Digital Research in consumer finance

From paying with cheques to trading stocks, digital technology has transformed how consumers and professionals do their banking. As David L Stearns describes in his book Electronic Value Transfer: Origins of the VISA Electronic Payment System (2011, Springer), the history of payment systems can be traced back as far as the 19th century, when wire services made it possible to send money quickly over long distances. Wire services provided the foundation for the interbank associations that became the Visa and MasterCard networks later on.

In the early 1960s, the banking industry was one of the largest users of paper, struggling under the weight of deposit slips and cheques. Coding cheques magnetically was an early move to use technology to reduce both paper usage and processing time.

Over the next few decades, as the cost of computing technology waned, these early technological developments were followed by the automation of other processes using On-Line Real Time (OLRT) computing. These included the use computers at the point of sale to process transactions, and in the back end to keep records.

However, digital finance did not take off in a large way until the World Wide Web became accessible in the early 1990s. Instead, ATMs were joined by an increasing array of credit, debit, and store cards. Charge cards have existed since the late 1800s, but they did not become widespread until after EFTPOS (Electronic Funds Transfer at Point of Sale) was launched in 1981.

Today, this field is huge and growing. It encompasses everything from how people use gift cards to how they purchase items in online games. It is made even more huge by the vast quantities of data now available about people's transactional histories, although many of those data sets are proprietary and often off-limits to researchers.

In order to gain access to such data, researchers must approach the institution that owns the data they want and request a Data Use Agreement (DUA). A DUA is a legal document that allows researchers to access, analyse, and publish data under certain conditions, usually focused on protecting the confidentiality of participants. Whether the owner is legally permitted to share their data with another party depends upon the legislation that is in operation in both the institution's and recipients jurisdictions. Researchers wishing to access a particular data set should contact the institution that owns it and request a DUA.

As digital consumer finance products and data have proliferated, the range of research topics and the tools we use to investigate them have also expanded. Service providers, such as banks and payments companies, have historically been geared towards identifying customer needs and preferences in order to develop better products and better delivery mechanisms.

In terms of methods, service providers have been particularly instrumental in developing survey techniques and ways to test user preferences. While earlier commercial research tended to focus on product development, later studies have shifted focus to customer relationships, including investigating how digital devices can be harnessed to create greater intimacy with customers.

With respect to consumer welfare, psychologists and sociologists have investigated the impact of digital money on people's spending habits, especially focusing on whether it increases indebtedness. Other studies find new ways to gather data, such as Joshua Blumenstock's work analysing mobile data in Rwanda and Afghanistan to discern wealth distribution and migration patterns.

Non-profit organisations have tended to be more focused on how digital consumer finance can increase the "financial inclusion" of the world's poorest people, including Information and Computing Technology for Development (ICT4D) and the use of digital devices for microfinance and money transfers.

Field experiments have been used extensively as a way of testing program effectiveness [see Experiments]. However, commentators have pointed out that unequal access to technologies (the "digital divide") means that some groups are unable to benefit from digital services, and may even face greater risks using digital services than transacting in cash.

Social scientists working in universities and government bodies have applied a broad range of qualitative and quantitative methods to digital finance research. These have included ethnographic observations of credit card use by anthropologists, interviews and surveys by sociologists, and lab experiments by economists and psychologists. They cover a broad range of topics, including financial literacy, digital service uptake, privacy issues, security risks, how digital services affect consumer choices, how digital interaction shapes financial decisions, the uptake of digital currencies, use of digital services for illicit activities, and so on.

What is Digital Research

  • Qualitative / quantitative
  • Multiple methods possible
  • Face-to-face or remote data collection

Digital research is not a method in its own right since technically all of the studies in this toolkit can potentially have "digital" aspects. Interviews, surveys, focus groups, ethnography, experiments, financial diaries, and user data analysis are just some of the methods that can be adapted to in digital research.

However, digital research but it deserves specific attention because it changes the ways in which we can carry out these classic methods, by providing new tools and avenues for communication with research participants and data. Moreover, consumers are using an increasingly wide array of digital payments services, and these are an important topic of study. These include:

  • Using ATMs
  • Multichannel banking
  • Shopping online for insurance
  • Making mobile phone payments
  • Use of digital currencies
  • Playing the stock market online
  • Creating a household budget on a computer spreadsheet

Digital Research can take place online, face-to-face, or both:

  • Studies carried out online may include interacting with participants in games, conducting participant observation in forums, or examining patterns of usage across social media
  • Studies carried out in person may include moderating a focus group about mobile money use, conducting participant observation of people using online banking, or interviewing a person about their use of a financial literacy application on a mobile phone
  • Some studies incorporate both modes, incorporating data collected online and offline, including users' interactions with researchers online and in person, and data collected by service providers

Digital Research therefore often blurs the boundary between "online" and "offline" worlds. Indeed, as the anthropologist Tom Boellstorff comments,

One thing that this kind of research demonstrates is that online interaction can be "virtually" face-to-face, and digital technologies are changing what it means to be "remote" in the first place. (personal communication)

An exciting outcome of the spread of digital consumer finance is that it has expanded the range of methods, tools, and techniques available to researchers. The fact that most people use computing technology (especially mobile phones) means that researchers can shift away from classic ways of collecting data, instead using online surveys, mobile apps, chat programs, video interviews, blogging programs, and other tools.

Strengths

Reduces geographic constraints

Internet-based studies can enable participation by people who would not usually be able to take part due to geographical distance from the research site. Recruiting and data collection can take place through a variety of platforms, including social media, electronic mailing lists, third-party websites, games, and video calls. This reduction of geographic constraints can assist with increasing the representativeness of a sample and can facilitate studies that compare geographically distant groups (say, in different countries or regions).

Expand the range of tools available for research

Digital researchers are innovating new ways to collect and analyse qualitative and quantitative data. There is now a wide array of web-based tools for data collection and analysis, including software that were deliberately developed to collect data (such as Revelation, a program that participants use to share information with researchers), platforms that were designed for other purposes but that researchers use for data collection (such as social media sites, discussion groups, or third party data sets), or analytical software that can be downloaded or used online.

Facilitate follow-up studies

Locating participants after a research phase is completed can be difficult. People may change their address or phone number, and locating people in person is expensive and time-consuming. Social media and email make it easier to contact participants for follow-up research or to share a study's results.

Limitations

Limited access to some groups

Digital research methods can only increase a sample's representativeness if the target population are online or own a mobile device. While mobile phone ownership is increasing rapidly, rates of Internet connectivity are far from uniform. This is especially true in developing countries, and sometimes also true of "wealthy" economies. Research design, therefore, needs to consider what the best method is to reach the target population.

Remote data collection can reduce data quality

Relying on remote data collection can reduce the quality of the data that researchers collect. This is because participants tend to share more information with people they trust, and it can be difficult to develop this trust online. This can present problems for both quantitative research (e.g., convincing people to take a survey or complete it accurately) and quantitative (e.g., conducting an interview with a stranger). These problems are not unique to digital research and have been faced by telemarketers for decades. Ultimately, whether or not remoteness presents a problem depends on the type and depth of information that the research aims to collect.

Case Study 1 — Studying Bitcoin using qualitative and quantitative methods

The evolution of digital payments has generated some interesting forms of financial transactions. Digital currencies, such as Bitcoin and Dogecoin, are perhaps the most controversial of these, since they are not created by government entities or banks, provide a means to circumvent currency control, and can be used (within limits) to hide transactions from the law.

Bitcoin has unique characteristics that make it markedly different from other payment systems in how it operates and how it can be used. Unlike with bank cards transactions, Bitcoin payments are anonymous; no user information is recorded in the transaction. Bitcoin transactions are not cleared through banks or any "central" location. Instead, they are cleared by a network of participants who compete with one another for a reward for authorising them.

Bitcoin is a matter of interest to policy-makers, central banks, and researchers who are interested in the technical aspects of this new form of money creation and transaction. It is also of interest to exchanges and investment banks interested in the possibilities of its distributed nature for increasing settlement speed. However, Bitcoin's novelty and technical complexity means that researching it requires both creativity and expert knowledge of computational mathematics.

A team of researchers from the University of California, San Diego and George Mason University have pioneered new ways of understanding Bitcoin and its users through combining qualitative and quantitative research techniques. The researchers were interested in identifying the extent to which Bitcoin lives up to its promise of pseudo-anonymity and to investigate how people's use of Bitcoin has changed over time. (Bitcoin is pseudo-anonymous because users have public addresses that anyone can see, but it is very hard to deduce who or what controls a public address.)

The researchers state explicitly that they did not aim to identify individual users. Rather, they used a combination of participation (using Bitcoin themselves) and algorithmic analysis of transactions to cluster users and the transactions between them. They identified service providers, but not users themselves. They used this information to identify the factors that compromised their pseudo-anonymity.

Methodology

In contrast to most payment systems, Bitcoin's users are pseudo-anonymous, but flows of value around the network are publicly visible. When a user makes a transaction, they use a public key that encodes their identity so that it is not passed on to anyone else in the chain. However, the Bitcoin blockchain encodes all transactions, past and present, and it is this feature that makes value flows publicly visible.

In this study, the researchers exploited the visibility of Bitcoin flows to cluster and positively identify Bitcoin service providers through applying algorithmic analysis and participating in transactions themselves. They used public keys to make 344 purchases from a range of sellers on Bitcoin, including mining pools, wallet services, bank exchanges, non-bank exchanges, vendors, and gambling sites.

These exchanges allowed them to cluster users, which in turn enabled them to identify major institutions in the Bitcoin marketplace and the interactions that occurred between them. The researchers were only able to identify those users that they interacted with, which were almost entirely third-party services like exchanges, not individual people.

The researchers began by carrying out a "re-identification attack" in which the researchers opened accounts and make purchases from a variety of Bitcoin merchants and service providers whose identities are already public (such as Mt. Gox and Silk Road). Since the researchers knew which public key they used themselves, they were able to positively label the public key on the other end as belonging to a particular service provider.

The researchers then turned to Bitcoin forums to locate cases in which vendors had identified their own particular key. They explain that many users list their addresses (or "tags") publicly. For example, charities list their donation addresses, and a company called LulzSec publishes their address on their Twitter account.

The researchers did not attempt to collect all addresses available but did amass a collection of 5,000 in total. They also searched Bitcoin forums (such as bitcointalk.org) to look for Bitcoin addresses of defunct organisations or ones that are associated with major thefts. According to lead author Sarah Meiklejohn,

The thefts show how criminal actors are engaging with Bitcoin; i.e., are they using it in a naive way, in which our attacks could be easily applied, or are they doing something more sophisticated like using mix services? Are they cashing directly out of the system using exchanges, or are they keeping the stolen funds in bitcoins? Basically, thieves were the most motivated users we could think of in terms of wanting to maintain anonymity, so it seemed natural to study their behaviour for this problem. (personal communication)

The researchers warn that these self-identified tags are not as reliable as the ones they collected themselves through making transactions, so they "consequently labelled users only for addresses for which we could gain some confidence through manual due diligence.

After this collection phase, the researchers analysed the data using account clustering heuristics. This enabled the researchers to identify 1.9 million public keys belonging to service providers or identities. They examined interactions with known Bitcoin service providers, and were able to identify 500,000 addresses as controlled by Mt. Gox, and more than 250,000 addresses as controlled by Silk Road. This did not allow them to identify the individuals making transactions per se, but it did allow them to observe interactions with particular services, such as deposits and withdrawals. In other words, the flow of Bitcoins in and out of the service was de-anonymised.

Findings

The main finding of the research was that, despite widespread belief that bitcoin is pseudo-anonymous, Bitcoin users can in fact be identified. The authors state:

Even our relatively small experiment demonstrates that this approach can shed considerable light on the structure of the Bitcoin economy, how it is used, and those organizations who are party to it. (Meiklejohn et al. 2013, p.12)

While they did not identify real-world accounts directly, their analysis de-anonymised users to a significant degree. In particular, the researchers analysed certain highly-publicised thefts to see if they could track the bitcoins to known services. In most cases they found that this was quite straight-forward. This has major implication for law enforcement:

demonstrating that an agency with subpoena power would be well placed to identify who is paying money to whom. (Meiklejohn et al. 2013, p.2)

This is largely because a small number of Bitcoin institutions (mostly services performing currency exchange) are becoming dominant, but it is also due to the public nature of Bitcoin transactions and the ability to label monetary flows to major institutions. Pseudo-anonymity therefore:

ultimately makes Bitcoin unattractive today for high-volume illicit use such as money laundering. (Meiklejohn et al. 2013, p.2)

The researchers suggest that a follow-up quantitative study could help to identify the scale of the issue.

Applications

Studies of Bitcoin and other digital currencies have clear applications for law enforcement, policy development, and understanding changes taking place in online trade.

With respect to law enforcement, this study suggests that it would be easier to confirm identity and, therefore, prosecute illegal activity carried out using Bitcoins than people tend to believe. Users may wish to think twice about whether Bitcoin really does protect their identity, and law enforcers may develop new approaches based on the findings of the study.

Policy development can also benefit from the study's innovations and insights. Research such as this advances our understanding of how Bitcoin works and provides us with new methods with which to study it. These kinds of studies could prove crucial to shaping public and monetary policy to take digital currency use into account.

In some countries, there is currently a tentative move towards incorporating Bitcoin into mainstream payments services, such as through contracting vendors to accept Bitcoin payments or installing Bitcoin ATMs. However, governments are legislating against Bitcoin use as much as they are legislating in favour of it. Understanding Bitcoin's potentials and pitfalls will help legislators decide its public value.

This study also increases our understanding of consumption patterns and of factors that lead to consolidation in payment service provision. One of the researchers' findings was that fewer, but larger, sellers are coming to dominate the Bitcoin market. It would appear that existing consumers share information with potential consumers regarding how to use Bitcoin and which sellers to choose.

This has the potential to drive customers towards Bitcoin from other payments systems and marketplaces. The increasing monopolisation of the Bitcoin marketplace by particular companies has the additional effect of decreasing anonymity, since large sellers are more readily identifiable. In other words, the Bitcoin marketplace is changing, and this changing market changes Bitcoin itself.

Ethics

Digital research, whether carried out in person or remotely, presents a broad array of ethical challenges. Privacy protection was the major ethical issue to arise in this Bitcoin study. To protect users' privacy, the researchers designed the study so that they would identify known service providers, but not individuals. Sarah Meiklejohn comments:

Our thinking was that we used public data, and as you say we identify users only by their Bitcoin addresses, and (intentionally) didn't identify any users who don't have a public-facing element (e.g., individuals rather than services). (personal communication)

By limiting their analysis to known service providers, and focusing on flows of Bitcoins rather than individuals' transactions, the researchers largely avoided issues of individual privacy and consent.

In many cases, however, judgements as to when social benefit outweighs issues of consent are subjective and problematic. Even where consent is given, it is not always clear that individuals will understand what they are agreeing can be done with their data. How will it be analyzed? Will the data be adequately anonymized and stored? What will the findings be used for? Will the data or results be shared with third parties, and for what purpose? Will it be used to increase company profit or to deny rights to certain individuals?

People are right to be concerned: there are plenty of ways in which data can potentially be misused. In September 2015, a story broke about how Facebook secured a patent that would allow banks to make loans based upon the credit history of their entire social network. Theoretically, a bank could deny credit if an individual's friends had a bad credit history, even if the individual themselves was in good financial standing. While such fears may be unrealistic, they point to the fact that few of us really understand the implications of data sharing and its effects on our financial lives.

Health insurance is another area of concern, and it is a good example of how social benefits can clash with social risks. Big data has the potential to provide enormous social benefits in the area of health care. If health care providers such as the NHS in the UK are able to access large amounts of data on public health, then they will be far better equipped to plan services for the future.

However, this kind of data is intensely private, and needs to be well-protected. An example of misuse would be if insurance companies could use the datasets to positively identify individuals with chronic illnesses or who need expensive treatments and deny them coverage.

Researchers and professional associations are responding to these kinds of issues by developing ethical codes and guidelines for digital research. For example, the Data Science Association has produced the Data Science Code of Professional Conduct to help researchers think through ethical issues.

Various specialised books now exist on the subject, such as The Ethics of Big Data: Ethical Reasoning in Socio-Technical Informatics. However, the issues are complex and are likely to become more so in the future. While it is crucial that researchers planning digital studies are up-to-date on current ethical practices and standards, it is equally important that regulators intervene to protect consumer rights.

For more information on this study, see:

Case Study 2 — Combining online and offline data collection on payments in Indonesia

A study examining how online behaviours are affected by offline lives was carried out in 2012-2013 by the American anthropologist Tom Boellstorff and his team. They collaborated with two Indonesian research teams to learn how Indonesians were combining social media, mobile phone use, and payments systems given that use of devices and the Internet was increasing rapidly in Indonesia.

The researchers chose this focus because device ownership, internet access, and their use for online shopping have grown rapidly in Indonesia over the past two decades. While many Indonesians continue to be left out of the digital revolution, the gap is fading fast and those who are connected often have multiple devices.

Due to the low price of SIM cards, many Indonesians have multiple smartphones with different providers so that they can obtain the cheapest calls possible. By October 2012, Indonesia had over 64 million active Facebook users, making it one of the top 5 nations in the world.

Indonesians use Facebook to connect with friends, but also to buy and sell consumer items through a variety of online stores and mobile apps. Boellstorff and his co-authors write that the International Data Corporation (ICD) showed that the value of internet-based trade in Indonesia reached $3.4 billion in 2011, and that a MasterCard survey in 2012 indicated that online shopping had increased 15% in six months.

However, credit cards played a relatively small role in trade: a Nielsen Online report showed that 57.4% of respondents were using online transfer methods for payment, but only 11.5% were using credit cards, and 13.1% preferred cash on delivery.

Boellstorff and his teams combined face-to-face interviews with analysis of the online purchasing and payments environments that people were using to gain insights into how people made purchasing and payment decisions. They focused on both the technology that people used and the social relations that shaped people's actions.

Methodology

The research took place in Surabaya (Java) and Makassar (Sulawesi). The researchers used qualitative methods including participant observation, individual semi-structured interviews, and focus groups. They supplemented these with the analysis of websites, mobile apps, and advertisements.

In Makassar, the researchers interviewed 54 respondents and conducted two focus groups with ten participants in each group. In Surabaya they interviewed 52 respondents and conducted four focus groups. All data were collected in Indonesian or in local languages and then translated into English by members of the research teams.

In each location, the researchers sought to recruit a diversity of people from different social groups, including those that they thought would give a range of perspectives on mobile social media and mobile payments (such as university students and "housewives" ). The best-represented group were heterosexual women, who are active in the world of online shopping in Indonesia.

The research teams held an initial meeting in September 2012, before research began, to decide on key interview questions to be included in all of the studies. The questions covered multiple dimension of device use. For example, to identify time discrepancies between when people began to use devices and when they began actually shopping online, interviewees were asked the following questions:

  • When did you begin using gadgets?
  • When did you begin using the internet?
  • When did you begin making online transactions?

To investigate users' behaviours, including how people decided to become re-sellers and any problems with addiction to online shopping, the researchers asked:

  • What motivated you to make transactions online?

In response to other studies that argue that online shopping almost always happens in combination with other activities (such as socialising or working), they asked:

  • Are there particular times when you cannot shop online?
  • What are you doing when online shopping?

People's reasons for using online transaction services tend to differ according to context. For example, while shopping may be popular in one country or among a particular social group, other services such as remittances may be more widely used elsewhere. To find out what people were using digital financial services for, without biasing their answers, the interviewers asked:

  • What kinds of transactions do you make online (shopping, sending money, paying bills, etc.)?

To find out how people saw themselves using services in the future, they asked:

  • Have you ever thought about stopping shopping online?
  • Have you ever thought about reselling things that you purchase online?

Finally, in order to find out how people actually pay for goods, why they choose one payment mechanism over another, and whether they set aside special funds for online shopping, they asked:

  • What funds do you use for shopping online?
  • How do you or your friends pay for online shopping if you don't have money at hand?

For the online part of data collection, the researchers analyzed websites, mobile apps, and advertisements. They collected and assessed this online data by going directly to the websites in question. However, they note that it is possible to do interviews and focus groups online as well.

Data analysis was synthesized as part of the overall research process. Boellstorff commented to us,

The key thing to remember is that the phenomenon being studied is already "synthesizing" the online and offline before we ever got there. So synthesizing the online and offline in this case (and in practically every case of such digital research nowadays) is not an artificial imposition. Ideally it should reflect the specific ways that in the case at hand, the online and offline are shaping each other. (personal communication)

In other words, the "online" and "offline" research should not be treated as separate data sets that are collected independently of one another and then brought together for analysis. Rather, the online and offline data are inextricably linked. For example, interviewees might tell stories in which events take place in the home and simultaneously through social media.

Similarly, an observation of a person using an ATM must take into account the "real" world, because factors such as time constraints and safety considerations will affect how they use that digital device. Moreover, people use multiple devices to achieve particular goals. Treating each digital interaction as distinct does not reflect how people use devices in the course of their everyday lives.

Findings

The research was designed to study the intersection of mobile phones, social media, and payments. As a result, some of their findings addresses digital consumer finance directly, while other findings address it indirectly through describing the context in which transactions take place.

Boellstorff and his team found that online shopping in Indonesia is made possible by the prevalence of devices. All of the interviewees owned more than one device, often a laptop, a BlackBerry, and at least one other smartphone.

Due to the low cost of SIM cards and the advantages of using multiple providers, many respondents had multiple smartphones or SIMs; for example, one to keep in touch with a romantic partner and one for other friends, or one for personal use and one for business use.

At the time, BlackBerry was still the most commonly-used handset, and all of their respondents had a BlackBerry and often other kinds of smartphones as well. One interviewee had 5 smartphones, each with a different provider.

The reasons why people began to shop online rather than in physical retail stores varied. The researchers found that respondents became interested in online shopping after seeing items their friends had purchased. The researchers explain:

This reflects a broader pattern in which friends and acquaintances play an influential role in online shopping practices not just as recommenders, but increasingly as customers and sellers. (Boellstorff 2013, p.12)

Buyers often knew sellers personally or at least lived in the same city, meaning that they could choose sellers based on personal knowledge or recommendations. This also made it easier to complain if there was a problem. Buyers often did not have to pay for shipping because items would be hand-delivered. Their buying practices therefore often mimicked physical shopping.

Facebook was often the first pathway to online shopping because respondents would see sponsored advertisements, information posted by friends on their own timelines, or comments customers had posted on the Facebook pages of sellers. Alternatively, people would be introduced to online shopping through the BlackBerry store app.

Once interviewees had begun to shop online they identified numerous advantages. The five primary reasons that respondents gave for wanting to shop online were 1) it is easy; 2) interesting things are sold in online stores; 3) it avoids the hassle of going to a physical store; 4) it is often cheaper; and 5) some items are hard to find in physical stores.

We often think of online shopping as something done by individuals or households, but the researchers found that groups of friends would also make collective purchases. For example, they discovered that groups of university students would purchase food (such as snacks) in bulk to receive lower prices.

The amount of money that interviewees spent online monthly varied from less than $1 to around $50. This sum reflects people's disposable incomes, but also their feelings about online shopping. Concerns about losing money tended to limit how much money people were willing to spend online.

As well as concerns about fraud, people worried that the items they bought would not match the online photographs or description and that they would be disappointed with their purchases. To counter this, one woman set a 500,000 rupiah ($50.50) limit on purchases. Beyond this limit, she felt uncomfortable and preferred to make the purchase in a physical store.

Some people also earmarked funds to be spent on online shopping. One interviewee, a man called Eska, gave his monthly salary to his wife to manage (a common practice). He had his own, separate bank account for his own expenditures ("men's money") that was funded primarily by workplace bonuses rather than his salary.

When he shopped online he always used this account, even when his wife asked him to buy something for her. In practice, then, it wasn't just his "men's money," it was also the household's online shopping account.

In terms of making payments, the researchers discovered that only 6 respondents used Internet banking services to make a payment, and only 5 respondents used a credit card. Instead, the vast majority of their respondents were paying for online shopping by making a transfer at an ATM or a bank counter. The main banks they used were BCA and Mandari; only 5 respondents used BNI.

Why use an ATM rather than making an online transfer? Some interviewees said that they were worried about credit card fraud, but the main reason cited was to avoid bank fees. If the buyer and seller used the same bank, the transfer could be made for free.

In fact, some sellers had accounts at multiple banks to ensure that their buyers could transfer them money with no extra charge. Some buyers reported that if they didn't have the same bank account as the seller, they would ask a friend or relative to complete the transaction for them. Note the pattern: we see a cost-savings behaviour taking place at both the ATM and the SIM card level, since, as noted, many Indonesians have multiple SIMs in order to save money by switching SIM cards depending on whom they are calling.

People sometimes used other people's credit cards, but that could create secondary problems. For example, one gay man reported that he preferred shopping for makeup online because he felt safer than when visiting a physical store. In order to pay for his purchases, he used his mother's credit card. However, he was concerned about his privacy here, because his mother could then see on her statements what he had been buying.

Online BlackBerry shops were sometimes used to both make and complete transactions. Interviewees stated that they preferred these shops because sellers would be identifiable by their BlackBerry PIN. They also tended to be more familiar with the sellers and had friends in the system who could provide recommendations. Moreover, BlackBerry Money allowed peer-to-peer cash transfers, which interviewees perceived as safer than using a credit card.

Applications

The team's combination of online and offline data collection has many potential applications in product design and marketing. For example, there could be opportunities to develop consumer finance products that reduce the costs of making transactions.

Consumer decisions are influenced by various factors: they may identify with the brands of their devices, their mobile phone carriers, or be swayed by the preferences of their friends and family. These identifications influence consumer decisions in different ways, and at different times. The combination of in-person ethnography and examination of digital sources used in this study is useful in finding out how people use products within real-life contexts.

Many of the study's participants reported having to put in a significant level of effort to complete payments, often maintaining multiple bank accounts to lower costs. The BlackBerry store presents a way to overcome some of these costs by smoothing out the transaction process.

However, as other brands are displacing BlackBerry handsets, customers will require other channels to make payments. Knowing where the pain points lie for customers, and which device to use when making a transaction, can assist in the identification of appropriate payment channels. We also see cost-saving behaviours that cross over between devices and services—in this case, ATMs and mobile phone SIMs.

Risk is another important area for product design and marketing. This study shows that consumers assume they are taking a certain level of risk when shopping online. They attempt to offset risk by using known providers and channels for their purchases, or through seeking advice from people they trust.

Building social proof into product design, such as through peer-to-peer transfers or permitting product recommendations from friends, could be leveraged to increase trust and encourage the use of particular payment channels.

Ethics

Most of the ethical issues raised in this study are the same as in any other face-to-face study. These include:

  • Giving potential participants sufficient information about the study so that they can make an informed decision about whether they would like to take part
  • Ensuring, as much as possible, that participants are not placed in physical or emotional harm during the study
  • Anonymising data so that participants are not identifiable unless they have given specific permission

However, as with the Bitcoin study described above, the "digital" nature of the research raises extra points of consideration. Boellstorff reports that one big issue he has encountered is that some researchers think that because something is online it is not "real" and so you don't have to protect people's identities.

For example, data gained from a public forum may be technically traceable, but if it is reproduced without permission, researchers have a responsibility to generalize the data into findings that are not traceable.

Another important issue in digital research is that social media tends to make people more visible, and so it can make anonymization and consent more difficult. Say an interviewee is demonstrating how they make a purchase in a BlackBerry store. When they show the researcher their own information, they are also likely to expose information about the people they are transacting with. These third parties have not given consent to take part in the research, and extra effort must be taken to discuss and record this data in the most general terms.

Safety issues may also be a consideration when studying money use. Following participants as they go about their daily tasks can yield valuable insights into their use of consumer products, but a foreign researcher accompanying a participant to an ATM may attract unwanted attention. These are not necessarily problems that were raised in this particular study, but they need to be considered at the outset and built into project design.

For more information about this study, see:

More about Digital Research

Examples in consumer finance research