Tigo BIMA Adoption Study

with the University of Edinburgh and Tigo Tanzania

For my Masters dissertation, I investigated the correlation between sense of agency and low adoption of an affordable mobile health insurance service by Tigo Tanzania (telecom company). I had already conducted a usability study for an assignment in my graduate course, Human-Computer Interaction, and the score was high. If usability wasn't the problem, then what factors were involved in the adoption of Tigo BIMA?

Background

Tigo-BIMA is a m-health intervention that offers short-term low-cost life and health insurance services, founded by BIMA, a provider of mobile-delivered insurance, in partnership with the mobile operator, Tigo Tanzania. Tigo BIMA has the potential to improve millions of lives by providing access to previously inaccessible medical care. The service was designed for low-income population who cannot afford to pay for yearly insurance and it uses the widely-accessible USSD code technology for feature phone users.

Unfortunately, technological capacity does not guarantee user acceptance. After launching, Tigo BIMA was facing low adoption from one user group specifically - women. Since I was working at Tigo Tanzania during the product launch, I was interested in investigating the low adoption challenge further for my dissertation. For an assignment in a Human-Computer Interaction course, I did a heuristic evaluation of the user interface with 5 users and the usability score was high. I used the findings as the basis for my dissertation research - "HCI for Development: Does Sense of Agency Affect the Adoption of a Mobile Health Insurance Service in Tanzania?"

Research Hypotheses

Higher beliefs in oneself as health locus of control (IHLC) is positively associated with Tigo-BIMA subscription
Higher beliefs in powerful others as health locus of control (PHLC) is negatively associated with Tigo-BIMA subscription
Higher beliefs in chance as health locus of control (CHLC) is negatively associated with Tigo-BIMA subscription
Higher beliefs in God as health locus of control (GLHC) is negatively associated with Tigo-BIMA subscription
Higher mastery (PM) is positively associated with Tigo-BIMA subscription
Perceived usability (SUS) is associated with Tigo-BIMA subscription (control)

Method

Step 1 - Secondary Research

I referred to 42 articles and academic research studies to learn about healthcare in Tanzania, alternative medicine, m-health innovation and impact, cognitive factors in technology, emotional factors in technology, designing for development, sense of agency and technology barriers for women in Africa.

Step 2 - Research Design

Informed by insights from the secondary research, I decided to design a survey combining different scales to measure sense of agency and a set of interview questions for contextual inquiry. All research materials were translated from English to Swahili. The following scales were used in the survey:

Multidimensional Health Locus of Control Scales (MHLC; Wallston et al., 1978) - measures belief about a person's sense of control over their health across 3 themes: Internal control (IHLC), leaving control to Powerful Others (PHLC) and leaving control to Chance (CHLC). The scale consists of 18 items rated on a 6-point Likert scale (1 = strongly disagree, 6 = strongly agree).

The God Locus of Health Control Subscale (GLHC; Wallston et al., 1999) - assesses the belief in God as a locus of control for one’s health. In light of Tanzanians’ over-determinism in spirituality (informed by secondary research), the GLHC scale was a useful tool to isolate the relevance of God in personal health.

The Pearlin Mastery Scale (PM; Pearlin & Schooler, 1978) - measures overall sense of agency and self-efficacy

System Usability Scale (SUS; Brooke, 1996) - to measure usability of the Tigo BIMA system

Step 3 - Participant Planning

I partnered with Tigo Tanzania to screen users so that I could have a list of female participants who fell into the following categories:

Tigo BIMA Subscribers
Tigo BIMA Non-Subscribers who were Aware of the service
Tigo BIMA Non-Subscribers who were Unaware of the service

*(Secondary research by Duggins (2011) argued that knowing about an opportunity could moderate the effect of sense of agency, such that those who are aware are more likely to take action simply because they know about the service. Therefore, I included the unaware group to control for knowledge of opportunity).

Step 4 - Research Procedure

I trained 5 agents from Tigo Tanzania's call center on the survey script, obtaining consent and mitigating biases. I also monitored the duration of the study to maintain ethical and empirical standards
221 telephonic surveys were completed by the agents and participant responses were entered into Qualtrics
I interviewed 25 participants (9 Subscribers, 8 Aware Non-Subscribers, 8 Unaware Non-Subscribers)
Interview participants were randomly chosen from 172 participants who completed the survey and expressed interest in a follow-up interview
The interviews were conducted face to face in a meeting room at Tigo offices
Interviews were recorded on using Voice Memos on my iPhone
Interview compensation was TZS 20,000 ($ 8.6)

Step 5 - Data Analysis

Quantitative Analysis using R: Descriptive statistics, Spearman's correlations, Generalized linear regression

Qualitative Analysis: Thematic analysis

Critical Insights

System usability does not influence subscription

SUS scores were high for Tigo BIMA and there was no correlation with subscription. Pricing was also not correlated.

Giving up agency for personal health to God had the highest correlation with low subscription

Interview responses supported the linear regression analysis that the more participants believed that God was responsible for whether or not they get sick and there is nothing they can do to stop it, the less likely they were to get health insurance.

Lack of trust in the Tanzanian healthcare system correlated with low subscription

Participants had bad experiences which made them lose trust in the healthcare system, which led them to lose trust in health insurance by association.

Research Impact

Changing the product name to Bima Mkononi (Insurance in your Hands)

The name was changed to Bima Mkononi to increase user's perceived sense of control of not only their own health but also the service itself and the healthcare system.

Redesigning the service

The service was redesigned to increase transparency by sending messages whenever customers use their cover. Bima Mkononi also introduced disability cover and focused the health insurance specifically on hospitalisation, which were both perceived as necessary regardless of belief.

Targeted Campaign increased adoption

Insights from this research helped the marketing team to design an effective campaign which made clear exactly what the insurance was covering, focusing on accidents and bone fractures, which are common and cannot be treated with alternative medicine. The campaign also included success stories from the target population.

Adding knowledge to the field of Human-Computer Interaction

This research highlighted the importance of investing in understanding underserved populations and the contribution of cognitive and emotional factors in design and technology.

Reflection

Conducting interviews in the user's natural environment

Doing the interviews in a meeting room at Tigo could have affected the results because participants might have responded with what they thought I wanted to hear due to the intimidating professional surroundings. In the future, it would be better interview participants in their natural settings in order to capture their unbiased thought processes and how they interact with their environment, especially if the research focuses on unique cultural contexts.

Read Full Dissertation