Brenda Oulo & Aubryn Sidle, The Girls' Agency Lab (GAL) Fernanda Gandara, Room to Read
Agency, defined as the ability to effect change over the nature and quality of one's life (Bandura, 2001) is a multi-faceted construct. A girl's agency is, of course, dependent on the opportunity structure of her environment, the intersecting forms of discrimination she might face (Crenshaw 1990), and the ways in which her agency is mediated by her community and social relationships. Yet agency is also related to something deeply personal: a young person's belief in herself, her belief that her environment can be changed, and the skills she possesses to navigate these challenges (Sidle & Oulo 2021). Reviewing how psychologists, educationists, and feminist social theorists have understood women's and girls' agency gives a few clear conclusions for practice.
First, agentic capacity (the dynamic combination of agency-building beliefs and skills) is something that all individuals have, and can develop. We know from the youth development and education literature that programs can be highly effective at improving young peoples’ ability to exercise agency in their own lives (Lerner et al. 2005, Temin & Heck 2020), suggesting that agentic capacity can be specifically cultivated or ‘taught.’ If this were not possible, then many of our efforts in the girls’ education space to build such capacity would be futile.
Second, agency is clearly constrained (or enabled) by the social, economic, and political contexts in which girls live. In other words, how girls exercise their agency and to what degree depends deeply on their setting– including tangible opportunities available to them such as access to financial or educational resources, in addition to the structural constraints around them such as racism, classism, harmful gender norms, etc. A young woman growing up in rural Afghanistan, for example, faces a very different set of challenges than a young woman growing up in a high-density settlement in Nairobi. Therefore, the way agency is exercised in each of these settings will manifest differently. So while agentic capacity might predict the likelihood a person exercises her agency, the form of agency exercised and the appearance of that agency to an external observer is determined by context.
So how should we measure agency? To know when, why, and how a girl might exercise agency, we need to understand both her agentic capacity and her context. These present two very different challenges for measurement and practice.
We argue that measuring agentic capacity is an exercise in maximizing commonality across differences. What are the most widely applicable attributes and competencies that make up agentic capacity? Much like self-esteem, these are likely to be more similar than divergent across geographies. On the other hand, ‘context’ is necessarily specific to place and is likely best understood qualitatively. Thus, any evaluation of agency must encompass measures that pertain to its internal attributes (agentic capacity) and its specific social or cultural environment (context) as both contribute to the ultimate outcome of interest: the actual act of exercising agency.
Which brings us to the million dollar question: if agentic capacity cannot be directly observed – only the exercise of agency - can we measure it?
Fortunately, there is an entire field developed for this exact purpose. Psychometrics – coined from the Greek words for mental and measurement – refers to the field devoted to developing methods to measure unobservable things (such as psychological constructs). Often, this work results in tests, assessments and other measurement devices that can serve multiple purposes in research, evaluation, and practice. The education sector is full of such tests measuring IQ, EQ, emotional intelligence, etc. A good test is one that proves valid for the purpose it is designed to meet. A test is not “valid” or “invalid” but, rather, valid for a specific purpose. There is a reason why the Myers-Briggs test is now famous. It is not because Katherine Briggs and Isabelle Myers developed a ‘valid test,’but because their test was incredibly accurate at measuring personality types in a way that is highly instructive for professional and personal success. Its validity is tied to that envisioned setting and purpose.
The role of test developers is to gather evidence that provides insight as to whether a test is valid for its intended purpose. Experts in the field recommend gathering validity evidence (both qualitative and quantitative) based on many different sources including content ‘experts,’ statistical performance compared to other similar tests, narrative evidence from test takers, etc. (see AERA/APA/NCME standards). Lately, the field has expanded its focus to include notions of social justice and equity (e.g. see JAV by Randall et.al). Overall, a test that is valid for a purpose meets certain standards of rigorous evidence, and depending on the purpose, more or less emphasis is placed on different types of evidence.
What aspects of test-purpose matter? If an instrument is developed to test agentic capacity, what do we mean then, by purpose?
Tests can be designed for the purpose of measuring change (e.g. pre/post tests to account for program effectiveness) or for diagnostic purposes (e.g. identifying who is in need of an intervention or not–such as IQ tests). Tests can also be intended to measure commonality across place or specificity in context.
First, if a tool is intended to be used across contexts, then it is designed to capture the commonalities of the construct across contexts, not the differences. There are many examples of such tests. In the case of agency, Zimmerman et al (2019) found common patterns in the psychometric properties of three dimensions of agency among adolescents 10-14 years old across 14 countries. Elements of the Rosenberg Self-esteem scale have been validated for use in dozens of countries and its limitations also clearly identified (Schmitt & Alik 2005). Our own work in psychometrics at the Girls’ Agency Lab, AMPLIFY Girls, and Room to Read suggests that internal psychological capacities can be compared across regions up to a point and our efforts in psychometric testing should focus on identifying the limits of comparability rather than assuming comparability is not possible. Why? Recent advances in cross-cultural testing and psychometrics, provide a gamut of options to design tools that are comparable across contexts relying heavily on qualitative methodologies, participatory and inductive processes.
Fortunately, the science of psychometrics equips us with tools to understand whether a measure is reasonable and appropriate for our needs or where it falls short.
In the interests of not only capturing a particular psychometric construct, but in best serving adolescent girls to navigate real-world challenges, we must make use of all the tools at our disposal for measuring agency (and, subsequently, understanding how to increase it). This means that we need both qualitative efforts to understand how context shapes agentic capacity, like those of the Learning and Action Alliance for Girls Agency (LAAGA) at the Brookings Institution, in addition to the quantitative efforts to understand the regional commonality of agentic capacity like the Adolescent Girls’ Agency Survey (AGAS) at the Girls’ Agency Lab (GAL) and AMPLIFY Girls, or the related efforts of the Adolescent Life Skills Assessment for Girls (ALSA) tool at Room to Read. The two approaches are not mutually exclusive, in fact, they are mutually dependent for understanding girls’ agency in a manner that serves both scientific study and the reality of adolescent girls.
A Note to the Girls' Education Community on Agency
A Note to the Girls' Education Community on Agency: Agentic Capacity and Structural Constraints to Agency are Distinct Phenomena