One size fits all?: On the merits of disaggregating data
Updated: May 19, 2022
Originally published April 10, 2020
Since 2017, our study of how diverse international schools impact students and impact the world has included the 18 UWC colleges as well as 11 non-UWC schools. We are using a mixed-methods design (both quantitative research, which aims to measure data numerically in order to generalize to a larger population, and qualitative research, which aims to describe the complexities and nuances of certain observations) in order to investigate various dimensions of this impact. This means that, quantitatively, over the past three years we have asked students and alumni from all of our participating schools to complete surveys probing their educational experiences, moral and intellectual dispositions, and ways of making a difference in the world. Qualitatively, we’ve spent time engaging in one-on-one conversations with students, alumni, teachers, and staff across the UWC movement as well as conducting observations of classes and activities.
The surveys for this study were particularly challenging to develop as we attempted to find ways to capture the ethos of each of the 29 schools in the study, while also making the survey applicable broadly enough to be generalizable to the field of international education at large. This question became particularly salient when thinking about whether, and how, to ask about participants’ demographics on the surveys, such as questions about race, ethnicity, and socioeconomic status.
Throughout survey development we were sensitive to the various issues that arise with demographic questions and the arguments that people might make against the inclusion of demographic questions altogether. These reasons include the possibility of appearing reductionist regarding racial/ethnic identities or other identity groupings, the potential for stereotype threat, the possibility of inserting a U.S or Western bias into questions, the danger of results that show differences between demographic groups being misinterpreted or oversimplified, and more.
Why, then, did we decide to include such questions in our surveys? Indeed, in both our student and alumni surveys we did ask participants about their gender identity, racial and ethnic identities, socioeconomic status, religious identity, national identity, regional identity, and more. To many it must have seemed excessive, and, as with much quantitative survey research, far too reductive to capture the nuances of the participants in this research study. (And indeed, the study population is incredibly diverse across the aforementioned metrics.)
While we would agree that such questions and the analyses derived from them can seem reductive, we still argue that for any educators who are interested in questions of equity in education—as we are—such questions cannot be summarily thrown out. In this study, we are trying to understand which educational practices bring about certain outcomes for students, so it is prudent that we investigate to the best of our abilities whether our findings hold true for different groups of people. Otherwise, we run the risk of applying a one size fits all solution when, in reality, one size may only fit one size.
For example, in Invisible Women, Caroline Criado Perez offers a thorough glimpse into the ways in which failing to disaggregate data—in her case, by gender—can lead to inequitable distributions of resources and outcomes for females: many city planners have traditionally failed to disaggregate data, thereby failing to account for the different ways women move through their environment when planning transportation options; non-disaggregated data has created tech recruitment algorithms based on male-normed patterns of behaviors, effectively guaranteeing a lack of female applicants; and obscured data has led to adjusted maternity leave policies and academic tenure policies around the world that fail to take into account the physical recovery time often needed by mothers. In sum, as Criado Perez argues, the lack of disaggregating data by gender in many different fields has led to an ongoing gender disparity in not only wages, but in many other fields of life.
Certainly such inequities are not always deliberate, and such is often—and hopefully—the case in educational settings. But, if we are to investigate whether an educational institution is achieving its educational mission and aims, should we not also investigate if it is doing this equitably for all of its students?
There are a variety of ways for us to disaggregate our data based on the data we collected from participants. Some of the key breakdowns we are considering investigating include:
Race, ethnicity, or region of the world
Fee-paying students and non-fee-paying students
Boarding students and day students
Diploma programme students and students who have attended longer than 2 years
While outcomes of the study could potentially vary by the above groupings, there are also more practical aspects of the study, such as response patterns or logistical concerns, that disaggregating data could shed light on. Perhaps female students were more likely to complete the survey, or male students were more likely to consent to participate. Perhaps those from certain areas of the world struggled more with the language of the survey and this should be taken into account when analyzing the data. It is only through taking a look at these groupings and breakdowns that we are able to know the answers to such questions.
A great strength of this project is its mixed-methods methodology; we are not limited to only being able to look at qualitative or quantitative data, but can let data from one methodology inform our analyses of data from the other. Thus, while quantitative methods always involve some degree of simplification and categorization that may feel incomplete to participants, especially when questions are related to complex experiences and identities, we can turn to the in-depth information from our personal conversations and field observations to make sense of the more complex nuances of this study. For example, we might find differences on the surveys in the types of school experiences students from different regions of the world report as important to them. We could then turn to our interviews to further understand why those differences might exist. In the meantime, disaggregating the quantitative survey data will move us forward in understanding how we can make internationally-oriented education more equitable for all.