VoteLocal: Helping Kiwis take part in the future of their cities.
VoteLocal was a Voter Advice Application (VAA) designed and built for use in New Zealand’s 2016 local elections, with the goal of helping young New Zealanders engage in the elections and increase voter turnout.
VoteLocal is an online tool that calculates and suggests which mayoral candidates are most compatible with the user based on their responses to a number of questions. It provides information directly from the candidates about what they stand for, informs users about what councils do, and debunks any assumptions that they aren’t relevant.
Through the 2014 iteration we learned a couple of critical things we never knew about our intended (and unintended) audience:
- the visual tone of the tool came across patronising and too childish
- users needed more transparency with what informed the results
Despite this, polls from the 2014 iteration told us the concept worked; it was shown to be successful in encouraging 30,000 non-voting 18–34 year olds to vote. VoteLocal aimed to take what proved to be a working concept, and the learnings from it, to a different space with different challenges—mayoral races in local elections.
Within this project I was tasked with a myriad of responsibilities that saw the product grow from its early concept to becoming a powerful tool for young people. This includes the design research, UX and UI design, coordination of web development, launch and dissemination tactics, marketing, and more. This case study focuses on the product’s design and impact, forgoing the dissemination and marketing aspects.
The Challenge of Local Elections
Local councils affect how we live our day-to-day lives. Despite this, there has been an ongoing decline in voter participation when it comes to local elections. In 2013, New Zealand’s national voter turnout for mayoral elections dropped to 41%, with young people (18–24) even lower at 37%.
After analysing an array of research in New Zealand on the topic we found that there are three main reasons why young people aren’t voting in local elections:
- They don’t have enough information on the candidates.
It was also identified that councils themselves often provide information that is too dense for people to make sense of, or simply not user-friendly.
- They underestimate the importance and relevance of local government.
Local government themselves identified that they need to “meet the current and future needs of communities”, so engaging young people at their level is pertinent to the success of local government itself.
- They are simply unaware of the elections themselves.
Research also suggested that to negate this it is important to make sure that information is easy to understand, issues important to electors are discussed online, and people are reminded online to vote.
VoteLocal was conceived to lower these barriers.
Ultimately our goal was to get young people voting. Responding to the issues found in our research, we positioned VoteLocal to carry out a number of objectives through the model of a VAA that would hopefully help achieve this:
- Identify regional political issues relevant to the capacity of local government, and form a series of questions to reflect these.
- Provoke users to indicate their own positions on the identified issues, enabling them to compare and understand where they stand in relation to the candidates.
- Match users’ personal values with mayoral candidates.
- Enable easy access to further information about mayoral candidates, the mechanics of their local government, and the local elections.
- Speak in an fun, approachable yet mature tone, different to typical political rhetoric—verbally and visually.
- Being a responsive website that will be accessible across all platforms and devices, and link up with social media.
- Provide mayoral candidates with another channel to promote themselves and make easily accessible their positions on the identified regional political issues.
Developing the User Journey
Early in the project it became apparent that the experience of the tool could follow card-game-like mechanics. This built on the idea that we’re trying to make this fun, familiar and approachable. This analogy lent itself well to developing the user’s journey through questions and information, as well as later informing the final user interface.
The challenge we ran into was we were trying to pack in too many ideas, or at least were struggling to find simple methods. The card-game testing method did help us understand this. Getting to a result for the user was paramount, but the budget was also a big constraint.
Question & Questionnaire Design
At the beginning of this project we realised we hadn’t understood VoteLocal in terms of what it actually was — a questionnaire. This made it predisposed to major factors of questionnaire design. An understanding of these factors and principles were used to ensure there was low survey fatigue (making sure they didn’t grow tired of answering questions), sensible user flow, good logic, and a high completion rate.
With regards to the questions themselves, analysis of annual plans, reports and briefings from our target councils provided an insight into the common responsibilities and roles of local councils in New Zealand. This was supplemented by reviewing local news reports of relevant topics. Gaining an overview of these provided a library of topics or issues to try apply to a generic set of questions.
Designing the questions themselves proved to be both the most difficult and most important task of the project, as the question format and length greatly impacted on the design of the tool and its build, as well as user comprehension. This was by far the most time-consuming process and involved quality assurance being provided by UMR Research and Prof. Jack Vowles in addition to testing with our target audience.
Much of the difficulty came from the immense task of condensing what are often highly-detailed local issues (quite different to national debates) into bite-sized questions that young people can engage with.
We created a prototype using basic InVision functionality. It provided a suitable level of interaction for testing both questions and question format with a variety of users. Through the prototype we could ascertain stumbling points with the sliding-scale format which we favoured in terms of content coverage.
With an audience of 18–34 year old New Zealanders in mind, VoteLocal was targeting a slightly more mature audience to that of On the Fence. Therefore, the visual style of VoteLocal sought to increase in maturity, both as a response to an older audience and critical feedback from some users of On the Fence, having called the visual tone “patronising”.
Through a variety of visual explorations we concluded with an engaging, fun and efficient vector aesthetic that maintained the gravitas of the content. We saw this as helping disarm the user’s preconceptions of political discourse.
The style was extremely helpful in providing balanced representation to mayoral candidates through individual illustrative depictions, and helped breaking the ice when persuading candidates to take part in the initiative.
Building on and refining an element from On the Fence, the ‘avatar’ was designed to be a feature of VoteLocal that pushed it from ‘just another questionnaire’ to something more joyous and rewarding. The avatar takes the premise of ‘helping Kiwis take part in the future of their cities’ by having their positions on issues generate their own cityscape.
The results page displays information in three tiers: the top three matches, the results card, and the comparison cards for each question that was answered. This responds to an objective of enabling users to compare and understand where they stand in relation to the candidates.
We found this required the tiering of information, from simple to complex, to cater for a range of user types—those who just want a result to those who want to get into the details of each candidate and each issue.
Algorithm Design, or, How Does it Work?
The matching algorithm is the core function of VAAs. This algorithm takes candidate positions and compares them with that of the user’s. In short, for this to happen positions are codified in a value between 0 and 100, or in some cases -1 and 1. From here the algorithm can work to produce overall match values between candidate and user. VoteLocal’s algorithm derives from what is known as the Hybrid model and takes into account a number of factors discussed by experts in VAA algorithm design.
During the 40 days VoteLocal was live before the 2016 local elections, over 20,000 people used the tool. We were pretty stoked with this, as we didn’t have funding for an expensive promotion campaign. The popularity of the tool largely came about through word of mouth and social media sharing.
Other metrics were gathered in the time and showed that the tool continued to be successful in achieving our outcomes.
- Of those who started VoteLocal, 83.38% completed it.
- 57.4% of sessions were completed by young New Zealanders (under 35) and 26.7% were under 25, the most disengaged group of electors.
- 31% said they had not voted in local elections before, and encouragingly 66% said VoteLocal did improve their understanding about what local councils do.
- Out of the 31% of respondents who said they had not voted in local elections before, 78% of them said VoteLocal had improved their understanding about what local councils do.
- 41% of users said VoteLocal motivated them to vote and 44% had their mayoral vote influenced by VoteLocal.
Although we can’t directly attribute increased turnout to VoteLocal alone, we can’t help but conclude we made some contribution. Remembering that VoteLocal was a prototype that we only rolled out in Auckland, Wellington and Palmerston North, we were pleased to see that of the metro centres, the two that had the largest increase in turnout over the 2013 local elections were Auckland and Wellington.
- In Auckland turnout increased from 34.9% (2013) to 38.5% (2016). That’s an increase of 3.7%.
- In Wellington turnout increased from 41.5% (2013) to 45.6% (2016). That’s an increase of 4.1%.
Atop these fantastic results, we’re also very pleased with the level of positive feedback from users. It’s comments like these that suggest we hit the nail on the head:
Very impressed with @votelocalnz — good way to identify whose policies align most with your own values. A+
VoteLocal is an initiative of the Design+Democracy Project, a research unit established within Massey University’s College of Creative Arts, and is made in collaboration with the great folks at Springload.
The core team consisted of Karl Kane, Tim Parkin, and Thomas Le Bas. The Springload team of Josh Barr, Bron Thomson, Vincent Audebert, Loïc Teixeira, and Charlotte Cami made things real. The beautiful illustrations were realised by Chloe Lassen. Thanks goes to an amazing number of wonderful people who helped make this initiative possible, especially: Claire Robinson, Sue Elliott, Richard McMillan, John (Neo) Anderson, Alexandra Hollis, Stephen Mills and the team at UMR, Jack Vowles, and Andy Asquith.