Hacks gonna hack: Obsessive Presidential Vote Overanalyis.
[DISCLAIMER: This is post of very niche interest. If you are among the 99.99999% of the human population who have no direct interest in the University of Birmingham Guild of Students elections, you may wish to look away now. Also in the past on TCOGB I’ve had a policy of “no Guild shit”. I’m temporarily waiving that policy].
[As noted in Part One, at the time of writing I have not read any other post-Guild Election commentary. This is to avoid developing any potential GroupThink-induced confirmation bias. The opinions and conclusions expressed are entirely my own.]
Note: This post contains Spoilers.
Each year the revelation of uncontested Guild Officer positions inspires disappointment. Perhaps it’s because our western-conditioned minds fall back on some imprinted value about the innate goodness of democracy. Seeing an election proceed with only one candidate and no opposition makes us slightly uncomfortable; it’s a bit too soviet.
For myself there’s a second reason to dislike uncontested elections – it completely eliminates the opportunity for preferential voting system multi-round redistributive fun!
Seriously, does anyone else just love the tension of the thresholds, the creeping bar charts, the anticipated eliminations? Will they squeak it in the final round? Will the front-runner’s lead grow or narrow? Single Transferable Vote (STV) is an election geek’s dream.
I’m also told that its better for democracy. But who cares about democracy when you’ve got charts?
This year uncontested elections robbed us of the tension in 3 of the 7 positions, and Activities & Development was a two-horse race. That left just Education, Welfare and the Presidential race to provide classic election night drama.
In this bit of election geek analysis I’ll be looking at the Presidential race – what the scores on the doors were, and what that could tell us about the preceding two weeks.
My prediction going into results was that this was going to potentially be very close. Five candidates, all of whom appealed to their own segment of the student electorate. The strong likelihood that this would take more than one round to settle, and therefore the potential for redistributions to be game changers.
Looking at the Presidential results as a whole, there are a few general points which stick out.
Firstly, a very low number of votes are “lost” between rounds (<3% of votes from round to round between 1st and 3rd). People are using their preferences, and STV is working to find the most preferred candidate.
The Presidential election showed an unusually total of RON votes (79, 1.7%). Historically I’ve observed a general base level of around 200 to 300 hardcore “Ronners”, a level which holds true regardless of who the candidates are or whether an election is contested or not. The RON vote is sometimes higher, but rarely lower. Turnout was somewhat down this year, but 79 is low even compared to other 2014 races, especially given that President was the highest turnout position.
Why a low RON vote? Perhaps because the number and (political) diversity of the candidates neatly covered the political spectrum? Further, with two candidates who could potentially be interpreted as “joke” candidates, the more subversive “down with this pointless shit” type RON voters may have found a new home.
Round by round
If it wasn’t clear from the bar chart on the night, or from the table of results, it should be very clear from the pie chart that Poppy succeeded in capturing an early lead and a large plurality of the vote – over 40%, more than 17 points above her closest rival.
It also becomes apparent when displayed visually that the “opposition” vote is much more divided. It basically comes down to “everyone else vs. each other, vs. Poppy”. It’s worth noting too that at this point only 764 votes separate Ed in 2nd place from Maximillian in 5th – fewer than separate Ed from Poppy (a full 800 vote gap).
With RON and Maximilian distantly trailing behind both the frontrunners and the middle tier candidates (George in 4th place is only 9 votes short of reaching double Maximillian’s total), both are eliminated simultaneously. The election proceeds to a second round
The c.400 Maximilian and RON voters from the first round split themselves more or less equally between the other candidates. Slightly more opt for Poppy, but the difference is too small to read too much into.
Here Poppy almost clinches victory, falling just 0.37% short of the 50%+1 quota for election. Here again we see the “divided opposition” nature of this election. At this stage a mere 34 additional votes for Poppy (or alternatively a 17 swing from Ed to Poppy, with turnout constant) would have been enough to secure election without the need for a 4th round.
The fourth and final round marks the only round where Poppy is not the predominant beneficiary of vote redistribution. From the first to third round her lead increased slightly and then held roughly stable. Only now does her lead narrow slightly – to its narrowest in fact – though by nowhere near enough the make the final result anything approaching “close”.
This round is the notable exception to the general redistribution pattern – here Ed captures nearly half of the redistributed Puusta vote (with just one quarter going to Poppy and the remainder being “lost”). Unlike classic examples of “Anyone but X” voting patterns, voters here weren’t strongly opposed to the eventual winner.
The equivalent of 1708 individual votes were redistributed by the 4th round. 47% of those alone would have needed to vote for Ed for the two top candidates to draw level. A further plurality of the remainder would need to have broken for Ed to change the ultimately result – and anything short of winning around 74% (47% + [53%/2]+1) of ALL redistributions would rely of votes being “lost” through non-use.
Going into the 4th Round Ed would have needed to have picked up 96% of all 3rd round Puusta votes to ultimately win. Not so much a mountain as a concave cliff face.
Poppy ultimately secured victory with a full 15-point lead over closest rival Ed. All but the most biased analysts would have to acknowledge this as a clear landslide victory. Poppy can also take comfort and encouragement from the fact that just as many redistributed votes added to her lead as tried to narrow it.
Is there a psychological benefit to winning by a large margin in the 4th round, over potentially just edging it in the 3rd? It probably depends on how seriously people in the Guild bubble take and apply notions of mandates and political capital. In practice victory is all that matters. By the time August handover rolls around people barely remember who was uncontested, who landslid, who took three rounds (with the sole exception of the candidate themselves, whose fate it is to obsessively overanalyse until the end of time/term).
Additionally, in the case of a reelected President, we literally have no precedent to judge these things by. That history-making achievement alone has to be worthy of congratulation.
The impact of STV
Looking back over the results data, the impression is less of a nailbiting suspense as we wait to see who captures those all important redistributed votes, but more of a stubborn trudge towards the inevitable result. Poppy’s initial lead was near-insurmountable. Such leads don’t – in the context of Birmingham Guild elections – get overturned in runoffs.
Every Guild Election this year was basically First Past The Post. Every candidate with an initial plurality won. Chasing the 50%+1 mark was a mere formality. Would the result have been any different under FPTP? Short answer: No. Long Answer: No, if people had voted the same way. Different electoral systems change how people vote. The problem with Guild elections is that its impossible to quantify just how much voters take advantage of their ability to vote preferentially, and just how much tactical thought if any goes into it.
[Of course, if a student media organisation were willing to conduct proper election polling in future… hint hint…]
STV works to find a most prefered candidate – in this instance that happened to be the already leading plurality vote candidate.
Changes in candidate positions as a result of vote redistribution are the exception rather than the rule. The circumstances which allow for such tactics in “real world” STV elections (endorsements, factional ties, established candidates and parties, clear ideological labels, electoral pacts, etc.) just don’t exist here.
Don’t bank on 2nd Preferences unless you already have most of the 1sts.
[To be concluded in: Part Three – View from the Sidelines: The Pirate Puusta Campaign.]