When performance is unequal

One can imagine a world where women and men are assessed fairly, a world shaped by data and analysis. Perhaps it requires being built from scratch, or almost. But it can exist. Here is how an artist, a data scientist, and an economist are working on it.

A group of serious, powerful men sitting around a table, questioning a woman. She stands before them, left alone and isolated. With simple lines, bold colours, and a few meaningful details, Zehra Ömeroğlu captures the plight of women in the workplace.

«In every sector, there’s always an attempt to isolate us, to weaken us» she explains. The Turkish artist has been on trial for four years for a cartoon about Covid-19 symptoms, published on 25 November 2020 in the Turkish satirical magazine Leman¹. «I believe that the State’s perspective on women and the role it assigns to them plays a significant part in this. Male cartoonists who draw similar scenes aren’t prosecuted for obscenity».

Sketching discrimination

What really counts in job performance evaluation? Who did what, and how?   Ömeroğlu’s experience, which she recounted at the European Cartoon Award ceremony², resonates particularly because of the field it involves: «In art, free thinking should prevail, but it does not. My male colleagues and I do the same job, tackle the same issues, yet when it’s a woman creating these works, society and the State judge her entirely differently» she says.

In her opinion, even reactions to her cartoons vary: «Though laughter is one of the most basic and uncontrollable reactions, I notice it changes depending on the artist’s gender» she notes. «Both men and women face strict limitations in my country, but in different ways. When a woman draws a harsh cartoon, she’s told it “doesn’t suit a woman” or that she should “be softer and more modest”. But if a man draws an even harsher one, it’s met with greater approval». Every time she draws political or current affairs scenes, Ömeroğlu encounters the same criticism: «Why don’t you draw women’s issues?».

The feeling is that «Just because you’re a woman, some people feel entitled to decide what you should draw. And in general, to tell you what to do».

Looking around for inspiration for her next cartoons, this artist continues to observe how difficult it is to find a field where women and men are evaluated equally on their performance. Yet this does not discourage her; on the contrary, it provides opportunities to «continue challenging the prescribed and internalised roles assigned to women» she explains.

Just a glance through her work reveals her strategy: she places women in different roles and portrays them in ways that diverge from familiar depictions. «I want to combat gender stereotypes by presenting women in unconventional lights, pushing everyone to adopt a fresh perspective» she says. Along with the pen, she uses humour: «the most powerful tool in this regard».

Despite facing a potential prison sentence of three months to six years and countless social media attacks, Ömeroğlu’s voice and art radiate grit and wit. And she never loses her irony. When asked how she would illustrate a world in ten years’ time, where men and women are treated equally, she does not miss the opportunity to highlight the…“performance” mistake. «Ten years is far too little for that! Women still have to pay a high price today simply to exist; there’s a long way to go before we’re assessed fairly! When it happens, I’ll be glad to draw it, but I still have plenty of time to imagine how» she jokes. Then she turns serious, keen not to come across as pessimistic. «Today things are changing rapidly, and the progress we’ve made in women’s rights and gender equality is remarkable – she clarifies – I have no doubt women will eventually be evaluated fairly, on par with men. It just takes time».

The source of distorted measurements

Time also emerges in the reflections of those who depict the role of women in the workplace with a different approach. They do not use pencils and colours but numbers, examining those used to measure job performance today that distort its quality, diminishing it when the protagonist is female.

To convey this picture, economist Marcella Corsi does not quote the usual gender gap statistics, gender pay gap, or gender role gap figures that we have all become accustomed to. Instead, she chooses to swim against the tide, tracing back to the source, to understand how job performance is measured and why this “how” is inefficient and thus, paradoxically, “unperformant”.

It is a process that requires patience: uprooting a deep-rooted problem means digging, getting your hands dirty, and not be in a hurry. «Evaluation can be based on time or the quantity of work delivered, or its remuneration» she explains, before showing how in both cases, the method used is not fair. The time and quantity of work delivered, for instance, are affected by the activity rate or inactivity rate (how many women say they want to work), the employment rate (how many women find work), and the unemployment rate (how many of the women who want to work actually find a job).

«All these rates in Italy reach worrying levels: gender gaps are chronic, and there are specific dynamics that make them so» Corsi explains. «Rather than not wanting to work, many women are likely still influenced by strong cultural stereotypes or forced to care for children or relatives. Their not working is not a real free choice». Corsi links the low female employment rate to the type of work available to them: «women often find only precarious and underpaid jobs, which impact the quality of performance at its core: they inevitably end up being less qualified and less qualifying» she explains. The same goes for the high unemployment rate, as it is equally high among those in unstable jobs. «Women are more affected by the employment crisis, often confined to what we call the key sectors of our economy like the tertiary sector, commerce, and hospitality» Corsi says. «As it happens, these are precisely the sectors at high risk of unemployment».

How much is equity worth?

If measuring time spent working does not lead to equity, the euros or benefits earned even less so. In 2024, according to the World Economic Forum’s Global Gender Gap Index³, Italy dropped eight places compared to 2023, falling to 87th globally. In Europe, it ranks 37th, only ahead of Hungary, the Czech Republic, and Turkey.

«Being paid less means lacking decision-making autonomy» Corsi emphasises

«There are laws to counter this phenomenon, but many forms of injustice, exploitation, and bargaining tactics during the contract phase see women often subject to reductions. These make them feel weak during negotiations». According to Corsi, those who attribute this phenomenon to women, characterising them as “suffering” from impostor syndrome⁴ , are mistaken. «It’s the result of growing up in environments still heavily male-dominated, where women are viewed and treated as outsiders. It’s an induced syndrome of someone who has already faced a challenge and knows what awaits her» she explains, before proposing actions for change.

Education in equality must begin with images in schoolbooks, early childhood toys, and family behaviours. «Children aren’t born with a sense of gender injustice; it arises later, with stereotypes» Corsi explains. She speaks of equity and uses only this term because «we are not identical. The aim should be for men to iron clothes and do things that are currently considered “women’s work”». As responsibilities move towards institutions, investment is needed in the public work sector, «where there is less discrimination» Corsi explains. «Today, however, there is a tendency to privatise everything. Then, we need a general education in more equitable time management and role-sharing, which are totally absent in Italy today».

Fixing the world with new data

Looking at both Italian and global data, the only way to ensure that women’s job performance is evaluated fairly seems to be to make the figures up. Does one have to cheat to eliminate the gender gap? Some have discovered how to do so legally and ethically, by adjusting the numbers that describe reality to make them fairer and more truthful.

This is a way to restore an objective view of the world, different from the bias-laden one we continue to accept and use to train AI models shaping our future. It is no trivial task but could offer a viable path towards equity in performance evaluation. Shalini Kurapati started out by venturing into the world of finance with Clearbox AI, the company she founded and with which she is trying to rebalance datasets. She achieves this by creating synthetic data, «generated via AI algorithms from real data samples, which retain the same statistical properties as the original data but allow for filling less represented areas, such as those relating to demographic groups or minority classes» she explains.

This “trick” impacts biases, including gender biases: «if real datasets are unbalanced, the model tends to favour the majority class, negatively affecting its fairness and creating prejudices».

«With synthetic data, it is possible to increase the number of examples from the minority class and improve the model’s fairness metrics: it is enabled to “learn” in a more balanced way, reducing the risk of making discriminatory or biased decisions».

Applying all this in finance means counteracting both direct and indirect discrimination in performance evaluations and financial decisions⁵. For example, avoiding a recurrence of the Apple Card case, which allocated different credit limits to men and women, «based on statistical models indicating a tendency for women to spend more, without considering the broader context» recalls Kurapati.

In finance, there remains a significant gender gap in crucial decisions such as mortgage or loan approvals. «As women generally earn less than men, models tend to penalise them, favouring men even under identical economic conditions or creditworthiness».

The unveiled picture reminds and reinforces the historical narrative described by Corsi, but also offers a way to draw a new one, where rebalancing is achieved by giving equitable numbers, not only in the financial sector. The pervasiveness of numbers, which until now has penalised women and distorted their performance, can now work to their advantage, particularly in sensitive sectors like healthcare and law.

«In the former, the availability of accurate and representative synthetic data enables the development of reliable predictive models, especially in medical diagnosis, drug research, or personalised treatments» explains Kurapati. «In the legal sector, fairness and accuracy of models can be significantly improved, especially when AI is used to support legal decisions, such as assessing the risk of recidivism or determining sentences».

Achieving performance equity in three sectors that significantly impact quality of life, like finance, healthcare, and justice, would be a dream, but one can dream even bigger: Kurapati sees success also in education and social policy. Thanks to that same technology that often acts as a vehicle and amplifier of inequality (not only of gender), synthetic data can restore equity systematically. It is an almost unexpected opportunity, yet sought after and desired, that we can seize today, without allowing ourselves to be hindered by prejudices about what is artificial. In this case, it could become the most natural method to attempt to “fix the world”.

---

1. On Zehra Ömeroğlu’s trial, Admin-Cfp, & Admin-Cfp. (2024, April 24). Alert Türkiye – Zehra Ömeroğlu – Cartooning for Peace. Cartooning for Peace – Dessins pour la paix. https://www.cartooningforpeace.org/en/alerte-turquie-zehra-omeroglu/ ↩︎
2. For the story of the European Cartoon Awards evening, Del Rosso, E. (2024, October 18). Uncategorized Archives – European Cartoon Award. European Cartoon Award. https://europeancartoonaward.com/category/uncategorized/ ↩︎
3. To read the report, World Economic Forum. (2024). Global gender gap report 2024. https://www3.weforum.org/docs/WEF_GGGR_2024.pdf ↩︎
4. Psychological perception of not deserving personal success and therefore “lying” about one’s abilities and skills. ↩︎
5. The inequity of financial performance evaluation in numbers is clear in Browne, R., & Sigalos, M. (2023, June 23). A.I. has a discrimination problem. In banking, the consequences can be severe. CNBC. https://www.cnbc.com/2023/06/23/ai-has-a-discrimination-problem-in-banking-that-can-be-devastating.html ↩︎