Kittiwake

tech culture society

Rules are necessary: Data Feminism and its seven principles

In my previous articles, I mentioned the concepts of Big Data and Data Feminism and the importance of these concepts. In this article, I will try to explain the principles of Data Feminism by presenting my own perspective under the lead of the Data Feminism book.

By Sibel Dinç

July, 2021

A look at the seven principles of Data Feminism 

The seven principles were put forward by D'Ignazio and Klein, authors of Data Feminism. Each item should be considered in order to ensure the equal visibility and application of the concept of feminism and mainly labour in the digital environment. As it is known, the concept of feminism is more than just a simple struggle for equality between men and women. Feminism is an umbrella phenomenon used for all disadvantaged groups except the powerful men of male-dominated societies. 

The phenomenon of inequality that we encounter in our daily lives has been transferred to the digital environment, technological devices and software alongside with our entry into the digital age. This necessitates the examination, research and production of solutions in the digital field of feminism. 

The seven principles identified are “Examine Power, Challenge Power, Elevate Emotion & Embodiment, Rethink Binaries & Hierarchies, Embrace Pluralism, Consider Context and Make Labour Visible”. Let's take a closer look at these items and try to raise awareness.

What is the purpose of the principles?

SpeakinData science deals with issues of interest in terms of quantity. However, in order to understand the root causes of the issues, the sides outside the quantitative framework should also be considered. In other words, it is a big mistake to consider data science only as number science. Each observation that can be described as a number -it can be a living thing or an inanimate object sometimes- (it) has qualitative outputs as well as quantitative outputs. At this point, the qualitative aspects of the data obtained should be handled with certain concepts.

In terms of Data Feminism, the seven principles determined were created to ensure that each subject examined is handled correctly in terms of data science. Let's take a closer look at the principles: 

1 - Examine Power 

It is argued by D'Ignazio and Klein that the purpose of studying power is not only to study power but also to dominate it. Knowledge is power and the field of data science allows to contain and analyse existing knowledge. At this point, examining the power will enable us to learn who has the power and whether it is manipulated, that is, to learn the current situation and to take an appropriate step.

2 - Challenge Power 

When we look at the world order, the existence of disadvantaged groups at all times protects the interests of those who hold power. Here, the authors emphasize the importance of examining the struggle with power as a second principle, and therefore the struggle with the phenomenon of inequality, in the data set. 

3 - Elevate Emotion & Embodiment

Scientists accept that the objective studies are reliable. Because they assume that there is not any emotional opinion in objective studies. But every single study in data science involves much more than a simple quantitative figure. In order to increase the impact and awareness of the studies carried out, it should be shown that there is more than just a number or a set of figures. For example, it is necessary to look at how a visualization example, in which the emotion phenomenon mentioned by the authors, changes their perspective on the event. The authors mention that two different visualization techniques and the effect of the news can create differences in an armed attack in a school in the USA in 2012.

From the perspective of feminism, we still have not been able to overcome some prejudices, no matter how much we object. One of them is that the phenomenon of emotion is associated with the female gender. And as it is known, where there is prejudice, the excluded group or groups are ignored. This situation prevents these groups from obtaining equal rights alongside the powerful and at the same time being represented to the required extent.

The perception among scientists that addressing emotions discredits studies is common, but human beings are emotional beings and events always have an emotional side. This emotionality should be accepted to the extent that it does not affect the objective point of view. The fact that emotionality is included in studies does not mean that the study is biased and lacks reliability. With the third principle of “Elevate Emotion and Embodiment”, the authors point out that the lack of emotion in data science studies and the devaluation of emotion is a deficiency. 

4 - Rethink Binaries & Hierarchies 

The most basic action for the representation of a group, namely its visibility, is to be counted and processed, that is, to participate in statistics. However, it is seen that stereotyped binary systems are still used in some cases in the field of data science and “other” segments are excluded. In fact, this both affects the representation and questions the accuracy of the data obtained. From the point of view of gender, although we use binary coding as a biological gender, it is necessary to include the concepts of sexual orientation, gender identity and sexual expression, depending on the suitability of the study. It should be taken into account how the person defines himself/herself, how he/she feels and how he/she wants to be represented. It may be right to do this without committing to an option for it. Facebook's step in this direction can be given as an example.

The fourth principle emphasizes the importance of classification and that hierarchical structures should be reconsidered in data science studies. As mentioned by the authors, “While classification and hierarchy are powerful methods of enumeration, they are also smart weapons used to dominate, discipline and exclude certain groups”.

On the other hand, I think that although classification is important for data science, it actually creates a labelling and thus supports discrimination between certain groups (gender, race, religion, language, etc.). However, I know that there is no clear step to be taken at this point, at least at the moment. Classification is the scaling technique underlying categorization science and data science. Classification is still used as the most basic statistical method to understand the current situation. Providing more options for accurate representation or obtaining open-ended responses from observations that received responses for the study will yield more accurate results, at least until other methods are found. 

5 - Embrace Pluralism 

The fifth principle refers to the necessity of pluralism. From the point of view of statistics, the selected sample should represent the population well so that the population can be understood. For representation, each group should be given equal weight in the sample. In fact, this principle is a generalization of basic statistical methods. By embracing pluralism, ignorance of any group is prevented, and real results are obtained for the research of interest.

6 - Consider Context 

In the sixth principle presented by the authors, it is said that data feminism is biased and contains subjective judgments. It is claimed that the main reason for this is due to unequal social relations. It is mentioned that instead of seeing data sets as a structure consisting of numbers only, we should look at our research question and the context of our field of study as a "mine". However, as it manifests itself in all principles, it is known that knowledge is in the hands of the dominant group or groups and can be shaped as desired. As the authors point out, data cannot speak for itself. For this reason, it is important to understand their contents well.

Another handicap is the open data concept, which makes the use of datasets accessible to everyone. This concept adopts the free presentation of data to the public without certain restrictions (copyright, patent, etc.). Although providing information freely and free of charge is a positive step, if it is not processed in the right hands, the outputs will be extremely biased and misleading. Open data can also mean a data set that is easy to manipulate. However, from another point of view, sharing the data sets with the public without any restrictions can also create a collective trend for social consciousness, awareness and solution stages for the solution of some events. The striking examples of sexual assault cases presented by the authors also demonstrate that the concept of open data provides a serious impetus for faster handling, analysis and presentation of data sets. In other words, presenting the data in an open way is still an extremely controversial issue for now. 

7 - Make Labour Visible 

Making the work visible is the last item of data feminism principles. In this regard, discrimination based on gender is also present in the digital environment. According to a study conducted through the GitHub site in 2016, it is seen that the gender-based acceptance of the contributions published on the site for code development differs. It has been determined that the acceptability of the female workforce in this field is at a low level compared to the male workforce. Apart from that, a similar example applies to technology companies. Although they have demographic content that includes discrimination in terms of race and gender, the efforts of the excluded group in the field of data science are undeniable. 


What have the studies taught us? And what can be done?

D'Ignazio and Klein's book Data Feminism addresses the seven principles of data feminism and the necessities of these principles with the help of examples. The common theme that all the principles of this book, which raises awareness about the studies in this field, are addressed, is the power imbalance and its reflections in data science. This does not only cover gender-based discrimination. There is also racial discrimination.

At this point, the first thing to be done is to raise awareness to the public on this issue, as is done now. While the principles of data feminism contribute to this awareness; the application of these principles to the works will be effective in creating more accurate strategies, thus using time, effort and money at an economic level, recognizing hidden labour and minimizing power imbalances. 

Let's think together! What other studies can be done to raise awareness in this area?

Suggested readings

1 “Seven intersectional feminist principles for equitable and actionable COVID-19 data” in terms of handling a topic that is up to date in terms of data feminism.

2 Another suggestion is no doubt to read the book "Data Feminism" in order to explore the examples and deep analysis it contains.

Next post :

Discriminatory language exists in the field of technology as well as in everyday life. The creators of course also have a share in the formation of discriminatory language. So, do you think a search engine can be a feminist? In my next article, I will talk about whether it is possible to apply the concept of data feminism on digital products with case studies.

Sibel Dinç is ongoing at King's College London in Big Data in Culture & Society MA program. She worked as a research assistant in Turkey. She is interested in social media studies. She takes an active role in The Institute for Internet and Just Society as a researcher.