Why work is one of life’s most powerful matchmakers by ABC

This story analysed Australian census data for roughly four million working couples. Provided exclusively to the ABC, the custom data spanned two censuses 20 years apart, including same-sex couples for the first time (something that ABS consultants originally advised was impossible). Our final analysis covered more than 70,000 job combinations across both censuses.

This project began when our data journalist noticed something odd about her married friends. A strangely high proportion had a partner in the same job – unlike she and her own partner. She wanted to know: “Am I weird or are they?”

But while assortative mating (the tendency to choose partners like ourselves) by occupation has been studied abroad, no research has focused on this question among Australian couples. This led to one of our key challenges: obtaining the right data and designing a methodology that would fill this gap in the research.

What followed was weeks of preliminary discussion with researchers in the field and consultants at the Australian Bureau of Statistics (ABS). We wanted a dataset that would allow us to track changes over time, while maximising the granularity of job classifications and minimising low values. Once we received the raw data, key steps in processing and analysing it included:
- Mapping jobs across two decades of workforce changes
- Accounting for “mirrored pairs” (e.g. a nurse-police couple duplicated as a police-nurse couple)
- Calculating percentages for every couple type, age group and job, as well as rates compared to the general population

Our next challenge was working out how to visualise such an enormous and detailed dataset for a busy audience. Each of our visualisations and interactives fulfilled one of two aims:
- To show overall patterns in the data (the “big picture”) or,
- To show people where they fit in the data (the “personal”).

Our early experiments with network charts, chord diagrams and other chart types convinced us that none would tell a clear story: we needed to create something new.

In the end, we adapted our main visualisation (“Who marries whom”) from blending a network chart with a treemap. One advantage of the treemap was that it better illustrated the class structure of the workforce. Drawing lines between jobs on this “social map” produced a clear snapshot of the “inwardness” or “outwardness” of any given job: romantic matches that crossed the class divide required a longer linking line, while jobs that kept to themselves or their industry had shorter lines.

The treemap also showed more common jobs as larger rectangles, which could be directly labelled. This invited interaction and made the viz easier to understand at a glance.

We brought the data to life by weaving real love stories into the numbers. Our public call out for stories (https://www.abc.net.au/news/2024-07-21/opposites-attract-or-exactly-the-same-couples-callout/104122752) received over 500 submissions. The final selection was chosen to represent a wide cross-section of jobs, locations, ages, sexualities, physical and mental capacities, and experiences.