Favourability voting
Favourability voting is an ultra high-definition family of electoral systems. It includes any voting system which includes separate individual scales for both approval and disapproval and also allows voters to express these two metrics on a single candidate. This is notably the only one in which individual votes can be displayed as a point in a whole entire arranged matrix, of approval versus disapproval.
Variants
Besides the most simplified form, Evaluative Favourability Voting (EFV), with no scores and only upvotes and downvotes, and the singular form, Noncomparative Favourability Voting (NFV), Favourability Voting (FV) also has an even more extensive, one-on-one matchup variant; Pairwise Favourability Voting (PFV), which generates a heavily granular dataset. These two both collect more expressive information and show greater differentiation between voters than just about any other voting system does.
Method
Favourability Voting is a cardinal voting method based on both score voting and approval voting in which voters numerically score each candidate and/or party on both of two separately divided scales: for approval (numbers from 0.00 to 100.00) and disapproval (numbers from 0.00 to -100.00). For example, in the more simple version, Noncomparative Favourability Voting, someone can simultaneously express +57.80% approval and -45.70% disapproval (for a net favourability of +12.10%) at the same time for any single candidate or party they wish. These two do not ever need to add up to each other. The positive percentages are then subtracted by the negative percentages to reach an election outcome, and whoever wins the highest sum (net approval) is selected. Pairwise Favourability Voting is even more intricate as this is where you freely measure how much you approve and disapprove of each candidate and/or party not just only individually but also in every single last possible one-on-one matchup there is. This is done and treated as wholly independent of each other and matchups such as A vs. B and B vs. A once again do not have to add up to 100 since they are not tied to each other (i.e. one might rate A in disapproval as -33.10% against B but B -66.80% against A). As you can see, each cell is treated as a different scale from each other and thus intransitive (circular preference) results in matchups (such as A > B > C > A) are fully allowed as the calculation process, which is different from other pairwise methods in that an overall score for each candidate is derived from the summation of their personal score and matchup scores together, and as such manages to bypass Condorcet's paradox. Scientists have determined that circles of preference are a natural occurrence in humans and this is in fact how many of our thought processes play out.
The final sum for each of candidates or parties is then deduced from the net favourability of not only the individual but also matchup scores, calculated by reducing the approvals by the disapprovals, and whoever has the highest rating then wins the election.
Ballot
The ballot form can be visually displayed in many distinct ways. For Noncomparative Favourability Voting, a voter may rate approval and disapproval on two slider bars for each candidate/party. That would look like this example of a single voter voting:
Candidate A:
- Approval: [ +75.24% ]
- Disapproval: [ –37.80% ]
Candidate B:
- Approval: [ 51.40% ]
- Disapproval: [ –27.86% ]
The net favourability given by the voter in this circumstance for Candidate A is 75.24 - 37.80 = +37.44% and Candidate B is 51.40 - 27.86 = +23.54%. These percentages given by the voter would then be summed into the candidate's overall rating.
In the case of Pairwise Favourability Voting, the ballot will look much more complex. There are very much lots of different ways yet again, but great one, for example, would be through a pairwise table:
| Candidate | A | B | C | Sum |
|---|---|---|---|---|
| A | +65.60 | +32.80 | +44.44 | +142.84 |
| B | +59.62 | +60.80 | +34.34 | +154.76 |
| C | +42.40 | +54.34 | +68.66 | +165.51 |
| Candidate | A | B | C | Sum |
|---|---|---|---|---|
| A | –33.58 | –64.40 | –51.69 | –149.67 |
| B | –41.40 | –57.51 | –55.80 | –154.71 |
| C | –36.80 | –45.20 | –66.32 | –148.32 |
The overall results for this person as in effects to the rest of the election's outcome from the three-way election example above would look like this:
| Candidate | A | B | C | Sum |
|---|---|---|---|---|
| A | +32.02 | –31.60 | –7.25 | –6.83 |
| B | +18.22 | +3.29 | –21.46 | +0.05 |
| C | +5.60 | +9.14 | +2.34 | +17.19 |
Candidate A would be set to lose -6.83pp while Candidate B would make a slight gain of +0.05pp, with Candidate C marginally winning over this voter, picking up a total addition of 17.19 percentage points to the candidates' sum.
An interesting, possibly better and wherein a perhaps slightly less time-consuming way lies would be to have voters place each candidate and party's performances against each other together on a grand interactive 2D Cartesian plane matrix, filled out by moving the their dots to whichever coordinates they choose. It would be scaled out this way: with the X axis as being disapproval and Y axis as being approval. Those with eyesight problems can also scale in and out (through a zoom in and out function).
This way, technically speaking voters do not have to open any more "windows" within this than they would with regular score voting, while also at the same time having the opportunity to express more deeply their preference and opinion.
Rationale
Favourability Voting encompasses all types of different possibilities, whereas with even the most detailed of other alternative voting systems, much of these are ignored. Let’s give an example of a single-winner election. Whereas in regular score voting, a voter who scores a candidate or party 50% could be interpreted as being three entirely different kinds of voters:
- Someone who both loves and hates everything in the platform (this can be caused by a conflict in which someone believes that all of these policies will lead to both positive and negative impacts at the same time: "side effects")
- Somebody who agrees with half of the platform but disagrees with the other half (for example, if someone is socially conservative and economically left-wing, then combining socially progressive with economically left-wing positions could turn this person into being half in support (on economic issues) and half against them (on social issues)
- Or even simply as a person who has neutral opinions (apathetic; doesn’t necessarily approve nor disapprove, just shrugs: some people may know about what a candidate stands for but they just still have no strong opinion about them) on the entirety of the premise
So, as you can probably see, these are pretty clearly three different feelings from each other which essentially have little to absolutely nothing in common. Pairwise Favourability Voting, unlike many other systems, is able to understand this and captures these three unique opinions separately: the one who both loves and hates an entity ("love-hate relationship/frenemies") would give it 100% on both approval and disapproval, and those who like half of something but dislike the rest of it ("meh/so-so") go 50% approval and 50% disapproval, whereas an indifferent participant ("whatever/I don't care") would put 0% on both approval and disapproval, and they can also be able to freely express different levels of these conflicting feelings when regarding comparisons, this level of expression allows for a better, more truthful way to sort out our preferences and for providing detailed statistical analysis.
Improving voter honesty, by fully including pairwise matchups, PFV distinctively allows smaller parties to gain more prominence by allowing a voter to express how much more they approve of them in comparison to the big party which they hate.
Enhancing the nursery effect, a lesser known party/candidate could easily be simply given a 0 under regular score voting, whereas with PFV, approval and disapproval are uniquely separated as two different measurements and therefore a voter is much more likely to give a 0 score on both approval and disapproval to candidates they do not have an opinion of, whereas if a candidate is hated they are likely to be given a 0 approval and 100 disapproval (this is an improvement in voter honesty, as while 0 could be interpreted as an "I do not know" score a 100 disapproval is much less unambiguous).
As a result of this deeper dive, this system ends up having a very intensive breakdown and factoring in of many different possible circumstances in order for a full grasp of exactly which one a voter may be or take place in this world. Which in turn leads to a much bigger understanding of the depth of a voter's thinking. For example, one takeaway about score voting has been that voters will have to look at the polls before taking a stand on the candidates. This would be a pretty common situation to happen in most of the world during loads of this century: let's say there’s one person running, Candidate A, who has a heavily polarized electorate of support. On the other hand there is Candidate B, who is also polarizing but slightly less so. However, while a voter may hate Candidate A, they may also not like Candidate B much. However, looking at the polls and if they are seeing that these are likely the two winning candidates, this voter will be encouraged to maximize Candidate B's score (100%) and minimize Candidate A's score (0%), even though they don’t really like Candidate B. Now, strategic voting aside, this is partly because in score voting, there is no actual explicit separation of personal and matchup scores, unlike with PFV, in which this voter could go around 50% approval for Candidate A personally but when taken against Candidate B, a number more close or even equal to 100%. Alternatively or alongside this, they may give 0% disapproval for Candidate B against Candidate A, while still going -50% disapproval for Candidate B personally. This represents how the voter may not like Candidate B (the lesser evil) themselves but fully supports them against the greater evil (Candidate A), unlike with score voting, which makes it impossible to express a maximal preference against a much less preferred candidate without that being the entirety of their score.
Here is a very good example by someone online of how cyclical preferences (which aren’t allowed by most voting systems, but are featured by Pairwise Favourability Voting) can occur: "Let's say I'm a Republican who prefers John Kasich to Donald Trump because I think Donald Trump isn't as trustworthy. But I prefer Donald Trump to Rand Paul because I'm a huge fan of the military. Yet Rand Paul is preferable to me over John Kasich, because he has a better policy on the free market; that may not have been a factor into Trump vs. Paul because my love of the military overwhelmed everything else, but let's say Kasich wasn't very hawkish, either, so the military didn't factor into that preference. One can see how "circular" i.e. intransitive preferences might be possible in a wide variety of circumstances for logical reasons in a complex world with complex choices."
History
The Favourability Voting family of voting systems was originally just an idea that occurred from a major revelation, but ultimately became more developed later and finally devised by a devoted User:DewyWind on June 13, 2021, and culminating in the creation of the this page, on August 25, 2021. Think of the Favourability Voting system as being a combined mix of score voting and approval voting: having a scoring range of 0-100 but with an additional differentiated positive layer and negative layer being on top of this aspect. If one decides to leave a rating blank then this is left out and not counted as being inside the vote. By this way of combining the two systems, it can serve as an easy and good compromise between both approval and score voting advocates, while also being able to swiftly capture and provide information at an even higher level, more so than both of these systems.