This project is maintained by amartrics
Final Word Count: 2470
In Amartya Sen’s book Development as Freedom, the famed economist discusses the tenuous balance that often arises between culture and progress on the path towards human flourishing and development. Though these two processes can exist in harmony, allowing for the cultivation of a well-functioning, socially-minded society, danger rears its head when the progression of one impedes on the other. After all, development itself depends entirely on the growth of human capabilities. When tradition begins to impinge upon the freedoms of minority groups, there is no longer leeway for historic norms to continually dictate the day-to-day life of an entire population. In recent decades, this conflict between preservation and developmental progression has risen in prominence as a topic of interest for activists, development experts, and data scientists in Southeast Asia, where longstanding patriarchal precepts dominate women’s social interactions, educational opportunities, economic activity, access to healthcare, political engagement, and human security. Top-down approaches to analyzing the statistics related to this issue, such as national censuses and household surveys, have already created an unsettling image of inequity. A brief examination of South Asia’s employment rates shows the shockingly low percentage (roughly 20%) of female workers in Pakistan, India, and Bangladesh [1]. But utilizing a broad brush, so to speak, is not an acceptable mode of analysis in this situation, where minute cultural shifts between zip codes can have a massive impact on the probability of a young girl receiving more than an elementary-level education. With significant gaps in gender-disaggregated data clouding the extent of gender inequality in areas like Northern India, Nepal, Pakistan, Bhutan, and Sri Lanka, it is difficult to ascertain whether the general populace of the region is meeting any of the United Nations’ Sustainable Development Goals related to gender equality. A substantial foundation of cross-tabulable, granular, high-quality, and high-coverage data is necessary to understand these countries’ progress towards gender-focused SDGs, especially as they relate to economic opportunities and female health & well-being.
New research and the testing of cutting-edge technologies suggests it is possible to create such a database by synthesizing priori from existing surveys with novel mapping techniques. Data imaging produced by a combination of Bayesian generalized linear models, artificial neural networks, and repeated cross-validation processes could be particularly useful in generating accurate representations of social, health, and economic phenomena across both tiny pockets and large swathes of land [2]. Before solutions to Southeast Asia’s massive wage gaps, poor rates of political participation and representation amongst women, belittled scholastic expectations and opportunities for girls, and lacking access to healthcare – especially in regards to female sexual health and domestic abuse treatment – are even proposed, it is vital to gain a complete and realistic understanding of spatially-diverse gender inequity. Recent data collected from social media, messaging platforms, and other methods of wireless communication through keyword-focused machine-learning algorithms demonstrates that over half of Southeast Asia’s population is decidedly uninvested in this particular facet of social change [3]. Without enough support from the human-driven side of development, self-supporting data is more relevant and necessary than ever in order to ground influential policymakers in reality, flip the stigmas associated with social transformation in rigid, unmoving systems like those found in Southeast Asia, and prompt individuals to act as agents of genuine progress.
One of the clearest, albeit crudest, indicators of gender-based duality in developmental progress is the wage gap. Across the world, women earn significantly less than men, with the average variance in rates of pay on a global scale clocking in at almost twenty per cent [4]. In Southeast Asia, this gendered distance in earnings is even greater; Nepal, for example, boasts an estimated imbalance of 28% between male and female incomes. Simply identifying this issue, however, does not explain its roots or the reasoning behind the correlative unfreedoms that accompany gender-disaggregated market activity on sub- and international levels. In “Gender (In)Equality in South Asia: Problems, Prospects and Pathways,” published in the South Asian Journal of Human Resources in 2015, authors Glenda Strachan, Arosha Adikaram, and Pavithra Kailsapathy define this issue as a lack of deep analysis concerning localized fiscal interactions, specifically those skewed by gender. According to Strachan and her cohorts, the importance of considering economic inequality alongside its tangential effects within the realm of development is often overlooked, leading to a deficit of legitimately relevant data. For researchers and deep learning programs attempting to interpret and enumerate the causation of explicit phenomena, this results in an inability to accurately replicate and, by extension, extrapolate the incubatory circumstances of such phenomena. Since much of the information regarding localized economic activity, especially in less developed regions of the world, is derived from national surveys without tangible foci, unreliable earnings reports from businesses that can easily misreport or skew data, and informal polls of consolidated households that generally do not divide the economic activity of members (e.g. differentiation between spousal earnings) [1], even the most well-funded of projects and organizations have routinely experienced difficulty in defining realistic, quantitative parameters of gender inequality. Incalculable rates of informal employment amongst women as well as widespread ignorance of the inherent, exclusively-female barrier to equal opportunities in the job market posed by gendered domestic roles and lacking child care services in developing countries further complicate the process of analyzing the impact of gender bias on development, as the data needed to quantify this information is not considered of import by most governments and thus, is generally nonexistent [5].
What is already known, however, can still be used to create estimates and predictive benchmarks of fiscal health for Southeast Asian women; a perfect understanding of this particular population’s level of interaction with the global market is not necessary to link, and subsequently understand, the process’ causal factors to other realms of development. For instance, researchers at Data2x, an organization that “aims to advance gender equality and women’s empowerment through improved data collection… [to] guide development policy,” published a report in 2014 detailing the interconnectivity of gendered development issues across economic, social, educational, and health boundaries. According to their research, some of the biggest indicators of future unemployment, particularly as a result of early marriage and adolescent pregnancy, stem from the troubles young women encounter during a key period of life: the transition from academic to working environments. By utilizing disaggregated rates of enrollment collected through informal methods, such as computerized recordkeeping in individual school districts, on a local scale and preexisting initiatives like UNESCO’s Learning Metrics Task Force on an international scale, future projects focused on women’s shifting social roles in Southeast Asia can easily obtain a rich pool of priori for programs analyzing the ties between levels of scholastic and economic engagement amongst women.
Quantifying gender-based discrimination in the context of a country’s economic and educational capabilities can help provide a baseline understanding of inequality, but will inherently fall short in creating an accurate image of women’s physical and mental wellbeing. Economic opportunities have been historically linked to mental health benefits – people with the freedom to pursue their own occupations report, on average, a higher sense of fulfillment, structure, and social engagement [6] in their day-to-day lives. Southeast Asian women barred from the workplace suffer not only as a result of lacking control over fiscal resources, but also from the psychological effects of social isolation and domestic expectations of submission. The cycle of gender discrimination is thus compounded: female children are viewed as economic burdens from birth because it is expected that they will not engage in economic activity, which further encourages the undervaluation of girls & women as a homogenous group. To better estimate the impacts of this aspect of development on women’s mental health, a database of detailed birth & death records, participation in health surveys, and records of patient monitoring disaggregated by gender must be cultivated. However, in areas already lacking in infrastructural support, taking the time and effort to increase the amount of available data on traditionally underrepresented populations may seem daunting, futile, or, in certain cases, entirely contrary to patriarchal norms and thus, a generally useless endeavor. In these situations, it is often up to external forces to organize the collection of gendered data through indirect methods. A comprehensive report on this issue authored by Bapu Vaitla, PhD, a visiting scientist at Harvard’s T.H. Chan School of Public Health and a fellow at Data2x, utilizes the results of several unique data collection methods to demonstrate the staggering imbalances between the “well-being” of men and women in the region of Southeast Asia. Though Vaitla’s article focuses broadly on the implementation of “big data” for the furthering of gender equity in southeast Asia, he brings specific attention to the challenges associated with the tracking of women and girls’ well-being in regions where social taboos prevent the discussion of gender-based disparity. Analyzing the results of a post-2015 study conducted by the United Nation’s Global Pulse program on the geographic clustering of certain topics in online conversation over social media platforms, Vaitla’s publication concisely merges development and data by mapping gender-based disparities in mental wellness, crime and violence, and access to healthcare within the region of Southeast Asia. By pinpointing hotspots of inequality between men and women on a geographic scale, Vaitla’s project acts as a trendsetting example of how “big data” can aid in quantifying development concepts formerly viewed as purely qualitative, such as sexist social stratification in “third-world” countries.
For this very reason, social media has become an irreplaceable tool in the process of understanding gender inequality, allowing unfiltered and geotagged conversations about topics normally censored by the presence of husbands, in-laws, and unspoken social policing to occur organically. In countries like Nepal, where the female-to-male ratio of psychiatric morbidity is 2.8:1, and Bangladesh, where the average woman is three times more likely to commit suicide than the average man [6], capitalizing on the removal of these social hurdles in a developmental context has the potential to save lives, as demonstrated by the second portion of Vaitla’s report. Beyond creating real-time maps of health and crime disparity based on online testimonials, Dr. Vaitla and his cohorts were also able to produce a machine-learning algorithm that can “accurately identify mental illness” through “genuine self-disclosures” posted on online social platforms with a 96% success rate [3]. Analytics collected from the algorithm, which compiled a pool of gender-disaggregated data from recorded frequencies of keywords delegated as indicators of suicidal intention, depressive behavior, or flat-out admittances of mental illness, show that women in general are 15.4% more likely to suffer from debilitating sadness or depression and 10.7% more likely to experience anxiety than men, with even higher rates of disparity in less-developed countries. This trend – a statistical increase in incidents of depression, suicide, anxiety – correlates with the negation of the female “biological advantage” of longer lifespans and lower rates of disease [7] that occurs in the same countries where women’s mental wellbeing is overwhelmingly poor.
Currently, data and social scientists are still working with far too little data to development applicable solutions to gender inequality. Instead of engineering quick policy fixes to eliminate age-old roadblocks to progress, big data firms and influential government figures should focus on pairing with human rights organizations to improve society’s general understanding of gender-based discrimination on a quantifiable scale. Aforementioned projects such as UNESCO’s Learning Metrics Task Force and the ongoing work of dedicated individuals like Dr. Vaitla offer glimpses of what can be accomplished by this merging of big data and social science. Ultimately, the goal of these endeavors should be to create high-quality and granular databases constructed from regional micro-surveys (e.g., a “bottom-up” approach) that detail in depth the condition of localized, sex-disaggregated development. Projects emulating this approach are already in motion across the globe. For example, in Uganda, a machine-learning program capable of recognizing human speech patterns and intonation is being used to monitor radio stations to collect, filter, and evaluate sound files concerning women’s involvement in politics [8]. A study of three years’ worth of public Twitter posts as a source of “real-time signals of discrimination against women in the workplace” processed by keyword-sensitive neural networks has identified four major factors that block women from occupational flourishing in Indonesia – inherently discriminatory job requirements, the requirement of social permissions from male family members for working, allocations of “appropriateness” to gendered lines of work, and the multiple burdens of women who run their own households [9]. In the specific context of Southeast Asia, a project testing the accuracy of “spatial interpolation methods based on geolocated clusters” has yielded remarkable success, allowing for the high-resolution mapping of gender-disaggregated development based on social, health-related, and economic phenomena in both tiny pockets and large swathes of land [10]. Conducted in 2017 through a grant from the UN Foundation, the study relied on a complex combination of Bayesian generalized linear models, artificial neural networks, and repeated cross-validation processes to estimate and model rates of literacy, income rates, contraceptive use, and child stunting amongst women in Bangladesh. Producing this type of easily-accessible information has the invaluable benefit of populating an empty data field while simultaneously making it available to the people who can contribute to and gain from it the most.
Tracking disparities in wellbeing between men and women is a key part of the United Nation’s Sustainable Development Goal 5, which aims to “[end] all discrimination against women and girls” on the grounds that it “is not only a basic human right,” but also that it “helps economic growth and development” [11]. While this definition effectively boils down a seemingly unquantifiable issue into a few palatable sentences, it also inherently allows for the misinterpretation of simplistic census figures and generalized, external analyses of non-disaggregated data as reflections of progress towards gender equality. Over the course of my research on this topic, I have sought to understand how quantifying the causes of gender-based discrimination – as opposed to continuing to focus solely on its more visible effects – can better data scientists’ understanding of this complex phenomenon, bringing me to the question: is it possible to quantify internalized factors of endemic, gender-based discrimination in a purely unbiased manner? Geospatially-concentrated methods of data analysis are particularly important to consider as catalysts for development in the region of Southeast Asia, where the presence of literally hundreds of unique cultural influences can cause massive variations in social norms between one small province and another. Besides utilizing the predictive and relatively unbiased data science methodologies detailed in previous portions of this review to illustrate gender-inequity in real-time and on a geo-specific scale, development experts can also work on increasing published literature regarding this issue through involving more women in the organization of future projects. By applying an integrated method of human and machine analysis to data collection, the gender-based gap in data collection may finally be bridged, allowing real-world disparities in women’s social, economic, physical, and mental health to be alleviated as well.
References [1] Strachan, Glenda, et al. “Gender (In)Equality in South Asia: Problems, Prospects and Pathways.” South Asian Journal of Human Resources Management, 2015. https://doi.org/10.1177/2322093715580222.
[2] Bosco, C., et al. “Exploring the high-resolution mapping of gender-disaggregated development indicators.” J. R. Soc. Interface, 2017. http://dx.doi.org/10.1098/rsif.2016.0825.
[3] Vaitla, Bapu, et al. “Big Data and the Well-Being of Women and Girls – Applications on the Social Scientific Frontier.” Data2x, Apr 2017. https://data2x.org/resource-center/big-data-and-the-wellbeing-of-women-and-girls/.
[4] “Global Wage Report 2018/2019.” International Labor Organization (ILO), 2018. https://www.ilo.org/global/about-the-ilo/multimedia/maps-and-charts/enhanced/WCMS_650829/lang–en/index.htm.
[5] Buvinic, Maya, et al. “Mapping Gender Data Gaps.” Data2x, Mar 2014. https://data2x.org/wp-content/uploads/2019/05/Data2X_MappingGenderDataGaps_FullReport.pdf.
[6] Niaz, Unaiza and Hassan, Sehar. “Culture and mental health of women in South-East Asia.” World Psychiatry, Jun 2006, vol. 5, no. 2, pp. 118-120. PubMed Central, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1525125/.
[7] Fikree, Fariyal F. and Pasha, Omrana. “Role of gender in health disparity: the South Asian context.” The BMJ, Apr 2004, vol. 328, pp. 823-826. PubMed Central, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC383384/.
[8] Lopes, Claudia A. and Bailur, Savita. “Gender Equality and Big Data – Making Gender Data Visible.” United Nations Entity for Gender Equality and the Empowerment of Women, Jan 2018. UN Women, https://www.unwomen.org/en/digital-library/publications/2018/1/gender-equality-and-big-data.
[9] “Feasibility Study: Identifying Trends in Discrimination Against Women in the Workplace In Social Media.” Global Pulse Project Series, 2014, no. 11. UN Global Pulse, https://www.unglobalpulse.org/document/identifying-trends-in-discrimination-against-women-in-the-workplace-in-social-media/.
[10] Bosco, C., et al. “Exploring the high-resolution mapping of gender-disaggregated development indicators.” J. R. Soc. Interface, 2017. http://dx.doi.org/10.1098/rsif.2016.0825.
[11] “Goal 5: Gender equality.” United Nations Development Programme. https://www.undp.org/content/undp/en/home/sustainable-development-goals.html.