11
RADIO GAGA
EVOLVING FIELD EXPERIMENTS IN MALI
KRISTIN MICHELITCH
▶ FIELDWORK LOCATION: MALI, BETWEEN MOPTI AND TIMBUKTU
In 2010, my colleague and coauthor Jaimie Bleck and I were interested in the effect of initial exposure to the national state radio station, the Office of Radio and Television in Mali (ORTM), on people’s political engagement surrounding the upcoming March 2012 Malian elections. Although Mali had been heralded as one of the most stable and competitive democracies in sub-Saharan Africa, village chiefs were also known to largely “gatekeep” political information and otherwise influence opinions and vote choice of villagers. However, over the last couple of decades, radio towers emitting ORTM had been spreading from the capital city of Bamako out to rural villages. Inspired by the seminal work of Daniel Lerner, Lucian Pye, and Eugene Weber on mass media expansion, we wondered whether access to radio would increase autonomous and pluralistic political engagement among villagers.
Our approach was simple.
Step 1: For a natural experiment, locate areas with no radio emissions, as well as contiguous areas that had just received radio emissions but where people did not yet widely own radios.
Step 2: In a field experiment within the radio emission area, randomly assign some men and women to receive solar crank radios to boost exposure and others to receive a placebo item as yet undetermined (eventual answer: flashlight).
Step 3: Collect novel outcome data related to political identity, knowledge, opinions, and behaviors.
Of course, this approach did not go according to plan.
Our main article1 is frequently used by colleagues teaching field experimental methods as an example par excellence of experiencing major disruptions of the original research plan—a coup and a rebel insurgency. However, this is a tale of facing and adapting to many different fieldwork challenges, from Mother Nature, to putschists, to gender dynamics in a cross-cultural team.
A NATURAL EXPERIMENT BLOWN AWAY BY THE WIND
We had already jettisoned the natural experiment before arrival in Mali in December 2011. It was literally blown away by the wind. What seemed feasible sitting at our desks in the United States—to locate a sharp division among otherwise similar villages that were both inside and outside the boundary of ORTM emissions—proved impossible. There were no media maps. We discovered this by hiring research assistants to drive around villages between Mopti and Timbuktu on motorcycles with a radio, trying to tune in to ORTM. These reconnaissance missions revealed that due to “winds off the Sahara” some villages got ORTM all the time and others got it some of the time. We’d have to drive much farther north to find an area receiving no ORTM emissions, but doing so would mean traversing the north-south divide in Mali, planting us firmly in the territory of a different ethnic group—the Tuareg.
Not only would the causal identification in such a natural experiment be confounded by ethnicity (the northern Tuareg are seminomadic traders and herders oriented on the Sahara, whereas the southern Bambara and Peuhl engage in farming and herding in the Sahel), but we deemed the security situation too precarious. Citing neglect from the southcentric state, the Tuareg had repeatedly rebelled to form their own state. Directly prior to our fieldwork, armed Malian-born Tuareg mercenaries had been returning home from fighting in Libya (following Muammar Gaddafi’s fall), and grievances about the lack of public service provisions were bubbling up again to create a low-intensity secessionist movement (more on that later).
With the natural experiment literally blown away by the wind, we moved on to Step 2, our field experimental design.
HOW TO GIVE A RADIO (TO A WOMAN)
We identified a group of ten villages that received consistent ORTM but where people had not yet obtained radios. Because of potential for spillover within a village, we decided to randomly assign treatment at the village level. Recruitment at the village level began with visits to the village chiefs (dugutigis) to pay customary respects (in the form of kola nuts) and obtain consent for the research team to enter. Within each village, we would hold a public lottery and randomly select extended family households (dus). However, we were a bit stuck when it came to the next part. Would we then randomly invite one person in each du to participate in the research and receive a radio (or flashlight)?
It might seem straightforward to most people—just give people radios or just give people flashlights. However, we anticipated that it would be tricky to give a radio to a woman and interview her about political opinions in this rural context. Indeed, women are often excluded from research in similar contexts because of the difficulty and expense. But we were committed to including women, both because we had theorized heterogeneous treatment effects by gender and because we are committed to making sure we afford women the same opportunities as men, even if it requires substantial effort. To find a solution, we knew it was absolutely essential to conduct qualitative work—focus groups and interviews—in out-of-sample villages to hone the entire research design. We did so closer to Bamako, where ORTM had been broadcast longer and many people had already obtained radios.
Elderly male patriarchs (dutigis) of the dus or other male family members often dominated the household radio. We heard stories of some dutigis carrying the batteries for the radio in their pocket so other household members couldn’t “waste” the batteries. Furthermore, we were warned that dutigis may insist that opinions be collected from the dutigi, the public face of the family, who provides the opinions on behalf of the du. When dutigis consent to allowing a junior household member to give opinions, they may also hover and influence the opinions expressed.2
We asked the out-of-sample villagers, “How can we give radios so they won’t be commandeered by the dutigi or other male family members?” First, the fact that we had solar, crank radios alleviated the concern about battery hoarding—radio use was “unlimited” due to the renewable energy, which is very important in nonelectrified villages. Second, we would definitely need to give a radio to the village chiefs, which could be used together with the dutigis in the “old men’s lounge,” a communal space where the dugutigi and dutigis spend large portions of the day communing. Third, participants suggested that the best strategy to ensure women’s access was to give two radios (or flashlights) per du: one to a man and one to a woman. That way, there would be a device for men and a device for women, who often work separately due to gendered divisions of labor. If the dutigi disagreed with the random selection, we would simply replace the du with another du. The recipient of the device would have the “job” of maintaining the device on behalf of the family (in the case of the radio, to make sure it was charged by cranking it or leaving it in the sun). Finally, we would need to hire female enumerators to interview female respondents, and we would need to convince their families that the enumerators would be treated respectfully while traveling in a mixed-gender team.
This strategy worked amazingly well. Although it was challenging, we were able to recruit both female and male enumerators who could travel with our team. Being women ourselves undoubtedly aided in the confidence that the female enumerators would be properly and respectfully treated on the road. Dugutigis and dutigis were agreeable to the random selection. The public lottery for selecting dus (as well as the presence of white American women in the village) was somewhat of a spectacle for the dutigis. We welcomed this distraction, which quelled their desire to helicopter over our respondents and alleviated the need for enumerators to shoo them away. Finally, our follow-up research determined that the radios were by and large maintained by the randomly selected respondents.3
LADIES, PLEASE STOP COOKING
After a two-week training of a sprightly troop of novice enumerators, we finally hit the road to enumerate the baseline survey and deliver the radios and flashlights in the ten villages. We were camping in some of the most beautiful terrain in the world—red desert, clay villages with thatched roofs accessed by four-by-four, pirogue, and donkey cart. Without electrification this far into the desert, the stars in the sky were the brightest I’d ever seen. I had my tent, sleeping bag, headlamp, food bars, etc. My creature comfort is a hot shower, so I would take a bucket of hot water with a small scoop behind a village wall to “shower,” using darkness as my privacy.
The days were long and hard for everyone, but exciting. Our enumerators got along fantastically. The enumerators seemed to feel empowered by doing the job, often the first formal money-making opportunity they’d ever had.
We were a bit like a big happy family on a road trip. Each night we would, per Malian style, eat with our enumerators from a common bowl in “eating teams.” Each gender had a common bowl. The carbs—rice, couscous, or the like—would be spread throughout the bowl, and the good stuff (meat, vegetables, and sauce) was in the center. Eating was done with the right hand only, and this hand needed to stay in one’s own area of the bowl. The person with the most status within each gender distributed the good stuff into your sector of the bowl for you to eat.
Yet these dinners became the source of controversy. We needed to bring in food supplies to every village to feed the team, and we hired village women to cook Malian cuisine. However, we found that our female enumerators kept cooking the food for the research team while the male enumerators took their leisure by playing soccer or relaxing.
As good feminists, we were very much against our female enumerators working a “second shift” and doing additional unpaid work. We protested day after day for them to stop cooking. However, day after day the ladies kept cooking, each time with a slightly different excuse.
LADIES: The village women here don’t know how to cook properly, we are much better cooks and you must eat a tasty meal!
US: We will make sure to hire the best village women and we are fine if the meal is mediocre—you can’t be cooking after working, you need your rest!
LADIES: The village women do not wash their hands, we will all get diarrhea!
US: We will provide them soap and instruct them to use it prior to the cooking. Please stop cooking, you need your rest!
LADIES: We were done early with work so we decided to cook, it doesn’t bother us, don’t worry!
US: You can’t keep cooking like this. See, you all and the male enumerators are treated the same and paid the same and have the same expectations from us. We don’t want you to cook. Please stop cooking!
We slowly came to understand why the female enumerators kept cooking. We knew that cooking was traditionally a woman’s job in Mali. However, we did not realize the depth of power women held by cooking. Namely, men control scarce household financial resources, but women control scarce household food resources. Even though the men’s common bowl might appear to have the same or better morsels than the women’s bowl upon meal presentation, women could preemptively divert some good morsels to their mouths during the cooking process—a first-mover advantage known to any good cook. In this context, women did not want to give up their source of power (and extra food) to the village women.
WHAT? A COUP? AND AN INSURGENCY?
We left Mali, heading back to our respective universities, excited about the upcoming election and the idea that men and women with greater exposure to ORTM would have more political knowledge, more opinions, a larger plurality of opinions, greater national identity, and a host of other outcomes. We hired research assistants to keep a journal of what was said on the radio to hone our measurement of outcomes and sat back to watch the election events unfold.
One evening, I got a call from Jaimie. There had been a military coup in Mali. On March 21, 2012, low-ranking soldiers staged a coup, disgruntled by the inadequate arms and supplies they had to battle the northern insurgents. The first thing they did was take over ORTM and the state TV station, even before taking over the presidential palace. They additionally cited corruption of the classe politique, and overall poor performance in delivering public services to the people of Mali. A trio of insurgent groups—the secessionist National Movement for the Liberation of Azawad (MNLA), the Islamist hard-line group Ansar Dine, and the terrorist organization Al Qaeda of the Islamic Maghreb (AQIM)—capitalized on the political chaos and marched out to occupy the north and declare an independent state, with our study area ambiguously on the border of state and rebel controlled territory. The villages remained unoccupied by either side, but they were only twenty-five to fifty kilometers away from towns that the rebels and army held.
What to do? First and foremost, it was shocking and disturbing—people that we knew were facing additional hardship in a context in which life was already hard. Civil servants such as teachers fled south, tourism evaporated, travel to markets was stymied, and people were fearful.
Should we and could we continue the research? If we did, we would need to adapt the research to focus on the effect of putschist-controlled radio on attitudes toward the coup and the status of democracy as a whole. People had examined media effects under consolidated authoritarian regimes in the past. If we could collect some outcome data, it would be an important opportunity to understand media effects from a “long-term” media exposure with good causal identification directly in the wake of a transition. We decided, yes, we would put in any additional amount of effort necessary to try to collect some outcome data as long as it was safe and respectful for both enumerators and participants.
It was challenging and stressful to monitor the situation and ascertain whether and when it would be OK to recommence with research activities. After things had come to somewhat of a political standstill between the new junta and the insurgency, local people were able to safely move around the study area. We decided it was safe to collect outcome data in May and again in July. Jaimie held a training in Bamako, and her Malian-born husband, Drissa, accompanied the enumerators to the villages for enumeration. Knowing the villagers’ food availability had declined, we “paid” respondents for participating in the final surveys by bringing in a cow for slaughter to each village, and enumerators distributed the meat to respondents and their families.
Our discoveries turned out to be professionally rewarding, resulting in multiple publications and visibility among colleagues.4 However, the professional success of the project juxtaposed against the unfortunate circumstances for Malians often put me in an odd emotional space. Presenting the work at conferences and have audiences amped up about the “unique opportunity to conduct novel research” felt weird. Of course, both can be true—research can be interesting, and the object of study can be unfortunate. Nonetheless, I felt bad that some of our personal success was in any way related to unfortunate circumstances for our respondents and ordinary Malians as a whole. It is easy for others to view your data as data, but you know your data as real people facing real political events.
LESSONS FOR BUDDING FIELDWORKERS
These are just some of the many exstoriences from that fieldwork—an exstorience is a portmanteau joining the words “story” and “experience”: a situation that is easier to tell as a story than to actually live through. The first takeaway lesson is that you will attract funding with Plan A, then quickly move to Plan B, to Plan C, to Plan D, sometimes even before arrival, eventually winding up on Plan X. This happens for lots of reasons, from wind to a military takeover of the regime. You will need to switch gears and sometimes couch your research in entirely different scholarship.
The second lesson is that you need to vet everything qualitatively in out-of-sample areas, especially experimental interventions that represent an intent to introduce change in a society. We felt ethically fine giving people solar crank (and thus sustainable) radios as an intervention—radios already existed in Mali and were considered desirable to own. However, even something as simple as giving a radio or a flashlight required a lot of qualitative research. Do not make the mistake of believing that your intervention or sampling strategy is obvious and needs no vetting.
Third, you will find yourself in conundrums where your own norms are challenged in diverse contexts, and it will take time to get to the bottom of these differences. There is no one answer here on what to do. Try to be gracious and work with people rather than steamrolling or waving away others’ contentions because they conflict with your values.
Fourth, you need to carefully consider the safety of both participants and enumerators in your research, especially under difficult political conditions. You can’t rely on the institutional review board (IRB) alone to make ethical calls. The IRB considers participant safety, but often they are unfortunately not familiar enough with your context and research to require a sufficiently high bar. Furthermore, even if the IRB is not looking out for enumerator safety, that doesn’t mean you shouldn’t be. Especially if your research occurs in difficult political contexts, it will be your responsibility to make good ethical calls with a commitment to never put enumerators or participants in harm’s way.
Finally, other people will not understand the physical, mental, and emotional hoops you will jump through for your field research, often while simultaneously juggling “regular” job duties. Nonetheless, there is no amount of book reading you can do that replaces the number of insights and discoveries that are often gleaned through your exstoriences working directly with the actors you study.
______
Kristin Michelitch is assistant professor of political science at Vanderbilt University.
PUBLICATIONS TO WHICH THIS FIELDWORK CONTRIBUTED:
•  Beck, Jaimie, and Kristin Michelitch. “Capturing the Airwaves, Capturing the Nation? A Field Experiment on State-Run Media Effects in the Wake of a Coup,” Journal of Politics 79, no. 3 (2017): 873–89.
•  ——. “The 2012 Crisis in Mali: Ongoing Empirical State Failure,” African Affairs 114, no. 457 (2015): 598–623.
•  ——. “Is Women’s Empowerment Associated with Political Knowledge and Opinions? Evidence from Rural Mali,” World Development 106 (June 2018): 299–323.
NOTES
1. Jaimie Bleck and Kristin Michelitch, “Capturing the Airwaves, Capturing the Nation? A Field Experiment on State-Run Media Effects in the Wake of a Coup,” Journal of Politics 79, no. 3 (2017): 873–89.
2. In our World Development publication, we show that bystander presence during survey enumeration is more common for female respondents throughout sub-Saharan Africa.
3. Bleck and Michelitch, “Capturing the Airwaves.”
4. Jaimie Bleck and Kristin Michelitch, “2012 Crisis in Mali: Ongoing Empirical State Failure,” African Affairs 114, no. 457 (2015): 598–623; Jaimie Bleck and Kristin Michelitch, “Is Women’s Empowerment Associated with Political Knowledge and Opinions? Evidence from Rural Mali,” World Development 106 (June 2018): 299–323, https://doi.org/10.1016/j.worlddev.2018.01.006.