Church minister, the NDP.)

Of the 75 candidates across the Maritimes considered, 45% were Roman Catholic, 32% were Protestant and 23% could not be identified according to faith. Catholics accounted for 60% of the Liberal candidates, Protestants accounted for 52% of the Conservative candidates, while the authors “could not identify the religious background of 80% of the NDP candidates.”

And here’s an interesting stat: in terms of victorious candidates, five of the seven PC winners were Protestants (and eight of the 13 Liberal winners were Catholics).

But in the case of the Nova Scotia Progressive Conservatives, at least, those wins didn’t necessarily depend on votes from fellow Protestants. The authors say data from the 1991 and 1995 NS PC leadership races did not reveal “much in the way of religiously based support.” When John Buchanan won in 1971, two of the four candidates were Roman Catholic and two were Protestant. In 1995, “the only Catholic candidate, Michael MacDonald, actually performed marginally better among Protestant delegates.”  (In that 1991 contest, Rollie Thornhill, a Protestant, did well in the province’s “most heavily Catholic region,” Cape Breton, but the authors say that was probably because he promised not to privatize Sydney Steel — a promise his Catholic opponents didn’t make.)

That four of five PC candidates in 2018 publicly identify as religious suggests they certainly see it as a factor. The Stewarts, after pointing out that Atlantic Canada in general is more religious than the rest of Canada (we’re apparently the “Bible Belt” of the country), sum it up this way:

Religious affiliation was not always associated with voting, but it seems that knowledge of a candidate’s religious background may provide some sort of common identification and, thus, enable leadership voters to be aware of whether or not they share a particular tie with a candidate. Given the strength of religion in Maritime society, it would be surprising if this did not occasionally affect voting.

Implications for Cecil Clarke:

Publicly identifying as a Christian (a Protestant, to be more specific) won’t hurt him and could help him in the leadership race but I’d be curious to know what role religion plays in actual provincial elections. I guess I’ll have to find another book.


The Rural/Urban Thing

I’d heard that the Maritimes tended to be more rural than the rest of the country but I’d never paid much attention to the stats (and as a Cape Bretoner, I always secretly believe that 95% of the province lives in the Halifax Regional Municipality anyway), so it was quite eye opening to look at the actual numbers.

The numbers in the book are from the 2006 census, so I thought I’d be clever and consult the 2016 census to determine how much the situation has changed over the past decade, but it turns out that what has changed is Stats Canada’s use of the word “urban” — as in, they don’t use it anymore. They now divide the population into those living in rural areas and those living in small, medium or large population centers. (The change makes sense — previously, anyone living in a community of at least 1,000 people was considered urban, a definition the Professors Stewart acknowledge as “intuitively implausible.”)

Until I can sort out the new definitions, I’m going to compromise by looking at the 2011 census data, which still used the categories “rural” and “urban.”

In 2011, even under Stats Can’s generous definition of “urban,” only 57% of Nova Scotians were considered urban dwellers. That’s practically unchanged from 56% in 2006.

PopulationUrbanRuralUrban (% of Population)Rural (% of Population)
Nova Scotia921,727521,338400,3895743
Newfoundland and Labrador514,536305,566208,9705941
New Brunswick751,171394,479356,6925348
Prince Edward Island140,20465,54374,6614753
British Columbia4,400,0573,790,694609,3638614
Northwest Territories (including Nunavet)73,36839,93833,4305446

Compare our numbers to the national averages and the contrast is striking — nationwide, only 19% of the population is considered “rural.”

What’s more, the percentage of “urban” dwellers in Nova Scotia has barely changed in 30 years — the 1971 census put it at 58%. As the professors say (I’m going to be calling them by their first names by the time I’m through, we’ve spent so much time together):

These large and consistent numbers suggest that rural voters are likely to be influential in the region’s politics; and, indeed, this has been a recurring theme in the thirty-two-year period covered by our study.

They found “significant voting divisions” in 14 of the 25 data sets. Unfortunately, only one of those data sets refers to a Nova Scotia PC leadership race — the 1995 contest won by John Hamm who, “outside of his home in Pictou County” was “far more popular with Nova Scotians who were not from rural areas.”

They also found the neighborhood effect was stronger for candidates from rural areas — where “metropolitan” candidates enjoyed support almost 10 percentage points higher among their neighbors, rural candidates got a 26 percentage point boost from their neighbors.

The authors conclude:

In the Maritimes, it is quite possible to win elections without carrying the major urban areas. The 2004 Nova Scotia election highlights this point, as John Hamm and the Conservatives were re-elected with a solid minority government, despite gathering only 30 percent of the votes in metro Halifax and winning just three of the seventeen seats there. Indeed, the fact that Halifax, by far the largest city in the Maritimes, contains just 33 percent of the seats in the Nova Scotia legislature illustrates the electoral dominance of rural areas.


Implications for Cecil Clarke:

Clarke is from Sydney Mines which has a population of 12,823 people, making it a “small population center,” so I’m thinking he could get away with claiming to be either urban or rural (of the three points he mentioned in his announcement speech — healthcare, education and high-speed internet and cellphone service — only the last addresses a particularly rural problem, the first two apply province-wide).

That said, John Lohr, a “lifelong farmer” from the Valley might have more rural appeal.

Bottom line: I don’t know what the implications are, but you are all smart enough to draw your own conclusions, I just thought I’d give you some interesting material to work with.



You have no idea how hard it’s been to stay (mostly) focused on the Progressive Conservative Party of Nova Scotia for this entire article. Conventional Choices is full of interesting information about the other Maritime provinces (PEI, for example, can hold an all-member vote but expect all members to show up in person for it because no community is more than two hours’ drive from Charlottetown). Not to mention information about the other political parties (most fascinatingly the Nova Scotia NDP, which is so distinct in so many ways — beginning with the fact that Nova Scotia is the only Maritime province in which it is actually a presence).

There are fascinating stories about leadership conventions past: like the 1991 PC Convention in which Rollie Thornhill lost to Donald Cameron on the third ballot only to be arrested a few days after the convention on charges of influence peddling.

A Cecil Clarke supporter hands out fleece tartan scarves that say, "Cecil Clarke: Ready to Lead." (Photo by Jennifer Henderson)

A Cecil Clarke supporter hands out fleece tartan scarves that say, “Cecil Clarke: Ready to Lead.” (Photo by Jennifer Henderson)

And don’t get me started about the expulsion of Paul MacEwan from the Nova Scotia New Democratic Party and the shift in its support base from Cape Breton to Halifax (Professor Ian Stewart is working on a biography of MacEwan which I look forward to reading.)

Then there was the Liberals’ 1997 experiment with “tele-democracy,” i.e. allowing people to vote from their touch-tone phones.

And the PEI liberal leadership candidate whose campaign was so “fervent, agrarian, anti-bureaucrat, anti-intellectual” that some likened it to “a low-key Island version of the Khmer Rouge.” (!)

No, I’ve tied myself to the mast and resisted these many siren songs to focus on factors I felt might be relevant to the 2018 PC leadership race.

So now I’m going to indulge myself with one little tangent, because it isn’t entirely without relevance to 2018: I found another sitting mayor who ran for the leadership of a provincial party.

His name was Ian “Tex” MacDonald, and he was a teacher, a PEI “sports icon” and the mayor of Charlottetown. He had also, until a year before the 1996 Liberal leadership convention to replace Catherine Callbeck, been a Tory.

MacDonald ran an enthusiastic but troubled campaign, with recurrent allegations that his team was using the incentive of free hats to recruit instant-Liberals from local high schools.

MacDonald had hoped to be a king-maker on the second ballot, but his aspirations:

…vanished when he received only about one-fifth of his anticipated total of one thousand votes. As Hillsborough MP George Proud observed: “The Roman Catholic Church accepts converts every day of the week but they don’t appoint them cardinal the next day.”

He ran unsuccessfully for the Liberal party in 1996 and 2000 and then, in 2010, was appointed executive assistant to Liberal Premier Robert Ghiz.

Implications for Cecil Clarke:

I really don’t think there are any — Clarke’s problem is not inadequate loyalty to the Tory party. And while he was handing out free fleece scarves at the Tory AGM and encouraging people who wanted him to step down as mayor to take out PC memberships and vote for him, I can’t see either leading to trouble.

But enough out of me. What do you think?