Cullen Narrowly Won Convention, But Mulcair Victory Already Assured
[Welcome, National Newswatch readers!]
Nathan Cullen narrowly won all of the real-time voting he appeared on during last weekend's NDP leadership contest, but his rival Thomas Mulcair's victory was already a foregone conclusion by the time the convention started, detailed vote breakdowns show.
Unfortunately for Cullen, the convention-day round-by-round voting never accounted for more than 17.5% of the total ballots counted, and he had too big a gap to catch Mulcair in the preferential ballots cast in advance.
Brian Topp and Mulcair dominated the mail-in ballot segment of the advanced voting, while the other candidates showed better in the online preferential voting that concluded on the eve of the Toronto convention, including a notably better vote-share for Martin Singh.
First Ballot
[Click on image to open full-sized version]
First Ballot Results by Voting Method, NDP Leadership Contest, March 24, 2012
| Candidate | Advance | Conv. Day | TOTAL 1st B |
||
|---|---|---|---|---|---|
| Mail-in | Electronic | Adv Total | Real-time | ||
| TOTALS | 15,197 (23.2%) |
39,690 (60.7%) |
54,887 (84.0%) |
10,487 (16.0%) |
65,374 (100.0%) |
| Thomas MULCAIR |
5,301 (34.9%) |
12,231 (30.8%) |
17,532 (31.9%) |
2,196 (20.9%) |
19,728 (30.2%) |
| Brian TOPP |
4,027 (26.5%) |
7,914 (19.9%) |
11,941 (21.8%) |
1,974 (18.8%) |
13,915 (21.3%) |
| Nathan CULLEN |
1,545 (10.2%) |
6,818 (17.2%) |
8,363 (15.2%) |
2,308 (22.0%) |
10,671 (16.3%) |
| Peggy NASH |
1,597 (10.5%) |
5,034 (12.7%) |
6,631 (12.1%) |
1,722 (16.4%) |
8,353 (12.8%) |
| Paul DEWAR |
1,415 (9.3%) |
2,527 (6.4%) |
3,942 (7.2%) |
0,941 (9.0%) |
4,883 (7.5%) |
| Martin SINGH |
0,364 (2.4%) |
3,185 (8.0%) |
3,549 (6.5%) |
0,272 (2.6%) |
3,821 (5.8%) |
| Niki ASHTON |
0,916 (6.0%) |
1,794 (4.5%) |
2,710 (4.9%) |
1,027 (9.8%) |
3,737 (5.7%) |
| Romeo SAGANASH |
0,032 (0.2%) |
0,187 (0.5%) |
0,219 (0.4%) |
0,047 (0.4%) |
266 (0.4%) |
[Click on image to open full-sized version]
The same pattern repeated itself across subsequent ballots.
Second Ballot
[Click on image to open full-sized version]
Second Ballot Results by Voting Method, NDP Leadership Contest, March 24, 2012
| Candidate | Advance | Conv. Day | TOTAL 2nd B |
||
|---|---|---|---|---|---|
| Mail-in | Electronic | Adv Total | Real-time | ||
| TOTALS | 14,553 (23.3%) |
38,538 (61.7%) |
53,091 (85.0%) |
9,403 (15.0%) |
62,494 (100.0%) |
| Unassigned | 644 | 1,152 | 1,796 | 1,796 | |
| Thomas MULCAIR |
5,890 (40.5%) |
15,430 (40.0%) |
21,320 (40.2%) |
2,582 (27.5%) |
23,902 (38.2%) |
| Brian TOPP |
4,540 (31.2%) |
8,892 (23.1%) |
13,432 (25.3%) |
2,192 (23.3%) |
15,624 (25.0%) |
| Nathan CULLEN |
1,877 (12.9%) |
7,657 (19.9%) |
9,534 (18.0%) |
2,915 (31.0%) |
12,449 (19.9%) |
| Peggy NASH |
2,246 (15.4%) |
6,559 (17.0%) |
8,805 (16.6%) |
1,714 (18.2%) |
10,519 (16.8%) |
[Click on image to open full-sized version]
Third Ballot
[Click on image to open full-sized version]
Third Ballot Results by Voting Method, NDP Leadership Contest, March 24, 2012
| Candidate | Advance | Conv. Day | TOTAL 3rd B |
||
|---|---|---|---|---|---|
| Mail-in | Electronic | Adv Total | Real-time | ||
| TOTALS | 14,183 (22.6%) |
37,631 (60.0%) |
51,814 (82.6%) |
10,922 (17.4%) |
62,736 (100.0%) |
| Unassigned | 1,014 | 2,059 | 3,073 | 3,073 | |
| Thomas MULCAIR |
6,596 (46.5%) |
17,141 (45.6%) |
23,737 (45.8%) |
3,751 (34.3%) |
27,488 (43.8%) |
| Brian TOPP |
5,256 (37.1%) |
11,234 (29.9%) |
16,490 (31.8%) |
3,332 (30.5%) |
19,822 (31.6%) |
| Nathan CULLEN |
2,331 (16.4%) |
9,256 (24.6%) |
11,587 (22.4%) |
3,839 (35.1%) |
15,426 (24.6%) |
[Click on image to open full-sized version]
Fourth Ballot
[Click on image to open full-sized version]
Fourth Ballot Results by Voting Method, NDP Leadership Contest, March 24, 2012
| Candidate | Advance | Conv. Day | TOTAL 4th B |
||
|---|---|---|---|---|---|
| Mail-in | Electronic | Adv Total | Real-time | ||
| TOTALS | 13,808 (23.3%) |
35,800 (60.5%) |
49,608 (83.8%) |
9,602 (16.2%) |
59,210 (100.0%) |
| Unassigned | 1,389 | 3,890 | 5,279 | 5,279 | |
| Thomas MULCAIR |
7,690 (55.7%) |
20,993 (58.6%) |
28,683 (57.8%) |
5,198 (54.1%) |
33,881 (57.2%) |
| Brian TOPP |
6,118 (44.3%) |
14,807 (41.4%) |
20,925 (42.2%) |
4,404 (45.9%) |
25,329 (42.8%) |
[Click on image to open full-sized version]
Growth Between Ballots
When it came to winning over the preferences of advance voters, or down-ballot support on voting day, Mulcair was the overwhelming second choice of Saganash, Ashton, Singh and Dewar supporters after the first ballot, while on the next ballot Topp proved somewhat better than Mulcair in winning the second choices of Nash supporters.
By the final ballot, Mulcair edged ahead of Topp in second-choices of Cullen supporters (45% vs. 39%, vs. 16% who had no remaining preference on the ballot, and whose votes were thus "unassigned"). To win, however, Topp either needed over 80% of the previous Cullen supporters (excluding the unassigned), or else he needed new voters who had missed the earlier ballots that day to cast a vote for him on the fourth ballot. It was a tall mountain to climb, though given that the 4th ballot started at supper-time in the BC time-zone, it was at least worth a try.
[Click on image to open full-sized version]
NDP Leadership Contest Results by Candidate and Ballot, March 24, 2012
| Ballot | RS | NA | MS | PD | PN | NC | BT | TM | Unas- signed |
|---|---|---|---|---|---|---|---|---|---|
| 1st | 266 | 3,737 | 3,821 | 4,883 | 8,353 | 10,671 | 13,915 | 19,728 | – |
| +2,166 | +1,778 | +1,709 | +4,174 | +1,796 | |||||
| 2nd | 10,519 | 12,449 | 15,624 | 23,902 | 1,796 | ||||
| +2,977 | +4,198 | +3,586 | +1,277 | ||||||
| 3rd | 15,426 | 19,822 | 27,488 | 3,073 | |||||
| +5,507 | +6,393 | +2,206 | |||||||
| 4th | 25,329 | 33,881 | 5,279 |
The Value of Various Predictors
We noted a few months ago that in 2003 the cumulative share of fundraising by the various NDP Leadership contestants had predicted their vote share on the first ballot to within 1.5 percentage points.
With that in mind, and now that the last of the four weekly sets of financial disclosure reports have been filed, how did the various possible predictors measure up?
We looked at the cumulative share of fundraising, the cumulative share of the number of contributions, along with the last weekly shares of both. Let's see how they wound up.
First Ballot Vote Share as Against Hypothesized Predictors, NDP Leadership Contest, March 24, 2012
| Candidate | Hypothesized Predictor | % Vote 1st B |
|||
|---|---|---|---|---|---|
| Wk to Mar 10 | Cumulative | ||||
| % # | % $ | % # | % $ | ||
| MULCAIR | 22.2% | 26.6% | 25.0% | 24.2% | 30.2% |
| TOPP | 6.3% | 9.7% | 11.2% | 18.8% | 21.3% |
| CULLEN | 27.0% | 23.7% | 24.9% | 17.3% | 16.3% |
| NASH | 21.2% | 15.8% | 15.6% | 15.5% | 12.8% |
| DEWAR | 10.2% | 12.0% | 16.0% | 14.8% | 7.5% |
| SINGH | 10.0% | 10.3% | 3.7% | 5.8% | 5.8% |
| ASHTON | 3.1% | 1.9% | 3.5% | 3.5% | 5.7% |
| SAGANASH | n/a | n/a | 0.4% | ||
None of the four hypothesized predictors fared as well as 2003, with only the cumulative share of fundraising even getting the order right. Still, fundraising appears to have been somewhat more reliable than endorsement point-counting schemes, or social media traffic (the latter in spite of the complete dumb-foundedness in some quarters that the world can't somehow be changed merely by sitting in front of a computer screen and watching the Twitter ticker scroll by).
Perhaps what the table shows us is the impact of Ed Broadbent's bombshell in the final week of the campaign, which seems to have had the effect of polarizing the choice back between Mulcair and Topp.
But if any candidate could really be said to have had a surge in the last week of the campaign, it must surely have been Mulcair himself, who outlasted and outwaited all his squabbling opponents. His first ballot vote-share exceeded that predicted by his fundraising to the tune of six full percentage points, as Nash wilted and Dewar positively collapsed. And, as his win looked more and more inevitable over the course of subsequent ballots, the air came completely out of any putative efforts to "stop him".
In any event, here's what the final fundraising charts look like, starting with the cumulative fundraising:
[Click on image to open full-sized version]
NDP Leadership Contestant Fundraising, by Weekly Report
| Candidate | Reporting to X number of weeks before Voting Day |
TOTAL | |||
|---|---|---|---|---|---|
| 4 wks | 3 wks | 2 wks | 1 wk | ||
| Thomas MULCAIR |
$238,003 1,085 |
$24,250 151 |
$41,240 272 |
$21,820 165 |
$325,314 1,673 |
| Brian TOPP |
$213,982 538 |
$11,652 58 |
$13,270 85 |
$12,910 69 |
$251,814 750 |
| Nathan CULLEN |
$154,976 948 |
$14,448 152 |
$38,511 355 |
$24,337 209 |
$232,272 1,664 |
| Peggy NASH |
$163,318 688 |
$20,664 136 |
$11,908 82 |
$12,441 137 |
$208,331 1,043 |
| Paul DEWAR |
$169,598 885 |
$8,795 41 |
$9,255 73 |
$10,600 70 |
$198,248 1,069 |
| Martin SINGH |
$65,916 161 |
$2,346 14 |
$10,169 74 |
$78,431 249 |
|
| Niki ASHTON |
$33,370 161 |
$8,772 22 |
$3,635 25 |
$1,825 24 |
$47,602 232 |
| GRAND TOTAL |
$1,039,163 4,466 |
$88,581 560 |
$120,165 906 |
$94,103 748 |
$1,342,012 6,680 |
Here are the weekly charts for fundraising totals and numbers of contributors:
[Click on image to open full-sized version]
[Click on image to open full-sized version]
And here are the charts showing the weekly shares of each:
[Click on image to open full-sized version]
[Click on image to open full-sized version]
Great analysis.
This result is a slap in the face to those who accused Nathan of recruiting “carpetbaggers” with no sincere attachment to the party.
The people who cared enough to wait until the actual convention…the people who braved the hack attack for hours & hours…these were the people who chose Nathan as their leader.
I would hope that the older generation understand that alot of us younger NDP members are proud of the history & morals of the party. To us, Nathan was a result of it…not a contradiction to it.
These fascinating convention numbers don’t surprise me one bit. I saved my ballot to vote online on the day, and changed my vote from Mulcair to Cullen for the first three ballots on the strength of the convention speeches (and that’s even having watched every debate, studied the websites, etc, throughout the whole campaign). Previously Cullen had been my second choice, but was simply the best at the convention and I could suddenly see him as leader in a way I’d never been able to before. I suspect I may not have been alone?
@Dan Tan:
As a member of “the older generation”, I can tell you I never assumed anything but the best of Cullen supporters. He was my second choice just behind Tom. I have no question at all that you are “proud of the history and morals the party”.
I am glad you chose to join and feel very confident about the party’s future in the hands of you and your peers!
Arthur Cramer, Winnipeg
Interesting numbers, for sure, and in keeping with our correspondent’s analysis of the convention. She was a Cullen supporter, but left with a sense that he’d performed the best of any.
http://www.thoughtoutloud.org/2012/04/young-smart-ethical-and-leaderless/
Dear Pundit’s Guide -
Thank you for releasing this information.
Were you able to see correlation’s between which candidates supporters were close supporters to other candidates?
For instance, there was much speculation that the supporters of Mulcair, Cullen, and Singh were similar-minded. Similarly, Topp, Nash and Dewar looked to have a similar supporter base.
Would you for instance have information saying that X number of Topp and Mulcair supporters on the 4th ballot were originally Nash/Dewar/Cullen/Singh/Ashton as their first choice?
Thank you
Oli, I just got a table of summary results, so this is all that could be gleaned from that dataset. Always hoping for more, but Scytl and the NDP have bigger issues on their plate right now than extracting a more complex dataset for me.
I love working the data,so seeing this excellent work is wonderful- it would be interesting if by chance there is final donations reports out bsed on the voting day in Toronto. For example BT stands at 337500 gros-net 287031. Bill
Hi Bill, We won’t see the full donations reports now until the candidates file their “final reports” in six months’ time. See the section entitled “The Fundraising Primer” in this blogpost for the full details on what gets reported when:
http://www.punditsguide.ca/2012/03/cullen-momentum-threatening-mulcair-ndp-leadership-donation-data-suggests/
Thanks for the final Topp figures. I did blog the reports of his final numbers somewhere, but didn’t have this final figure until you posted it now.
Thanks for this Pundits Guide. Appreciated your excellent coverage during the NDP LDR. It would be great to see a vote by province breakdown of the first ballot. Curious to see the % of the Quebec vote that went Mulcair and a voter turnout analysis by province. Curious to prove my theory that his concentrated block of Quebec support + a way higher voter turnout in that province gave Mulcair the extra boost needed to get that necessary to win 30% on that first ballot.
The full size graphic link for “NDP Leadership Contest Results by Candidate and Ballot, March 24, 2012″ is broken.
Very interesting, but I wish the party would release the provincial numbers. It would be interesting to see which candidates did better in each of the provinces. You wouldn’t be able to find that out would you? I could make some maps for you :)
Everyone: I have asked for a more complete dataset for research purposes. But it was unclear how much extraction work this would entail for Scytl. As I said, they have bigger fish to to fry right now. Patience, all you data junkies!
Thanks for the bug report, Jim. All fixed up now.
Great analysis, Alice. I was intrigued from the beginning when you noted that Mulcair was a “foregone conclusion” as a result of the mail in and electronic ballets that were locked in at the beginning of the convention. You also noted that if the convention had been driven by the convention votes only, the outcome might have been different.
To me, this raises the question of the wisdom of the traditional “convention” method. We all saw what happened to the Liberals in their last leadership race, and it begs the question, does the feeling on the floor, and the to and fro of successive ballots lead to a better decision or a worse decision? Clearly, there is no short term answer to this question (except, perhaps in the case of the last Liberal convention!).
As you know, the Green party uses a single preferential ballot, and members attending the convention can wait until they hear the final speeches before casting their vote, but, as in this recent case, most members send their votes in ahead of time. There is less of a “show” for the viewers, but I wonder, do you think it produces better results?
Respectfully,
Jim
Interesting that about 20% of the advance ballots for candidates other than Mulcair or Topp didn’t rank other candidates, or at least not Topp or Mulcair, and had to be counted as “unassigned”. That’s a high percentage of spoiled ballots.
Martin, that’s 9.6% of the advance ballots were exhausted by the final count, not 20%. That’s 5,279 votes as compared to a victory margin of 8,552, not a factor.
Pundit, what these figures do not show is the breakdown between those who voted in person at the convention in ballots 2 through 4 as compared to those who voted in the same ballot electronically.
And the results you show also do not cast much light on the real question of the voting: To what extent did the voter suppression through denial of service attacks that persons currently unknown launched after the first ballot, succeed in discouraging and/or preventing voting by electronic means in the 2nd, 3rd and 4th ballot?
The subsidiary question is who benefited from this suppression and who lost?
The argument by Scytl and the NDP spokespersons that the voter suppression attacks did not alter any votes cast is only a partial answer; the attack appears to have been designed not to alter votes already cast but to deny the opportunity to vote to those who wished to vote via internet on the ballots that followed the first ballot. It was skilfully executed, and based on anecdotal evidence (from blogs and media reports) did indeed screw things up considerably.
Cat, you have misunderstood the process. The delegates at the convention voted through the online gateway, just the same as those from home who did not cast a preferential ballot in advance.
Really the only two categories of voters we can distinguish are those (i) who voted using a preferential ballot in advance, either by mail or online, and (ii) those who voted one ballot at a time on convention day (whether from a series of computers at the convention site, or on their smart-phones, or from a computer at home).
The impact of the Distributed Denial of Service attack would have been equally felt anywhere on convention day. You are right that, by definition, the purpose of a Denial of Service attack is to bombard a site with traffic to prevent legitimate users from getting through. The NDP’s auditor Price Waterhouse considered that if a roughly comparable number of people voted in the second, third and fourth ballots as had voted ballot-at-a-time in the first ballot, then the results could not be said to have been dramatically affected. And indeed, the numbers of people who voted in the first to fourth balllots one-ballot-at-a-time did not change substantially (10,487 to 9,403 to 10,922 to 9,602)
I believe you are under the mistaken assumption that the first ballot was composed only of preferential votes in advance. Not so. People could vote ballot-at-a-time for the first ballot, but the voting was stretched from Friday evening to Saturday morning at 9 AM eastern. It was the second through fourth ballots that were subject to the Denial of Service attacks.
While I do know of legions of people who were frustrated by the lengthy delays, I have not heard even anecdotal evidence of anyone who couldn’t eventually get through. I know that in many cases, the idea that someone would try and disrupt the vote only redoubled people’s resolve to make sure they got through, frustrating as it was, though I’m sure that those who did give up are probably not trumpeting that fact either.
It is truly unconscionable that someone would try and disrupt a democratic process in this fashion, but such attacks on very public, very time-limited voting windows are alas not uncommon. The attack might have had a political motive, or it just might have been an irresistably high-profile target for a competitive hacker who likes to brag and show off.
I’m not sure if the attack could have been said to have affected some candidates more than others. Certainly the leadership candidates themselves were making no such claim on that day, as I heard each say that they recognized it affected everyone equally.
Let me take this opportunity to clear up another common misconception. In shortening my earlier tortuously-long headline down to just saying that Cullen had “won the convention”, I had intended to convey that he won the ballot-by-ballot voting during the convention, while Mulcair had won the advance voting. Some people in the media and elsewhere apparently only read the headline and not the entire blogpost, and/or misunderstood the voting process and thus misinterpreted the headline that way. Nathan Cullen did not win the majority of the convention delegates so far as we know, nor would we have any publicly available data that would allow us to make such a conclusion. I didn’t realize what interpretation was taking hold “out there” the day I wrote this post, but apparently it created some headaches for exhausted party spokespeople who had to explain what actually happened all over again, and for that I apologize.
Interesting for me was that, in comparing candidates advance results to their convention day results (as percentages), we get (approximated, I didn’t pull out a calculator):
Martin Singh – down about 66%
Thomas Mulcair – down about 33%
Brian Topp – down about 20%
Romeo Saganash – par
Paul Dewar – up about 15%
Peggy Nash – up about 33%
Nathan Cullen – up about 40%
Niki Ashton – up about 100%
Now, obviously some of this has to do with the effort the Singh, Mulcair and Topp campaigns put into advance voting as the surest means of GOTV.
But Niki doing twice as well in live balloting as advance is surely an outlier. While our GOTV was inevitably less resourced than Mulcair and Topp, that doesn’t explain the difference.
I’d like to say it was her strong showing in the convention showcases (and I have anecdotes to that effect), but for her to do twice as well seems a little much.
Thoughts, O Pundit’s Pandit?
I have a theory on this Malcolm. Partly because this fits my decision process.
I would normaly vote live. But my choices settled with both Topp and Mulcair in my top three. Being a close follower of the race I knew it was virtually impossible that one of my top three was not going to be in the last round. So I just went ahead and voted in advance.
Lets say your top choice is Ashton, or you at least lean strongly in that direction. You dont know how long Ashton is going to last in the race, and you are much more likely to want to see how things are sorting before you make your choices after the first ballot.
Your following ballot choices are more likely to be effected by assesments of momentum than someone like me. And that would be magnified if you had a strong wish that Mulcair [or Topp] not win.
In other words- first choice Ashton supporters tended strongly to have a bigger stake in waiting to see what was happening at Convention before making down ballot choices.