Female participation and attrition in open online courses in Latin America

• In Event: Equity issues in Latin American higher education

Proposal

Sustainable development requires gender equality, both as a goal in and of itself, as well to generate innovative solutions to entrenched problems by tapping half of the world’s talent pool. In addition to SDG 5, other goals such as fighting climate change, promoting health, building sustainable communities, and ensuring access to clean water will require innovative solutions in science and policy. Without investing in the entire population’s potential as scientists and decision-makers through high-quality education, we are unlikely to meet the global goals (Stevens, 2010).
But despite the urgency of achieving a sustainable future, demand for higher education continues to outpace enrollment, and in low- and middle-income countries women and disadvantaged groups face large disparities in access to higher learning (UNESCO GMR, 2017). While women often participate in higher education in equal numbers to men, their participation in scientific research is lacking and they remain a minority of skilled workers in climate-change related sectors (UNESCO, 2015). Specifically, while Latin America and Caribbean regions top the world in female participation in scientific fields, participation is declining (UNESCO, 2015). Overall, women are more likely to participate in scientific fields related to health but remain a small minority in engineering and computer science, and studies of these industries in Latin America find that women face a glass ceiling (Bonder, 2015; UNESCO, 2015; ECLAC, 2014).
Massive Open Online Courses (MOOCs) have been touted as one solution to closing gaps in access to higher education for under-represented students, as the founder of EdX promised “MOOCs make education borderless, gender-blind, race-blind, class-blind and bank account-blind” (Agarwal, 2013). Many studies have found that MOOCs are dominated by the world’s most advantaged, concentrated in high-income countries, they are more likely to be highly educated, affluent and male (Christensen et al., 2013; Chuang & Ho, 2016; Hansen & Reich, 2015; Kizilcec et al., 2013; van de Oudeweetering, & Agirdag, 2018). On the other hand, MOOC users are much more varied than a traditional higher education classroom both in their demographic characteristics and motivations, and large numbers of users do come from the global south (DeBoer et al., 2014; Schmid et al., 2015; Garrido et al., 2016; Jiang et al., 2016).
Online learning is on the rise in Latin America, as users from these countries access MOOCs from foreign institutions such as EdX and Coursera, and as universities in the region turn to ICTs to meet increasing demand (OECD, 2015). However, it is not well understood how sub-groups of users such as women do or do not benefit from such courses, and how contextual factors related to culture, course type, and demographic characteristics may predict their success.
This proposal aims to help close this gap in understanding by looking to nearly 700,000 users of massive open online courses in the Latin American and Caribbean region, 30% of whom are women. Previous work has shown that users from less-developed countries or disadvantaged backgrounds face stereotype threat that inhibits their success in Western-provided online courses, and others have looked to how course design that does not account for women’s cognitive development can disadvantage female users (Kizilcec et al., 2013; Kizilcec et al., 2017; Byrne, 2018). Other work has found women are more successful in STEM courses in certain countries based on their economic conditions (Jiang et al., 2016).
In this study, I investigate these theories in the Latin American context. Previous studies of online learning environments have often suffered from relatively small sample sizes, restricting their ability to look at specific sub-groups and country-level variation. By employing a large dataset of users across six years I am able to describe trends in usage of these MOOCs as well as estimate factors that correlate with completion or attrition in open courses.
The study is guided by two questions:
1) What describes female users of MOOCs in Latin America, and how might they differ from male users?
2) What predicts the attrition of MOOC users in Latin America? Are women more or less likely to stop out of open courses than men, and how do the patterns vary by course and country context?
To answer the first question, I calculate a number of descriptive statistics of users as a whole as well as country-level statistics. In preliminary analysis I find that women are more likely to enroll in courses within health, government, and social science while men are more likely to take computer science courses. I also find that the strongest predictor of certification rates within a country is its human development index, and that the gender gap index is not a significant predictor of female certification rates at the country level. I do find, however, that female certification rates are higher in countries where more women with less than a college education are enrolled, rather than those where the users are already highly educated. This trend is stronger in Latin America than in other regions.
In order to further investigate these relationships, I answer the second question using discrete-time survival analysis (Singer & Willett, 2003). This method estimates the likelihood of users stopping engagement with a MOOC from week-to-week. It this provides conditional probabilities of women versus men, and utilizing random effects and a multi-level approach, I can model whether the relationship between gender and attrition varies within or between countries and courses.
These results provide a more holistic picture of women as users of open online courses, and while it is entirely descriptive in nature, the results can illustrate a more nuanced picture of how contextual factors like gender, course type and geography intersect in online learning environments. This can help inform future research into course design as well as policies that are targeted at increasing provision of higher education through ICTs, particularly for women. Better understanding how sub-groups of learners may be impacted by contexts should be informative for those who are committed to improving higher education opportunities and equitable access to careers that will improve sustainable development.

Authors

• Eileen McGivney, Harvard Graduate School of Education

• Andrew Ho, Harvard