“Up to 70% of the labor involved in coffee farming is supplied by women.”
You’ll see that statistic reported far and wide. We’ve shared it in our communications at TechnoServe, too.
But is it actually true? Where does this stat even come from?
If you trace the statistic back to its original source, it first appeared in a 2008 survey published in the article “Women in Coffee,” which ran in the “International Trade Forum” journal. While that article no longer appears online, the survey is described in the 2012 International Trade Centre publication “The Coffee Exporter’s Guide.”
The survey asked “twenty-five persons, mainly women, in 15 coffee producing countries” to estimate the share of labor provided by women in different activities, and published the aggregated estimates in the following table:
Over time, those second-hand estimates — provided by an average of 1.66 respondents per country — have been transformed into a widely accepted fact about the coffee sector writ large.
Beware the zombie stat
This is an example of a “zombie stat” — an outdated, inaccurate or incomplete statistic that simply will not die. Often, as in the case of the International Trade Centre report, the original publication is transparent about the methodology used to arrive at the statistic, and is clear about its caveats and limitations. But over time, as the statistic is repeated and adapted from secondary, tertiary and quaternary sources, that context falls away, leaving a misleading data point.
The statistic on women’s role in coffee production is hardly unique: The global agrifood industry is rife with similar examples, which can also be found across the broader international development sector. That statistic you’ve probably seen about smallholder farmers producing 70% of the world’s food? The real number is probably closer to 30%. In fact, the use of these unreliable stats has become so pervasive that in 2017, researchers at the World Bank mined new survey data to challenge and rebut many zombie statistics related to African agriculture.
Why zombie stats flourish
It’s understandable why these zombie stats arise and thrive: Those of us in the agrifood sector want data to help guide our decisions and support our arguments. Often zombie stats support something that feels true — it’s certainly the case, for example, that in many places, women provide a large share of the work in coffee production. And these zombie stats fill a gap where more reliable data is not available.
However, using zombie stats in lieu of better data can give us an inaccurate picture of reality. The actual picture of women’s roles in coffee production is much more nuanced and context-specific, and understanding that nuance enables us to design more effective interventions.
For example, in Ethiopia, while women perform a lot of work in coffee production, the fact that they don’t own the land means that they often feel a lack of ownership over the crop, and are not seen as farmers. We saw that in our first training cohorts when TechnoServe began to work there in 2011: Just 4% of the farmers who attended this initial training were women. Our team realized that it needed to address constraints to women’s participation in training — including inconvenient scheduling and lack of child care — but also that it needed to cultivate a sense of joint ownership of the coffee farm within the household. Today, more than 40% of the participants in our coffee programs in Ethiopia are women.
Solving the Problem of Zombie Stats: The need for better data
How can we in the agrifood sector address the problem of zombie stats? In part, the solution starts with ourselves and how we communicate. It’s easy to fall into the practice of using stats that are “too good to check” when attempting to draw attention to a development challenge or persuade an audience to take action. But while it’s tempting to pull an unsourced statistic or blindly trust a citation we read in an article, we should be rigorous about going back to find the original source. Statistics can still have power even when we frame them in their full context.
To really kill off zombie stats, however, the sector needs to invest more in research. For example, there are a lot of key unanswered questions about gender dynamics in agricultural value chains, including a lack of understanding about which interventions can be scaled across geographies to advance women’s economic empowerment in agrifood systems.
There seems to be a growing awareness of this need, as in the last few years, we’ve seen a new focus on boosting investment in this type of research in agriculture and food production. For example:
- FAO’s “The Status of Women in Agrifood Systems” report provides a comprehensive review of regional and global evidence on the value of women’s participation during and after food production, exploring how agrifood systems can contribute to advancing women’s empowerment.
- USAID’s promotion and review of impact evaluations are also generating new insights on promising approaches to increase women’s access to and control of income in agrifood systems across geographies.
While these global and regional studies are extremely valuable, continued investment in country- and program-level gender and value chain assessments — relying on quantitative and qualitative data — is essential, if we hope to design and implement approaches that respond to local contexts and norms.
If we invest in answering these questions and obtaining trustworthy new data about these issues, we will have a clearer picture of the reality facing emerging markets communities and the development initiatives that serve them. With this new awareness, we can more effectively address the challenges that women and men face in fully participating in — and benefiting from — agrifood systems. And perhaps we can also spark a broader movement in global development, away from zombie stats, and toward a deeper, fact-based understanding of the issues we’re collectively addressing.
Cait Nordehn is Senior Manager at TechnoServe’s Global Gender Practice.
Photo courtesy of Tara Winstead.