Apps, also known as mobile applications, have become a relatively new media format, developing parallel to the increasing popularity of smartphones throughout the last 2 decades. They are distributed and downloaded from a variety of so-called ‘app stores’ which represent some of the world’s largest software shops currently in the air for commercial use. The two most popular and widely-used stores are Apple’s App Store and Google’s Play Store, and both boast an extensive range of over two million applications and have content added by a wide variety of contributors across the globe on a daily basis (Blair, 2019). Combined, the two mobile marketplaces represented a staggering 105 billion downloads in 2018 (Iqbal, 2019).
Varying in both function and price, the apps are globally available to almost every person owning a smartphone and play diverse roles in users’ lives and daily practices. Morris & Elkins state that the small units of software “insinuate themselves into our routines and habits”, making apps so-called ‘mundane software’ (2015). Currently making up for a significant amount of their daily spent time, app users can be said to partly live through the lens of their mobile applications and app stores of choice. Every new addition to a user’s inventory of apps holds the potential to functionally contribute to whatever aspect of life one sees fit. From translators to chess to nutrition scheduling, there’s an app for that.
As the small pieces of software are becoming increasingly discussed objects of study, the relatively new media format and its corresponding stores do not fail to facilitate a wide range of academic potential. The last five years have even seen the University of Amsterdam present an annual research on the format, with 2020 not being an exception. While app stores play host to the apps, discourse and meaning is implicitly loaded as apps can be seen as sociocultural products. Therefore, app stores as arguably neutral media spheres affect to billions of lives on a daily basis and can be studied like active shapers of future practices and knowledge (Lupin, 2014). Considering these aspects, the way controversial topics might or might not be represented within the stores seems definitely worthy of further research.
Therefore as part of this year’s Winter School Data Sprint, and to conform to the overall direction of our project group concerning Apps and Their Practices, we have thus decided to look into what role gender bias possibly plays within ‘the health & fitness’ category of mobile app stores. With western global pay gaps still in place and biased workplace hierarchies just as pertinent, media spheres are definitely not left out when it comes to gender bias (Connson, 2019). Not to mention the recent surge in worldwide discussions surrounding gender bias, our direction of research was initially fueled by earlier outings of sexism and gender bias within app stores like the overload of disempowering period tracking apps and an app for monitoring Saudi women that are at risk of leaving the country (Hall, 2017; Owen, 2019).
To shed light on a new aspect of the practices surrounding apps and their stores, our project group has made efforts to extract data from the two largest app stores currently available and, as a result, present a variety of details surrounding the concerned app category.
How are gender roles represented in health- and fitness applications across app stores in the top 10 countries with the highest smartphone usage worldwide?
Multiple methods have been applied to this research in order to gain relevant data fitting the research question, among which quantitative as well as qualitative methods, overall a mixed-method approach. In order to determine a usable scope of the research, a valuable dataset was to be gathered. Firstly, a list of health & fitness apps was to be composed. Ultimately, this list was found at App Annie, a platform for app analytics and market data, in the ‘App Store Rankings’ section’s top charts, filtered on the ‘Health & Fitness’ category. In order to narrow this list down to a useful dataset, the decision was made to focus specifically on free apps as opposed to paid or subscription apps, as free apps are accessible for everyone with a smartphone and downloaded more frequently.
Secondly, in order to narrow our research scope down, the demographics were to be defined, for which Statista’s list of top countries sorted by smartphone usage was consulted. These countries were limited to the top 10, which consisted of (in descending order): China, India, the United States, Brazil, Russia, Indonesia, Japan, Mexico, Germany and the United Kingdom (Statista 2019).
As the research focuses primarily on a cross-application analysis on gender bias, it was important to create a dataset for multiple app stores, as these act as app spaces on meta level. It was decided to limit these to the Apple iOS App Store and the Google Play Store, as these are by far the two biggest app stores.
Furthermore, we decided to look at the top 20 free applications per country per app store. Ultimately, this led to an analysis of 400 applications, from which the following characteristics were collected: app name, app description (practices), whether the application was predominantly for male users (M), female users (F) or both/non-applicable (X), the color of the app icon, whether the app icon displayed a male (M), female (F), or none, and if a subscription was possible – if so, how much this would cost on a monthly basis.
In order to obtain app metadata, the app names and rankings were extracted from App Annie. For the iOS store, a list with the app names with “site:https://apps.apple.com” added behind every name was put in the Digital Methods Initiative (DMI) Lippmannian Device in order to obtain the iOS store ID’s. Subsequently, this data was converted to a .CSV file which was opened and edited in Excel with split cells, from which the app store ID was then extracted. For the Play Store, the DMI Link Ripper Tool was used in order to obtain the app ID’s.
By running the obtained ID’s through the Play Store and iOS store, additional data (such as the app description and app icon) could be gathered. The app description consisted of textual information and screenshots.
After collecting the data and assembling these in Excel, streamlining of the coded practices was next step. We split up app practices into individual segments, having all the listed application practices lined up per app and per country (in appendix). Consequently, each individual app practice was linked to a standardized category. By doing so, we were able to distinguish patterns within countries in regards to the distribution of gender representation.
Finally, there was a total of 22 categories, distributed over six main categories (tracking, guidance, consultancy, community, medical and beauty). Tracking included the following subcategories: tracking activity, tracking exercise, tracking nutrition, tracking sleep, tracking weight, tracking menstruation, tracking pregnancy, tracking water, tracking mental health and tracking intimacy. Guidance included the following subcategories: guidance diet, guidance meditation, guidance work-out, guidance planning, guidance parenting, guidance mental health, guidance sleep, guidance spiritual.
Each standardized category of a particular application was linked to the targeted gender of the application, thus showing how prominent certain application practices are within each gender. This allowed us to visualize these standardized codes with their respective gender link in a matrix, resulting in a visualization in which the popularity of application practices is ranked from most prominent to least prominent across the total pool of applications. The other axis of the visualization is representative for the countries that we used as sample groups, which are also distinguished. This allows us to analyze application practices across countries.
Considering the vastness of the data set that we collected, one of the big limitations of this research project was that we did not have enough time to digest all the data into a usable form and subsequently use it to draw conclusions in regards to gender representation. Data such as the subscription fees were therefore disregarded in our findings and discussion. Another limitation that we encountered was that China and Japan had a selection of applications in their top charts with their Mandarin and Japanese, respectively, in the app title and description. This could have potentially lost some of the nuances of the language when translating to english, thus possibly letting us miss some of their fundamental application practices. Lastly, the coding process was done by four people. Though we applied thematic coding to keywords related to application practices, having a clear (binary) code book may have been beneficial to the large data set. This way to reproducibility would have been higher and thus also the reliability of our results.
Having gathered and coded the data for the top charted application within the health and fitness application across the ten countries with most smartphone usage, we are left with a lot of data to interpret. Though our research question specifically questions the representation of gender within the most downloaded health and fitness applications throughout the world, to put this question it is also important to observe trends on a bigger scale. This allows us to situate our research question in a larger perspective.
In lieu of this, our first finding pertains to the differences between the application stores. Throughout our coding, we noticed a difference in the practices that applications had between the iOS App Store and the Google Play Store. Table 1 in the appendix highlights this difference. In particular an interesting difference to note was that the Play Store had more practices related to the “fitness” side of the category than the health side. The application practices “Guidance work-out, Tracking exercise, Tracking activity, Tracking weight” constituted 39% of the application practices within the Google Play Store and 31% in the iOS App Store. In turn, the iOS Store had a lot more app practices related to mental health (14%). In the Play Store this was only half of that (7%). This could indicate a characteristic of the population who make use of both these application stores. Whereas users of the Play Store may make use of their phones more often during fitness, users of Apple products may be involving their devices more for the benefit of their mental state of mind. In both application stores the most common application practice within the top downloaded apps was a community function, relating to the ability of interconnecting with other users or apps outside of the application itself.
With regards to the representation of gender within the apps that we analyzed we were able to observe some interesting trends. Visible in table 2, in the top 5 most prominent app practices, the overall spread of female, male, or other targeted genders are the most balanced.
Table 2: Application of all practices organized per country and subsequently gender, ranked by most common to least common
As the app practices become less common across applications, applications that are geared towards males become less. What was also observable is that during the coding process, there were no male-specific app practices found. Whereas this was the case for females. Applications to track one’s menstrual cycle for example were so common that the practices could be categorized together (tracking menstruation). Male applications were solely geared towards promoting ways to make physical changes to one’s body through exercise. Furthermore, another notable finding was that the application practices linked to mental health were sometimes female but never male. For example, in Germany the application “Period Tracker Flo. Ovulation Calendar & Pregnancy” had a tracker for mental health as a practice within this female oriented application. In no male targeted application were there traces of mental health practices found.
On a cross country scale, other observations can be made. In specific, application practices in regards to targeted gender can be compared across the 10 different countries analyzed. Looking at the country with the most gender-neutral application practices, the UK has the highest percentage in this category with 88% of the application practices in their country belonging to gender-neutral applications. At the other side of this question is Mexico in which only 53% of application practices were related to gender neutral targeted applications. Mexico in turn also had the highest number of app practices belonging to female targeted applications that amounted to 43% of their total coded application practices. Lastly, both the UK nor Japan has application practices that pertained to any male oriented applications. This suggests that in both cases male health and fitness application were not as popular in either application stores.
Although the results of the research are quite straightforward and definitely succeed in representing gender bias, there are multiple ways of interpreting this data. Principally, the fact that some countries show a higher division in the binary gender distribution is evident. In order to further research this, there are three main critiques that are significant. The first is globalization critique. What does it mean that certain applications and the gender that these are targeted at, appear in the top 20 of most downloaded apps, on a regional or local level? How many local applications are even found in the top 20? Apps are able to reflect a great deal of country-specific norms and values, and the polishing of app markets increases the reliance on global systems. Globalization critique therefore merely is based on critique on capitalism, cultural homogeneity and cultural imperialism – in which power relations between the dominant and the peripheral model technological infrastructures in ways that only benefit the dominant. By means of this, a significant question for further research is to look at mobile applications from a post-globalist point of view.
The second critique consists of an app store-specific critique, as the research pointed out some clear differences between the iOS App Store and the Play Store. How do the rankings per app store differ? Which app store would, for example, offer a greater supply of paid subscriptions, or enable an environment in which subscriptions are encouraged? Exploring gender bias within these hypotheses could add an interesting economical perspective to this research. This economical perspective could be extended to a cross-country analysis as well, as a striking finding was that across countries, prices for subscriptions varied a great deal.
The third critique is a cultural critique: this research laid the basis for a cross-country analysis for gender bias across the top 20 apps per country, but this could be further deepened to a cross-cultural critique. Across some countries female health applications for tracking menstrual cycles or pregnancy were much more prominent than in others, as well as beautification applications or male fitness apps. A linguistic analysis could be helpful in order to gain more cultural data. What terms and definitions are mostly associated with males or females? How is this embedded in different cultures? For example, in China, India, Indonesia and Mexico, an application called Height Increase - Increase Height Workout, Taller, a male-targeted workout guidance application for exercises that would increase height, is amongst the top 20 applications, which is not to be found in other countries. What cultural logic is behind this? In Western countries, people generally are taller than in East-Asian countries, but a general statement which often is heard is that ‘taller men would be seen as more attractive’ as well (Case & Paxson 2006). How are these cultural codes embedded in applications and the countries that these are most favored in? And thus, what stereotypes do developers accumulate in software interfaces?
A final approach to further research could be to look at the acquired data from a queer perspective. Within this research, the focus was primarily on the gender binary of male/female, but as this excludes many expressions and forms of gender, it could lay the foundation for a queer critique on app spaces.
This project attempted to define in what ways binary genders are represented in the applications that people download on their phones. Looking at the top charts of the applications within the health and fitness category allowed this project to understand what qualities within the presentation of applications in both the iOS and the Google Play application stores were favoured. There were significantly more female targeted applications than male targeted applications found during the research which is relevant to the the overall indication of the application stores representation of gender. Though this could possibly indicate that the female applications are more popular to download, or male applications might be the norm rather than the outlier. When looking at the amount of “gender neutral” applications, it could be that health and fitness application that are gender neutral may have more visualisations that portray men because that has become the norm, thus painting a biased picture toward the female gender.
Furthermore, because most applications are similar in the top charts, it is safe to say (as mentioned earlier) that there is a certain globalization trend at play. Globalization could then in turn influence the gender norms that are perceived on not only a national, but also a global scale. The repercussions of this could possibly take gender norms to a new level, displaying even more stereotypes throughout the app stores. By buying and downloading applications in these stores, a behaviour that has become so integrated into daily lives, users are subconsciously playing into these gender normative roles that developers chose to give the applications they create.
Though this project has just scratched the surface of the possibilities of studying app stores, gender inequality, stereotypes and exclusion are important factors to keep in mind while downloading an application in the future. In a subsequent project it might be interesting to do a (digital) ethnographic research of the people that the applications within the health and fitness category was intended for and the actual users of these applications.
Blair, I. (2019, January 8). Ultimate Mobile App Stores List (2019). Retrieved from https://buildfire.com/mobile-app-stores-list/
Case, A., & Paxson, C. (2006). Stature and Status: Height, Ability, and Labor Market Outcomes. doi: 10.3386/w12466
Hall, M. (2017, July 25). The Strange Sexism of Period Apps. Retrieved from https://motherboard.vice.com/en_us/article/qvp5yd/the-strange-sexism-of-period-apps.
Iqbal, M. (2019, November 19). App Download and Usage Statistics (2019). Retrieved from http://www.businessofapps.com/data/app-statistics/
Locke, C. (2019, July 5). Why Gender Bias Still Occurs And What We Can Do About It. Retrieved from http://www.forbes.com/sites/londonschoolofeconomics/2019/07/05/why-gender-bias-still-occurs-and-what-we-can-do-about-it/#4edee3925228.
Lupton, D. (2014). Apps as Artefacts: Towards a Critical Perspective on Mobile Health and Medical Apps. Societies, 4(4), 606–622. doi: 10.3390/soc4040606
Morris, J. W., & Elkins, E. (2015). FCJ-181 There’s a History for That: Apps and Mundane Software as Commodity. The Fibreculture Journal, (25), 63–88. doi: 10.15307/fcj.25.181.2015
Owen, M. (2019, February 8). Apple & Google accused of supporting 'gender apartheid' by hosting Saudi woman-tracking app in App Store. Retrieved from https://appleinsider.com/articles/19/02/08/apple-google-accused-of-supporting-gender-apartheid-by-hosting-saudi-woman-tracking-app-in-app-store
Table 1: Store comparison of most common application practices