Methodology

WhereNow?'s analytical framework is grounded in a multidimensional approach – quantitative research (the index), public consensus (our interviews), and user input (your experiences) – to ultimately help young people decide where they want to live.

The WhereNow?^TM City Index (WNCI) represents the quantitative aspect of this approach, designed to rank cities globally according to their attractiveness for young people to live, specifically.

The idea was to apply the same analytical techniques used extensively in business, but at a social level, with the following tenants kept central:

(A) Data quality over quantity: Every Data Input has been meticulously evaluated and sources with inadequate methodologies and/or weak assumptions have been excluded. Similarly, cities with incomplete data have also been excluded.

At its core, we (the WN team) designed WNCI to help geographic decision-making for ourselves, friends, and family, placing elevated impetus on data quality. This meant compromising on coverage. Still, WNCI covers 158 cities across 72 countries, driven by millions of data points.

(B) Clarity over prescription: Some things are not quantifiable and nothing is objective. Indicators chosen by the team are based on our judgement, and weights applied (while adjustable online) reflect our collective preferences. If you want to critique our methodology/rankings, please do contact us. WNCI is designed to help, not tell people where to live.

The Process

Having established the goal: Build a definitive index to rank cities globally according to their attractiveness for young people to live, we started by asking one another: Why do people love where they live? What makes a city great?

1. Data curation: After several conversations, we developed a lengthy list of approximately 50 Indicators. That list was then iterated for parsimony (simplicity), then reiterated based on data quality/availability. Indicators were then categorized into 6 Dimensions to form the following taxonomy:

WNCI Framework

Data Inputs feed into Indicators -> Indicators feed into Dimensions-> Dimension scores are aggregated to calculate final WNCI scores/rankings

Data for each Indicator was correspondingly sourced from a variety of central databases and national sources, cross-analyzing city coverage. In instances where source quality was not compelling, that source was replaced/removed. In instances where city coverage was incomplete, cities were excluded from WNCI coverage.

This stage was not linear. We continuously substituted data sources, edited Indicators, and excluded cities (where necessary) to optimize data quality throughout the research period timeframe.

2. Design choices: Having set the foundation for WNCI, Indicators were then further refined, and data was chosen selectively to reflect young people's preferences. Being 25 (Ollie), 25 (Ned), and 26 (Nic), these are preferences that we can speak to – depending on who you ask… For us, we defined young people as those aged 18-30, but recognize that age does not always correlate consistently with life stage.

For instance, Relationship Opportunity scores were estimated using gender equity ratios for the demographic pool ages 20-30, sourced primarily from UN consensus. Similarly, Education scores are calculated using a weighted ranking of QS University Rankings, tagged to defined city metropolitan areas.

Some design choices were more/less simplistic. For instance, Sunlight data is measured in yearly sunlight hours, pulled directly from Global Solar Atlas. Alternatively, Transport and Strength of Economy Indicators were derived using more sophisticated, in-house models pulling from several Data Inputs, such as OpenRailwayMap and the IMF.

3. Synthesization & weightings: Once design choices were quality checked, and assumptions were bulletproofed – through either (a) source substitution; (b) iterations in Dimension categorization; (c) Indicator development – all data was consolidated into a single comprehensive model (See – WNCI Framework).

Thereafter, all Data Inputs were normalized (0-10). Most Data Inputs are normalized via a bounded linear method – higher nominal values asymptotically approach a score of 10 (good) and lower nominal values approach 0 (bad), or the inverse depending on the Indicator's polarity.

Alternatively, modelling Temperature, for instance, required setting optimal average centigrade bounds, with smoothing and asymmetric curtailment – effectively, cities with warmer temperatures score higher then curtail, given heat is positive to an extent.

4. Final output: To calculate final WNCI scores/rankings, we (The WN team) weighted each Data Input by Indicator, then each Indicator by Dimension, based on our collective outlook. Dimension scores are weighted evenly, but Indicators are not. And critically, final weights were assigned prior to final city rankings being calculated to omit any cherry-picking / ranking manipulation, free of any commercial compromise or financial bias.

Ultimately, WNCI is a quantified reflection of how we see the world, keeping the principles aforementioned (A & B) at the forefront of every analytical decision made. The process has been incredibly informative for us to guide where we want to live – we wish every user the same.