Why data is a feminist issue
5 March was International Open Data Day. 6 March was Mother’s Day in the UK and Ireland. Today, 8 March, is International Women’s Day, a public holiday in many countries including Russia.
We already know that ‘Poverty is Sexist’, as ONE’s new report puts it. But where does data come in?
The answer is because official data, on everything from national income to death rates, is systematically biased against women, and can exclude them entirely. Here are three examples: care costs, how surveys are designed, and how deaths in childbirth are measured.
Researchers at the Overseas Development Institute looked at data from 66 countries, and found that women did 3.3 times as much care work as men – ranging from ten extra days to ten extra weeks a year. Yet none of this work is counted in national income. There is an old joke in economics textbooks that “if you marry your cleaner, GDP goes down.” But it’s no laughing matter for the women who do the bulk of caring for children and elderly relatives, often with little support.
There is more bias in the surveys that statisticians use to see how countries are developing. For instance, surveys often define a ‘head of household’ as ‘the person in the household acknowledged as head by the other members’. This is difficult to define or measure consistently: in some communities, men might be heads of household by custom, in others women might be, while in many the idea of having a ‘head’ seems outdated. Moreover, most surveys look at income or food consumption for the whole household, so if boys get better food than girls (or vice versa) we don’t know about it.
We know that many women around the world die in childbirth – but we don’t know how many. Bill Anderson at Development Initiatives found that four out of five African countries did not maintain civic death registers. Instead, they estimate maternal mortality from GDP, fertility rates and the number of midwives. That’s heroic but unlikely to be accurate. In 2009-10, Sierra Leone introduced free healthcare for pregnant women and children. Over the next 3 years, the proportion of women giving birth in a health facility doubled, but there was no significant change in the death rate. Is that because of poor quality medical care, a lack of drugs, or something else? Without good data, we’ll never know.
These omissions have grave consequences. A panel advising the United Nations noted that when we fail to count something, the message we send – intentionally or not – is that it doesn’t count. Care work is a good example. The McKinsey Global Institute estimates the value of unpaid care work at $10 trillion. When women do get paid for their work, it’s 20 to 40% less than men.
So what needs to be done to stop data from excluding women?
First, count more. Few people get excited about building civic registries or running a census, but they are critical to making sure everyone is counted. It’s not expensive: about a billion dollars a year for all developing countries together. That’s the same as the cost of political opinion polls in the U.S. alone.
Second, count better. When surveys are disaggregated by gender, the results can be surprising. The OECD has a paper by three researchers (all men, as it happens) who found that households headed by women in Thailand and Vietnam were better off, on average, than households headed by men. This was only true for women whose spouses who migrated for work, though. Widows and single women were worse off.
Third, change what is counted. National income accounts were designed in the 1940s when most people worked in farms or factories; it is time to update them. ODI suggests gathering more data on care work. Why not go further and change the definition so that it counts towards GDP? That might encourage employers to take it seriously and men to contribute to it equally.
Data has excluded women for as long as people have been keeping records. It’s time to change that.