MobileActive.org

This guest post was written by Nithya Sambasivan, a Ph.D candidate in the Department of Informatics at the University of California, Irvine. A link to the complete paper (PDF) is here.

I spent a total of three months last summer conducting an ethnographically-inspired study of urban sex workers (USWs), where we designed, implemented and evaluated a phone-based broadcasting system for urban sex workers.   I "hung out" at the solicitation locations and the drop-in shelters of Pragati, called Swati Manne. These Swati Mannes were not only places to rest for the USWs, with beds and TVs, but they also included a medical clinic and a payment area for MFI loans. The internship project was with Microsoft Research India. None of the interviews were audio-recorded. Only hand-written notes were taken.

Development organizations are always looking for ways to reach out to their target populations. Communication infrastructures are crucial in reaching out to these communities for notifications, announcements, advertising, reminders, and emergency services during crises. Communications are also used for disseminating information about welfare services and for effecting behavior change (e.g., public service announcements). We collaborated with one such non-governmental organization, Pragati, in Bangalore, India. The organization is dedicated to assisting its target population—urban sex workers—primarily for healthcare, but also offers ancillary services such as microfinance and counseling.

Briefly, CHWs (Jeevikas and Jananis), who are senior USWs themselves, go out to the field regularly to monitor health checkups, distribute condoms, collect microfinance payments, and offer counseling. Each Jeevika is allotted a particular geographical focus based on her social network, and about 10-15 USWs come under her wing. She travels to the public solicitation areas where the USWs typically solicit, such as bus stops, parks, or lodges. However, the system suffers from inefficiencies such as delays in communication, because Jeevikas have to physically travel to the field to disseminate information and distribute goods.

In our study, we found that the USWs tended to forget the payment dates for their microfinance loans and would skip their monthly medical checkups for sexually transmitted diseases. The USWs we interviewed were a unique population in that they were socially stigmatized and economically poor. However, mobile phone penetration among the USWs was 97%, which is an unusually high number for both: a developing nation population (the average wireless tele-density in India is 55.14% as of July, 2010 and women (who are 37% less likely to own a mobile phone in South Asia, as of February, 2010). Mobile phones are excellent tools for reminders and announcements since they form a pervasive infrastructure; are very effective at delivering short messages; and are portable. For the above-mentioned problem areas, we took into consideration the exceedingly high mobile penetration among USWs and narrowed down on designing a reminder and announcement system.

In coming up with a solution, we wanted to leverage the existing trust among the USWs in the Jeevikas and in Pragati. At the same time, USWs lead unique lifestyles and several factors need to be taken into consideration: they maintain double lives by not revealing their professional lives to their families; instead they tell their families that they work "innocuous" jobs, such as domestic work. Such boundaries in identity seep into technology usage. USWs maintain two separate devices or dual-SIM phones, by turning off the work phone at home or switching SIMs. This phenomenon poses the design issue of choosing the right device for communication and the larger issue of communicating at the right time, for reaching out to a USW when she is at home may cause severe consequences to her life. Finally, the issue of migration needs to be taken into account. Several USWs live out of a suitcase. Unlike populations that cohere through shared spaces, such as neighbors, street-based USWs are either homeless or scattered across the city. There is a pronounced lack of social capital that arises from sharing a geographical space.

Considering the above factors, we decided to build a phone-based broadcasting system. The system would call a list of phones at once, and play a pre-recorded audio message. Some design principles that we chose are:

• Server-side based - to avoid software compatibility issues and not to leave traces of any linking application on the phone.
• Audio-based - most of the USWs were non-literate
• Made use of institutional trust - we recorded the voice of the local field coordinator, a much-liked person.
• Calling at a good time - based on the overlap between street- and home-based USWs, we decided to broadcast the audio messages at late afternoon-evening, when the USWs tended to relax and were not at home.
• Anonymous content - we did not reveal any identifying information, such as names, profession, disease, etc.

We recorded 20-30 second audio messages for four different deployments: 1) a reminder for a new Swati Manne inauguration (pilot test); 2) a reminder for microfinance loan deadlines; 3) an advertisement for HIV testing; and 4) an announcement for a computer training session. We had to carefully balance several factors: noise while recording, such as traffic, phone or construction noise (overcome by noise removal and background music) and be clear about the message without giving out revealing details (for example, upon a suggestion by Pragati staff, we modeled our medical testing reminder on radio adverts, without revealing sensitive information like the organization name, profession, or the STI types. Please see below).

“Namaskaram, I am Geetha speaking from Swathi Mahila Sangha. We are organizing healthcare camps. They are taking place at the following locations: every Sunday at the Leprosy hospital, every Wednesday at Swati Manne, every Thursday at the Kamkshi Pallaya government hospital, every Friday at the Ayurvedic hospital, and every Saturday at the Kamala Nagar BBMP hospital. We will conduct blood tests and health examinations. Please attend without fail. Thank you.”

Results

Our deployments over multiple verticals is unusual for ICT4D, but we were keen on learning about the nature of the phone broadcasts across healthcare, microfinance and training. We connected with 85.25% of members who provided phone numbers. 80.7% members listened entirely (we called 35, 23, 230, and 627 mobile numbers for the 1st, 2nd, 3rd and 4th deployment). Our users strongly indicated that the field coordinator's voice was received positively, by amplifying the existing familiarity and trust built over several years by Pragati officers. Contrary to our initial preference for using a film star's voice to read out the content, given its sensitive nature, we found a strong proclivity among our users towards using the field coordinator's voice, especially, a female voice. Users had control over their phones, and we did not hear of any instances of intruding into their lives at the wrong time. Finally, we noticed that the system managed to reach more people than we had actually called, because of secondary diffusion among the call recipients to non-recipient friends.

There is a lengthier discussion about what our design and deployment can tell us about M4D and HCI4D in general, and more details on the USW lifestyles, method, design, design, deployment, and results in our CHI (Conference on Human Factors in Computing Systems) 2011 paper (PDF, coming up for presentation in May).

Note: This work was done while Nithia was interning with the Technology for Emerging Markets group at Microsoft Research India in collaboration with Ed Cutrell and Julie Weber (from the University of Michigan).

Photo courtesy Nithya Sambasivan.