I walk into Starbucks in Achrafieh, Beirut and feel all eyes on me. I tug at my top self-consciously, probably making things worse, and wonder a) do I look like an easy Westerner; b) do I look like a ragamuffin (in comparison to the groomed Lebanese); c) are my shoes weird for this country (I think so); or d) have I got something on my face? This feeling says as much about my state of mind as it does about anyone’s judgement, and I know from past fieldwork that the self-consciousness lessens over time. But it’s early days, so even my order is whispered and has to be repeated, shrivelling inside.
Then I find an empty seat too quickly, so that I don’t notice I’m actually boxed in and can’t easily approach anyone at all. This is all going fairly disastrously. Coffee will surely help, I think; but after drinking it I can’t say I feel any better. The girl over the way is stuck into her Kindle and won’t make eye contact. There’s a man making rather too much eye contact and I don’t want to speak to him. There are two old men, deep in a discussion I don’t want to interrupt. There is a haughty and very sexy woman I feel shabby even looking at, tapping at her diamante iPhone with long red clicky nails.
The longer I sit there the worse it gets, as I’ve been looking around for far too long and then not doing anything. Do I move onto the next cafe? That’s admitting defeat and I know it won’t be any easier anywhere else. One last look around and I spy a lovely pretty face, sitting by herself, looking out of the window. I go over and she smiles up. “I’m sorry to interrupt, do you speak English?” I say, in English, pathetically. She nods. “Are you Lebanese?” Another nod. “I’m doing some research…”
What happens when the material, the subject matter, is itself responsive to the conditions of investigation?
My first questionnaire: 119 to go—if I am to get the requisite numbers for statistical analysis, according to psychology colleagues. I myself am there for the anthropological research experience, but the same fieldtrip is being used for a psychology experiment—both approaches used to investigate moral values and inter-group conflict. Given that the groups in question are sectarian (I am looking specifically at Christian, Sunni, and Shia, the three biggest confessional sects in Lebanon), this is touchy stuff, in a country which was riven by a bloody, sectarian civil war for 15 years (1975-1990), only concluded with a ceasefire agreement and a modicum of political reform, keeping the basic sectarian political system intact. The questionnaire for the psychology experiment has been meticulously prepared, with many pilot versions and repeated checks from a variety of academics, but prepared partly with a view to antagonism in order to elicit certain reactions. “The questionnaire makes me uncomfortable too,” I would confess to participants, “and I much prefer talking to people to find out what they think.” This would make people laugh and open up, or laugh and ask why on earth I was doing research I didn’t like—which in turn would make me laugh and get the conversation going.
The questionnaire seems a simple instrument, albeit for a complex experiment: Sit down with people, and ask them a consistent set of questions. This work is for an international and interdisciplinary consortium to which I belong, Artis International, which studies conflict in the hope that public policy can be shaped to minimize it. But as I worked my way through cafes and towns, I began to wonder—was I gathering information to be neatly sorted, classified, and processed by models constructed by psychologists and anthropologists? Or was I engaging in complex interactions with people at an arbitrary moment in time and space, in ways that were impossible to capture precisely, by me or anyone else?
The question of how to divide up knowledge has occupied many a philosopher, although it seems straightforward based on subject matter. Zoology studies animals, history the past, psychology the mind, geology the earth, literature—well, literature. Universities largely follow these disciplinary divisions for historical and practical reasons; that is, they are inherited from medieval, Aristotelian universities, but the system also has the advantage of gathering people together who are thinking about the same things, and can share ideas and results. The world is divided into things, and our investigations into these things mirror these divisions.
Then, of course, are the boundary areas or overlaps. Palaeontology and zoology look at the same bivalve collections, both fossilised and more recent, to study and teach evolution; history and literature delve into social mores and customs from the past, often using the same texts; anthropology, psychology, and sociology all investigate people and their interaction with their cultural contexts. So how do we distinguish further between the disciplines when the subject matter is the same? Methodologies, we might think, or, intertwined with this, the questions asked of the material and the answers attempted by each discipline.
This picture relies, though, on the assumption that there is solid, robust, and unchanging “material” to interrogate, with each practitioner coming up in turn to look and probe and ask questions and return armed with their conclusions, in response to their questions. But what happens when the material, the subject matter, is itself responsive to the conditions of investigation? The most famous example of this is in quantum physics and wave-particle duality; particles such as electrons behaving like waves when looked at with double slits (interference patterns emerge), and particles when scattered by an electro-magnetic force. Neither form can explain all characteristics of a particle.
People may emerge as a set of characteristics and behaviors that they themselves might not recognize.
Humans as well as sub-atomic particles react to how they are being investigated: We know this from personal experience as well as academic, and disciplines have developed legion and sophisticated ways of circumventing these tendencies. Psychology repeats its experiments as much as possible, the persistently resistant subjects seen in the statistics as outliers.
“Historically,” says Hammad Sheikh, a postdoctoral fellow in psychology at the New School for Social Research and my colleague at Artis International, “the methodology derives from astronomy,” referring to averaging results of repeated observations to compensate for error in imperfect instruments or environmental conditions. “Psychology borrowed this approach (and terminology): a number of observations are made (the respondents are seen as a repeated measurement) using psychological instruments (e.g., questionnaires), and the results are then aggregated to remove measurement error.”
There is a clear danger in this approach: the risk of over-generalizing in the search for underlying structures, and dismissing anomalies (“statistical outliers”) instead of tracing their significance. And there is worse. “The search for universals in psychology,” says Scott Atran, anthropologist at France’s National Center for Scientific Research, Oxford University, John Jay College, and the University of Michigan, “is vitiated by essentialism, the belief that the phenomenon in question has a fixed set of inherent, inalienable traits that make it what it is.”
Anthropologists are attracted by the lure of the universal, too. For example, Harvey Whitehouse, an anthropologist at the University of Oxford, believes in finding the underlying patterns of human behavior, but using historical information rather than psychological, taking his research into evolutionary cognitive psychology. But many are wary about such methods, uneasy about looking for such all-encompassing theories. “Social scientists are reductionist in their hearts as well as their methods,” Baruch Fischhoff, Howard Heinz University Professor at the Carnegie Mellon University, tells me. “They want to have simple descriptions for complex phenomena,” he says.
The worry for Fischhoff, myself, and Atran alike is that essentialism can mean over-simplification and blindness to crucial features if they lie outside the scope of existing theory. This is especially true when applied to the complex, shifting, obfuscating entities that are humans. People may emerge as a set of characteristics and behaviors that they themselves might not recognize. Respondents in Lebanon would ask why there was no question about their international background, for example, which simply would not be a factor in many countries but is huge in a country where so many people moved away during the civil war, bringing up children, now adults, in Europe, North America, the Arab Gulf, North Africa, and so on; where so many are educated in English or French and subsequently study abroad, perhaps for a year, perhaps settling in their new homes for years; where so many work around the world sending money home (Lebanese do not need a visa to work in other Arab countries). This international flavor and its effect on responses is not captured when the object is to understand an underlying structure of human behavior based around essential, not accidental, characteristics.
Sheikh is unperturbed by this problem: “Having this information is useful because it can be potentially controlled for in the data analysis,” he agrees when I point out this response, “but it is not crucial ... All these factors which may alter individual responses are considered ‘noise’ in the measurement. The methods are designed to discard this noise, and work as long as the sample is large enough.”
In the field the “inaccurate instrument,” the source of measurement error, is the experimenter herself: The interaction between the interviewer and respondent cannot be kept precisely the same in any possible world where humans are humans—the so-called “experimenter bias.” This can be kept to a minimum with standardizing the introduction, questions, environment, and so on, never fully, but, with lab conditions, enough to eliminate error with aggregation and averaging. What complicates the process enormously is conducting the experiments outside the lab, especially in a culturally diverse, socially divided, conflict-plagued country such as Lebanon.
I kept my opening gambit the same (more or less) but there couldn’t possibly be an identical script for each explanation, not least because people react very differently to being asked by a stranger to fill out a questionnaire that takes at least half an hour. Some accept with alacrity, barely listening to instructions. Others very naturally ask “What is it for?” or “What is it about?” or the more suspicious “Who are you and where do you work?” (the eternal burden of being a white westerner in the Middle East is the suspicion that you are a spy). Others want to get into a conversation straight away about the topic, from which I have to gently dissuade them, assuring them there’ll be time for talking afterward. But crucially I cannot reply to every person with the same phrases as they ask different questions in different ways, and my explanations vary according to educational background, linguistic levels, and the extent of a shared cultural context. For example, people may make a joke that I respond to, establishing a different relationship to someone I switch into Arabic with in order to communicate, a language in which I would no doubt miss a joke if it were offered, let alone respond to it.
And this is not to mention other impossible-to-control variables such as the background noise, an electricity cut, what the interviewees have done that day, how much sleep they had, the appearance of a friend, a phone call, and so on. Fieldwork is a messy business—to the delight of the anthropologist, but the sorrow of the psychologist. This isn’t a disaster, of course. Yes, the data is more “noisy” in the field, Sheikh admits, but for the analysis you just “need more participants than in lab experiments to compensate for the lack of control.” Hence the desire for 120 respondents, minimum, a number constantly murmuring away in the back of my mind during my fieldwork, tallied up agonizingly slowly. But the numbers game brings other challenges, for often friends and family are passed on to help me out, which brings a lack of “independence of observation,” as Sheikh describes.
Subjectivity? Sheikh, a psychologist, is cynical. “Most research psychologists believe that the subjective experience of humans is not directly related to the ‘underlying causes’ of their behavior,” he tells me. He points to a landmark study in 1977, in which participants were asked to choose from a number of pairs of nylon stockings, and then give the reasons for their decision. Participants were quick to oblige, describing a host of qualities like softness and warmth. But in reality the pairs were identical, and most participants picked the pair on the right. “The subjective reasons for the choice did not have any connection with that choice,” says Sheikh. “Since this study, there have been a large number of studies demonstrating the independence of insight and causes. So, psychologists are very skeptical about subjective insights of participants. Instead, they try to uncover relationships between psychological variables using statistical methods.”
When I come back five months after my first experience of psychology in the field, things seem—subjectively—very different. Not just in the lowering, humid heat of August Beirut versus the spring air of Mediterranean March, but in the palpable sense of unease. The Islamic State group had just attacked the town of Arsal, on Lebanese soil, briefly capturing some ground and killing and kidnapping Lebanese soldiers before being driven back over the border into Syria. Given that the population’s experience of civil war was still strong in many memories, the fear was real and visceral: They knew what war could bring to their country.
I had spent the first week or so in March refining my approaches to potential interviewees, and developing responses for any resistance I encountered once people had started the questionnaire and discovered its sectarian nature. And so I had thought that I would be able to hit the ground running six months on. But no.
After a few days of a lack of sectarian self-identification on the returned questionnaires (which made them useless for the stats), I had to conclude that this wasn’t just a string of unlucky encounters. There was something going on. I began spending more time on each person, using the same sorts of phrases I had developed earlier but much more thoughtfully and with more time for each respondent. After a longer initial conversation, and the completion of the questionnaire, participants often wanted to talk more, which I was keen to do, but it meant the research was often very slow. 120; 120; 120 was still beating away at the back of my mind and would not go away—so I worked longer hours (and finally employed research assistants—opening a whole other can of worms).
I began to ask: “Why is my research so much harder this month than in March?” I would tell participants that the questionnaire had been adjusted to remove many of the problematic questions, but the responses were actually worse. Many started evasive, “You are probably just talking to the wrong people.” But on being pushed, the attack by ISIS on Arsal earlier in the month would come up. It was affecting everything in Lebanon, I was told: a backdrop of fear.
After weeks doing research in the field, one of my most vivid observations was simply missing in my own data.
“There is no sound louder than the sound of war,” one Christian 20-something said; “we cannot think of these sorts of questions when we can hear such a sound.” His friend chimed in: “The war has come to Lebanese soil, and we Lebanese know about war.” Others echoed these sentiments: “We must unite at these times as we know what the alternative is, and know that we can destroy each other through civil war.”
This is perfect fodder for my psychology colleagues, I think: A comparison with responses from March would yield all sorts of information about people’s attitudes to out-groups during a time of stress. Maybe there is increased unification in the country in the face of a common enemy?
Not so fast. The sectarianism I observed in March emerged as strongly as ever, as conversations unfolded. If I asked about more specific fears than civil war (“Do you think the Islamic State group will try again? Do you think the Lebanese army is strong enough?” and so on), fears of how the army would split arose, especially amongst Christian respondents (“The Muslims have strong militias and would defect to join them rather than keep loyal to a Lebanese force”); the Sunnis would often fear reprisals given the sectarian identity of the Islamic State group and the sectarian brutality shown in Iraq and Syria (“this is not our Islam, we have never had extremists here” I was told many times, “But still we will be identified with ISIS because we are Sunnis”). The Shia Hezbollah were named far less frequently in the surveys as the main threat than in March, but in longer conversations they would still be blamed for many of the threats to the country (their role in Syria, their control of politics and their military might, their intimidation tactics), and feared for further power grabs and reprisals against other communities.
This might sound callous, but again I had the thought that this would be good for the research and my questions concerning sectarianism, moral values, and attitudes to the out-groups within the country. I probed as many people as I could, sometimes meeting with respondents again, as an ethnographer would appreciate, though it didn’t add to the tally.
Then I reached, then passed, my benchmark: 144! Success beyond my dreams! All scanned, uploaded, and delivered; hard copies sent registered delivery—my luggage doesn’t take that sort of weight. I headed back to New York pleased with the repeated experience of drawing out clear, often extreme, responses, which I felt must emerge from any method of analysis, statistical or other. If my respondents were showing such clear patterns face to face, surely the patterns would also be seen in the responses in the surveys. And so I predicted with confidence to my colleagues that the statistics would point to clear conclusions concerning how people respond under increased threat of conflict.
The psychologists got to work once the data entry had been done—and there was nothing. No correlations at all, despite the promising pilot in March when my conversations had been less clear-cut, more varied. Nothing to connect fusion to the individual’s community, which I had seen as so very much stronger in August, with their moral values, or hostility to other sectarian groups, or willingness to commit extreme sacrifices for the group (unto killing and dying for the group)—although again I had experienced this, albeit “subjectively.” There was no supporting evidence either for or against my predictions. The differences I thought I’d seen—that I’d subjectively experienced—were not reflected in the completed questionnaires. After weeks doing research in the field, one of my most vivid observations was simply missing in my own data.
I cannot relinquish my own understanding of Lebanon gained through painstaking interviews of a range of people, whatever my colleagues across the social science divide might conclude from the data. But I can appreciate the critiques of subjectivity and limitations in my methods: My conversations were neither straightforward nor transparent and colleagues clarified this for me. People responded to my status as an academic, a westerner, a woman (many of the questionnaires, the data entry research assistant later told me, included phone numbers and invitations in the margins), and no doubt other features I am not even aware of. These all of course contributed to the nature and the content of the conversations which I must reflect on in assessing the information, but I still trust many of my own conclusions—especially those to do with fear and sectarianism which no one, I believe, could miss. The challenge for me now is to work with colleagues to understand why the stats did indeed miss it, and in this way develop our understanding of the context, rather than abandon the collaboration.
I suppose it is in the nature of the academic to be optimistic. Atran tells me that he “would love to be able to focus on actually figuring out what a causal science of human beings might actually look like, linking individual cognitions, social networks, local institutions, and ecologies in causal, materially identifiable chains.” Is this possible? I think so, but it will likely require a new approach. I would like to see, for a start, both the majority and the outliers included in the analysis; the different realities and assumptions of subject, interviewer, and all involved researchers taken equally seriously; the long and short term, from the evolutionary timescale to the political, given attention; statisticians given a free rein while adapting to ethnographic information; “subjective” experimenters understanding the demands and rewards of numerical analysis; and a contribution from the humanities in learning how to understand the micro and contingent.
With these goals in mind, here I am, planning my next fieldtrip in Iraq. There I will gather up my flawed instruments and use them as exhaustively as I can in collecting the stories, in their own words, of those caught up in conflict—not because I’ll get it exactly right, but because their stories need to be heard and preserved. Who knows what light some radical new interpretative framework might cast on them?
Lydia Wilson is a research fellow at the Centre for the Resolution of Intractable Conflict, University of Oxford; research fellow and field director at Artis International; and visiting scholar at the Ralph Bunche Institute, Graduate Center, CUNY. She edits the Cambridge Literary Review.