Even in Norway, widely perceived as progressive and egalitarian, gender imbalance in content is a reality – that’s what media company Amedia found after investigating its own output. The publisher also discovered evidence to support their belief that better gender representation makes business sense. After analysing stories from 19 newspapers published across a 21-month period, they found that publications which featured more stories containing women sources had higher readership among women. This interesting correlation was identified as part of a wide-ranging investigation into gender balance across 660,000 stories from 64 newspapers published over the same 21-month window – a massive endeavour, made possible through data science and automatic classification. Using natural language processing techniques, Amedia was able to automate the process of identifying and counting the number of women and men mentioned in its content. They first identified so-called entities referenced in stories, such as people, businesses or organisations, in order to be able to differentiate between, say, a business with ‘Nina’ in its name, and an actual person named ‘Nina’. The second step involved matching recognised names with Statistics Norway’s public database of men’s and women’s names. With this data in place in their centralised data repository, they were able to run analyses towards all aspects of data.
They found that across the 660,000 stories, on average, only 34% of names mentioned belonged to women. But further analysis of the individual newspapers revealed significant variation. They found that the content of smaller publications tended to be more gender balanced than larger, regional ones.
For instance, the top performer had a 42% share of women’s names. The differences in gender balance became even more apparent when they mapped gender across different story topics. In ‘Education’, which encompassed some 30,000 stories, 47% of all names belonged to women, compared to only 21% for the 37,000 stories under the topic “Disasters, emergencies and accidents’. The category dubbed ‘Society’, which covers topics such as communities, families, and welfare, was the only one in which women’s names were in the clear majority.
These differences in gender representation may well be a result of authority figures skewing towards being women or men in different sectors. However, they could also stem from journalists’ own gender biases influencing who they choose to interview.
Future plans include offering all of Amedia’s journalists and editors running data on gender imbalance. Some editorial teams already use a specially developed dashboard, which informs them about the gender gap in readership in real-time. (See AMEDIA)