Now he’s running six large-scale trials across six countries alongside his teaching at Stanford. Why does he go to such lengths to get the data?
When Stephen J Anderson stepped off the plane in Cape Town with his wife Sophia and eight-week-old baby in tow to start his PhD research into how small business owners can benefit from marketing training, he didn’t expect that they would live there for three months. “It was a lot harder than I imagined,” he says. “I didn’t know what I was getting into.”
“Most academics are quite hands-off,” says Janine Titley, one of Anderson’s research managers in South Africa. “Stephen really cares about the detail. His commitment and the quality of his research are awe-inspiring. I have enormous respect for him. I also think he’s a little bit crazy.”
There are definitely easier ways to gain a doctorate, which don’t involve travelling further than your desk. “The usual approach to research is to take an existing data set and strangle it till it talks,” says Jeff Skinner, Executive Director at the Deloitte Institute of Innovation and Entrepreneurship (DIIE). “Stephen does something very different.”
Anderson adds: “In marketing ‘field experiments’ have typically involved taking one company’s list of customer emails or database of online users and randomly assigning different ads to different individuals to see which are most effective. My approach is to work with hundreds of small companies to evaluate if and how marketing solutions stimulate business growth in emerging markets.
“A key challenge, however, is the lack of records in these contexts. So we literally go out into the field to construct our own sample from scratch, then collect performance data over several years using one-on-one interviews and electronic surveying methods.”
Rajesh Chandy, Professor of Marketing at London Business School (LBS), was Anderson’s PhD advisor. “Stephen’s work is an exemplar of “action research” in business: it involves active engagement by the researcher in the design and implementation of an intervention involving real businesses in the field. It’s not easy to pull off – especially in far-flung, unfamiliar settings.
“A researcher needs to have at least three qualities to succeed: a knack for picking important problems, the ability to frame novel research questions and unusually strong skills in project execution. People who combine these qualities are rare. Stephen has all three in abundance.”
Anderson recalls: “I remember slamming open Rajesh’s door, saying: ‘I’m gonna focus on business skills for entrepreneurs! There’s only a few working papers on it and no-one’s looking at it solely from a marketing angle.’ I didn’t realise my naivety. There were actually many people already doing work in this space who were further ahead of me with funding secured and projects midway. It’s been quite the learning curve but I keep trying to improve and come up with novel ideas for new projects.”
The Deloitte Institute of Innovation and Entrepreneurship provided vital seed funding that enabled Anderson to make initial trips to South Africa to discover and learn with low risk. “London Business School was really generous. There were small pockets of funds you could get as a PhD student to kick things off. I took calculated risks, started new projects and tried to work with new people.”
Anderson now has six randomised controlled trials running in parallel across several different African countries and India, and has set up his own cloud-based data platform called ENTREinfo to carry out the research. He is also teaching at Stanford Graduate School of Business.
Like any junior researcher on tenure track, Anderson has professional reasons for wanting his projects to stand out. He’s also driven by very personal reasons. “I was born in the projects in Vancouver. My biological father left my mom alone with five kids when I was very young and I remember being poor. I won the birth lottery by being born in Canada where even the poorest kids can get access to food, a good education and healthcare. But at seven or eight something clicked and I thought, ‘I’m getting out of here’.”
He worked his way up through school. “I’m not naturally smart. I would wake up super early to be up before my mom to study or go to the gym to train for basketball. I’ve never been gifted academically or athletically (I’m short for a basketball player), I just worked hard. I’d go to the library after school and then I’d help the janitors clean the gym so they’d let me continue to practise at night and then I’d study again.”
Anderson’s discipline and persistence paid off through athletic and academic scholarships to the University of Victoria: he became the first in his family to gain a degree. When it came to choosing his PhD at LBS he thought, ‘How can I have an impact? How can I differentiate myself and my research?’
Just after he arrived in London, Anderson’s beloved mother died. “I was kind of lost,” he says. “I was aware I’d had all these opportunities, partly thanks to the mentors and professors who’d given me a chance. Maybe I could do more than just care about myself and make money. That’s what pushed me to go and have this adventure.”
In collaboration with Professor Chandy and Bilal Zia of the World Bank, Anderson set out to design a large-scale randomised controlled trial comparing a group of small business owners who received marketing training with a group who received finance training and a group who received no training. With no initial data – no sales records, no accounts – and without the “before” it would have been impossible to accurately calculate the “after”, ie what effect each intervention had on business performance over time. Anderson considers this a blessing in disguise, in retrospect.
“The world is a messy, noisy place. If I’d only had archival data, then my analysis might have been limited to examining correlations instead of causal effects. For instance, I wouldn’t have been able to tell if sales performance had gone up because a company’s leaders had greater marketing skills or the improvement was due to external factors. Running a field experiment allows us to overcome these challenges and better isolate the effect of interest, in this case the impact of marketing skills on firm performance. But to do so required building our sample from scratch and collecting primary data over multiple years.”
Titley describes the team’s rapidly evolving method on the ground: “Plan A was to sign up small business owners through five branches of a local support network but there wasn’t enough footfall. That’s when Steve came up with his ‘pound the pavement’ strategy. We mapped out all of the business areas and markets in greater Cape Town then set off, a team of 15, and told people, ‘Here’s this project, are you interested? Sign up’.
“We limited it to small business owners working out of a physical structure in the hopes that they’d be easier to track down for future surveys and better able to apply the skills they learned to improve their businesses. We made many mistakes along the way, but continued to iterate our approaches and eventually pulled it off.”
She recalls “No-Sleep September”, when she and fellow project manager Christy Lazicky spent “nights planning, days implementing” to get the sample recruited, registration completed and the baseline survey conducted. The logistics are dizzying: the team approached about 10,000 small business owners and interviewed nearly 2,000 of them to end up with a study sample of 850. The next challenge was keeping the participants coming to class for all 10 sessions. “At first we offered biscuits, then they petitioned for muffins. The snacks were the bane of my life.”
“There are tough logistical challenges that make this kind of research hard,” says Anderson. “There’s the back-end admin around partners, institutional review board ethics approvals, research agreements, memorandums of understanding and more just to get a new project off the ground – a whole bunch of headaches. And most critically, you’ve usually got to get the local stakeholders to agree to scale up to numbers they’ve never done before, in order to have the statistical power to detect effects.
“Then there’s raising the money. Initially you think a couple of smaller doctoral grants for US$10,000 or $20,000 will suffice but no: big projects with the potential for big impact typically involve samples of 1000+ small businesses and cost hundreds of thousands of dollars. It’s a massive hurdle.” The project secured further funding from the STARS Foundation, the UK government, and the World Bank.
Anderson maintains that, if the research is properly designed, the final analysis should be relatively straightforward – and genuinely useful. He also believes in getting his hands dirty. “I don’t just evaluate. I collaborate closely with partners to improve an existing programme or shape it by adding new offerings – so I can study these interventions to address novel research questions or issues important to managers and policymakers. I ask myself, ‘What can we do to make this a stronger intervention?’”
The road less travelled has its own rewards. “I’m very close with the partner. Sometimes we are the partner. In Ghana, I spent weeks in the field seeing how small business owners waste or divert money from loans – then I designed a new investment product and pitched it to different banks until someone agreed to launch it with thousands of clients. Now here I am getting to see if it works or doesn’t. All of these business solutions have a marketing angle so I get to study problems that are relevant to my field and will hopefully make an academic contribution.
“Intellectually it’s a creative challenge but I also have to believe it will make an impact. If it’s not interesting academically or doesn’t matter to managers and policymakers, why do it?"