Share this article.

Exploring gender stereotypes from an intersectional perspective

ArticleDetailDownload PDF
Gender is one of many identities and experiences that overlap and intersect to form the human experience. Notions of gender are thus “intersectional” because of the multiplicity and complexity of these components. Intersectionality poses a challenge to many statistical modelling methods traditionally used in research, which are best suited to the study of simple, additive cause and effect relationships. Damian Kelty-Stephen, from Grinnell College, together with Sebastian Wallot and other colleagues, used multifractal modelling to explore the intersectionality of gender, specifically the way in which gender stereotypes influence our expectations.

Intersectionality theory aims to articulate multiple overlapping constraints on personal identity. It seeks to capture the diversity of an individual’s social experiences and explore these in a less simplistic way. Intersectionality has inspired new interest in our identity as diverse social and cognitive beings who navigate dynamic environments. An intersectional perspective, however, poses a significant challenge for research. Traditional linear statistical modelling techniques, which have dominated research, cannot account for complex, multi-causal constructs. Damian Kelty-Stephen, together with Sebastian Wallot and other colleagues, aims to address this problem by using a statistical technique called multifractal modelling to acknowledge the intersectionality of gender. His work using this technique demonstrates that we can actually leverage this multi-scale structure to explain and specifically predict our expectations of gender.

This image illustrates how self-paced reading looked in the task – words are loaded individually and only one word can be seen at a time.
The complexity of gender
The concept of gender challenges traditional scientific wisdom. Identities and ideologies surrounding gender emerge at complex intersections of social-cognitive constructs and biological inheritance, attached to both but belonging to neither. Until recently, science has been at odds with any formulation of gender not adhering to a binary distinction of male and female. Modern gender theory challenges these notions of so-called‘’natural’’ binary distinctions. Furthermore, historical analysis has demonstrated that notions of ‘’natural’’ binaries written into genetics are culturally rather than chemically based.

Intersectionality and gender
Intersectionality originally referred to a matrix of intersecting oppressions constricting an individual. Intersectionality theory proposes a framework in which social constraints (oppressive or otherwise) meet and interact with each other to create a unique experience of identity. It suggests that an individual’s experience unfolds through overlapping constraints and boundaries built at different scales
of social and cultural variation.

Intersectionality theory has inspired new interest in our identity as diverse social and cognitive beings who navigate dynamic environments.

Gender is one of many identities that intersectional theory looks to model, and intersectionality thus offers an appropriate framework to explore the complexity of identity and the expectations we hold. However, the practical application of intersectionality theory poses two key challenges to the research literature. Firstly, the traditional dominance of linear modelling in scientific research and subsequently the inability of such models to study gender because of their treatment of behaviour as merely the sum of various influences. The study of gender expectations provides one way of addressing these issues in the scientific literature.

Different perspectives on time-series modelling can allow us to interpret the very same sequence of word-reading times differently for a sequence of 11 words. The top panel schematizes that the 11th word-reading time depends only on independent word-reading times of previous words. The bottom panel schematizes that effects of those previous words might depend on the previous reading behaviour at a larger grain of time.
The importance of expectations
Expectation plays a major role in reading and especially in the way in which we approach narrative. Our minds attempt to make coherence from little information. We use our expectations in an attempt to achieve a sense of coherence.For example, we begin reading each sentence anticipating the next words that we have not yet seen. These expectations form a bridge which is continually extending outwards as we read from initial uncertainty to eventual coherence. This bridging entails both filling in details which are not present in the text and ignoring details in the text considered to be unimportant.

Expectation requires much more than traditional linear modelling offers
Building this bridge requires us to use backgrounds and contexts to project our guesses. Linear models can only explain behaviours as the sum of independent causes. There is no room in a recipe of independent causes for contexts and backgrounds. Linear models cannot project guesses and so, in order to model expectations, a new framework is needed.

>Expectation plays a major role in reading and especially in the way in which we approach narrative.

Multifractal modelling
Multifractal modelling exploits the tendency for construct-bearing systems, such as weather systems, natural hazards, ecosystems, galaxies and humans, to clump unevenly across time or space. Clumps follow nonlinear relationships with time scale called “power-laws.” “Law” means that the clump size increases in proportion to time scale raised to exponents or “powers.” The “powers” in this context are fractional, or “fractal” for short, and the systems carrying these constructs, such as gender, exhibit multiple power-laws. These “multiple”, “fractal” powers yield the term “multifractal.” Multifractal clumpiness in measured series often differs from what strictly linear modelling simulations of measured series would show. Multifractal modelling thus allows for the empirical estimation of the degree of multi-layered, multi-causal structure giving rise to constructs of interest, such as gender.

Average reading times per each word, both on the “reveal,” that is, the use of the first gendered noun referring to the protagonist, and for 15 post-reveal words. Average reading times belong to participants in the “fulfilled stereotype” group who saw the word “man” (dotted line) or to the participants in the “surprise” group who saw “woman” (solid line). There was no difference between these groups on the “reveal” word, but there was over the next four words and again for the 10th word (“nonchalantly”). Multifractal modelling significantly predicted how strong these group effects were for each participant.
Expectations of gender and multifractal modelling
Gender-stereotypes can inform our expectations when reading. Kelty-Stephen, Wallot and other colleagues used multifractal modelling in their work to explore the influence of prompting certain gender expectations and later having these expectations challenged. Expectations about gender are one example of the way in which a plot twist can be engineered for readers. Multifractal modelling was a useful statistical method in this context because it could make explicit the multi-scaled, multi-causal experience of a growing expectation about gender. Multifractality might predict individual differences in readers in response to finding their expectations violated.

Prompting gender expectations
In research by Kelty-Stephen, Wallot and other colleagues, words appeared in sequence from one of two 2,000-word stories that the researchers composed. Both stories described an interview between a first-person narrator and the protagonist. Words 1 to 999 were identical in both stories, omitting to indicate the protagonist’s gender but giving the protagonist stereotypically male professions (i.e., a soldier) and mannerisms (i.e., an aggressive temperament). The 1,000th word revealed the protagonist’s gender as being either male or female, depending on the story. Reading speed was measured for individual participants on each word across the entire story. But an individual’s ability to assume the gender expectation would depend on their own proficient navigation of narrative text at many different scales at once (i.e., word, phrase, sentence and social situation). And it was this multi-scaled behaviour that multifractal modelling could properly assess.

All participants read the crucial word indicating gender at the same speed. However, for the group who had read the word “woman”, their surprise caught up with them leading them to read more slowly over the next 15 words. Multifractal modelling of the word-by-word reading times allowed prediction of reading speed after the violation of the gender expectation for each reader. Specifically, it was the multifractal evidence of interactions across time scale in the reader navigation of a text which set up reader expectation of gender. Furthermore, analysis showed that the reading approach of those who had read the word ‘’woman’’ increased multifractality following the surprise, perhaps as a means of guarding against future surprise.

The work of Kelty-Stephen, Wallot and colleagues highlights the usefulness of applying multifractal modelling to the study of gender. In contrast to traditional linear statistical modelling techniques, multifractal modelling approaches allow for the implementation of an intersectional understanding of how we conceptualise identities. Specifically, multifractal modelling was able to make individual-level predictions about the way in which readers respond to having gender expectations violated. Results showed both immediate and longer-term responses were altered after the surprise. Such approaches hold great promise for the study of gender from an intersectional perspective but also for the investigation of other concepts which are the product of multiple, complex and interacting influences.

Other than gender, what are the other intersectional concepts that could be investigated using multifractal modelling approaches?
Damian Kelty-Stephen: Well, most things scientists are interested in are actually intersectional. There are very few smoothly delineated, easily separated things out there. Whenever we in science try to ratchet down on parts of the world, you find more variability, more context-sensitivity. You either ignore it and replace it with a more intuitive model that is easier to read, or you take a breath and learn the multifractal math that lets you quantify that context-sensitivity. In my lab’s research, multifractal intersectionality has been unlocking secrets underlying thought (e.g., mathematical reasoning), movement coordination, perception (visual and touch), and most recently, how we identify speech sounds.

What do you most enjoy about studying gender?
Co-author Hannah Brown: The fascinating part about studying gender is that it is so ubiquitous in our lives, yet its form and shape so hard to pin down. What I loved about this project was finding a way to demonstrate the gender biases that percolate throughout our everyday, trivial experiences. It is always exciting to see evidence back up theory, for gender theory especially.


  • Booth, C. R., Brown, H. L., Eason, E. G., Wallot, S., & Kelty-Stephen, D. G. (2018). Expectations on hierarchical scales of discourse: Multifractality predicts both short-and long-range effects of violating gender expectations in text reading. Discourse Processes, 55(1), 12-30.
  • Brown, H. L., Booth, C. R., Eason, E. G., & Kelty-Stephen, D. G. Multifractal signatures of intersectionality: Nonlinear dynamics permits quantitative modeling of hierarchical patterns in gender dynamics at the cultural level. In E. Mitleton-Kelly, A. Paraskevas, & C. Day (Eds.), Handbook of research methods in complexity science and their application (pp. 254-266). Cheltenham: Edward Elgar.
  • Wallot, S., O’Brien, B. A., Haussmann, A., Kloos, H., & Lyby, M. S. (2014). The role of reading time complexity and reading speed in text comprehension. Journal of Experimental Psychology: Learning, Memory, and Cognition, 40(6), 1745.
Research Objectives
Dr Kelty-Stephen, Dr Sebastian Wallot and colleagues apply statistical methods to their work that account for the complexity of gender stereotypes.

The researchers acknowledge the generous funding of Grinnell College’s Mentored Advanced Project program.

Co-authors on this paper:

  • Sebastian Wallot at Max Planck Institute for Empirical Aesthetics in Frankfurt.
  • Chase Booth (graduated Grinnell College in 2016; was a Thomas J. Watson Fellow, Class of 2016).
  • Hannah Brown (graduated Grinnell College in 2016; now works at the John F. Kennedy Center for the Performing Arts in Washington, DC researching the benefits of arts education).
  • Elizabeth Eason (graduated Grinnell College in 2017; now a Pricing Analyst at Nationwide Insurance in Des Moines).

Collaborators not on this immediate project who are nevertheless important in helping develop these ideas:

  • Anne Fausto Sterling (Brown University), an important source of feedback on my ideas on how to develop a multifractal end to an existing dynamical-systems view of gender.
  • James Dixon (University of Connecticut), my major advisor.
  • Zsolt Palatinus (University of Szeged in Hungary), a coauthor, fellow graduate student at U. Connecticut, and frequent collaborator concerned with the place of multifractality in psychology.
  • Navin Viswanathan (Kansas University), another fellow grad student at U. Connecticut, he also studies speech perception but shared key ideas with me about context colouring how humans use language. So, it’s no coincidence that this project has led me to think about speech perception in similar terms more recently.

Damian Kelty-Stephen earned a BS (2005; College of William & Mary) and PhD (2010; University of Connecticut) in Psychology. He worked at Harvard Medical School’s Wyss Institute for Biologically Inspired Engineering until 2013 when he joined Grinnell College as visiting faculty. As of 2017, he is a tenure-track Assistant Professor.


Dr Damian G. Kelty-Stephen
Assistant Professor
Grinnell College
1115 8th Avenue
Grinnell, IA 50112

T: +1 641-269-9525
Twitter: @OtherFoovian

Related posts.