Commentary on Townsend, Taylor, Merrilees, et al. (2020). Youth in Northern Ireland: Linking Violence Exposure, Emotional Insecurity, and the Political Macrosystem.

About the Author
J. Lawrence Aber

Applied Psychology Department,
New York University

Dr. Aber is Willner Family Professor of Psychology and Public Policy at NYU Steinhardt and University Professor at New York University.  His research: examines the influence of poverty and violence on the social, emotional, behavioral, cognitive, and academic development of children and youth, examined at both the family and community levels. In addition, he tests the impact of complex social interventions on the development of children who have been exposed to poverty and violence. 

Lessons from Youth in Northern Ireland:
Building Measures, Methods, and Interventions

I remember being a doctoral student of Edward Zigler’s at Yale in the 1970s and learning for the first time about the emerging work of Urie Bronfenbrenner at Cornell. Ed had invited Urie to give a departmental colloquium on a book he had begun to write and would later publish as The Ecology of Human Development (Bronfenbrenner, 1979). I was thrilled to hear someone proposing a conceptual framework for empirical research on human development that captured the person-process-context-time complexities that I was already convinced must shape and be shaped by human development.

But my heart began to sink as I started to contemplate the enormous practical challenges confronting researchers who wished to conduct empirical research faithful to this ecological framework. Which constructs at what levels of human ecology were critical, and which outcomes of human development should be studied? How could constructs like these be measured reliably, validly, and feasibly? If researchers could identify and measure these constructs across time, how could they analyze the data to estimate the dynamic, multi-level, reciprocal effects of context and process on developmental outcomes? Finally, and of great concern to scholar-activists like Zigler, how could this emergent science of human development be harnessed to understand and address the real challenges facing children, youth, and their families? 

Happily, during the last half century, there has been great progress in meeting these research challenges. As evidence of this progress, witness the remarkable new monograph by Townsend and colleagues (2020). It draws from and extends the rich longitudinal study on the impact of political violence on youth in Northern Ireland that had been initiated by Cummings and colleagues a quarter century earlier. Their monograph combines creative, cutting-edge methods that measure features of the macrosystem with data that address the microsystem of adolescents. They use multilevel dynamic modeling of longitudinal data on processes-in-context to explore the impact of political violence in relation to youth’s personal qualities, identities, and experiences. Urie would undoubtedly have been gratified to learn that his ecological framework could stimulate and help organize such ambitious, innovative empirical research on such an important phenomenon. We have come so far…and yet there is still so far to go.

It is helpful to begin by reviewing key questions posed by the authors: (1) Does youth exposure to sectarian violence (in Northern Ireland from 2006 to 2011) result in heightened emotional insecurity about their communities? (2) If so, are these effects further heightened in a tense or threatening sociopolitical climate? Four decades ago, it would have been impossible to answer these questions in rigorous empirical work. But it has now become possible, in part because investigators like Townsend and colleagues have developed new ways to measure key constructs—exposure to sectarian violence, emotional insecurity in the community, and exposure to sociopolitical tensions and threats. Their new conceptualizations and measures not only allow them to provide new empirical insights; they also inspire important refinements and extensions, some of which I consider here.

Exposure to Sectarian Violence. The authors make a strong case for why they chose to develop a new measure of violence rather than using one developed by others. Like Richters and Salzman (1990) who developed a measure of community violence for use in the US, and like Macksoud and Aber (1996) who developed a measure of exposure to war-related events in the Middle East, Townsend and colleagues developed a measure that was culture-specific, this one designed to study sectarian violence in Northern Ireland. Fair enough. I understand and admire the authors’ epistemological humility and ethical care in focusing first on within-culture salience and validity of exposure to violence. But will such culturally specific measures ultimately limit the field’s ability to synthesize findings across studies of political and community violence? When, for example, researchers find different associations between exposure to violence and developmental outcomes, how will we know if the observed differences are rooted in fundamentally different processes across contexts or if they are instead the consequence of non-comparable measures? As the field has long recognized, what is needed is a taxonomy of types of exposure to violence and measures that allow researchers to operationalize and examine important variations among these types.

Emotional Security in the Community. The construct of emotional security is critical to the authors’ theoretical formulations because it is proposed as the mediating mechanism by which exposure to sectarian violence impacts adolescents’ adjustment. But are the measure and the conceptualization of the construct well-matched? Recall that the survey did not involve interviewing youth themselves, but rather asked mothers to report on their children’s felt security. The authors explicitly recognize this as a limitation of their method, but likewise note the value of this measure for their work given that it offered data across all five waves of their longitudinal study. I agree about its value, but it may be wise to consider the possibility that the measure is tapping a somewhat different construct, perhaps one that might be labeled “mothers’ worries about their children’s safety in the community.” This would be a different theoretical construct, but one which is also informative, and one which can be profitably examined within family systems theory and attachment theory. It is another reminder of the breadth of the person-process-context-time complexities captured in Bronfenbrenner’s model.

Tension and Threat of the Sociopolitical Climate. The authors’ most significant methodological innovation was assessing the degree of sociopolitical tension and conflict at the level of the macrosystem by systematic coding of newspaper accounts. This method is a creative way to measure important features of the macrosystem in which youth of Northern Ireland live. But as the authors noted, theirs is a proxy measure of the construct of interest—political violence. I encourage these authors as well as other researchers to develop, validate, and compare additional measures of tension and conflict in the sociopolitical climate. Ideas for such measures may be drawn from two large-scale data-collection projects that produce cross-national, georeferenced, disaggregated events-level conflict data: the Uppsala Conflict Data Program Geo-referenced Events Dataset; and the Armed Conflict Location Events Dataset (see Eck, 2012). If developmentalists are to create measures of key macrosystem constructs, it would be wise to collaborate with other researchers of macrosystem phenomena (e.g., sociologists, political scientists). Developmental scientists should strive to bring the same rigor to the “sociometrics” of context-level constructs that it has already brought to the “psychometrics” of micro-level constructs.

Cross-cutting Analytic Issues. Another enormous cross-cutting challenge in this study (and in all such studies, including my own) is the issue of omitted variables. What key constructs are not included in this study that could—were they included—significantly affect the pattern and interpretation of results? For example, exposure to sectarian violence is probably associated with other factors that could affect youths’ (or their mothers’) perceptions of how safe and secure those youth are in their community. Three illustrative candidate constructs are exposure to nonsectarian violence, exposure to family discord and violence, and anti-social peer networks. When unmeasured and hence uncontrolled for, the variance these constructs share with exposure to sectarian violence becomes part of the association of the measure with emotional security in the community. Similarly, mothers’ perceptions of their children’s emotional security in the community are likely to be confounded with maternal anxiety and depression, a confound which could serve to inflate its observed links to adolescent adjustment problems. These are well-known challenges in developmental research as is the broader problem of endogeneity in developmental research (Duncan et al., 2011). The authors admirably use multiple baseline covariates and time precedence to limit these problems, but they remain nonetheless, and are well-worth worrying about.

Developmental science has long struggled with the problems of omitted variables and their potential to inflate, deflate, or suppress associations between variables at the person or microsystem levels. But as a discipline, we are in new territory when thinking through important omitted variables at the level of the context or the macrosystem. For instance, in addition to tension and threat in the sociopolitical climate, there is documentary evidence of the presence of positive countervailing factors at the macro level, for instance, efforts mounted by elements of civil society to bridge the divide between Catholic and Protestant communities and to build a sense of peace and social cohesion. Are there other countervailing forces to a tense and threatening sociopolitical climate that affect youth and their parents across religions? Could coding of news accounts help identify positive features of climates such as peace building and social cohesion?

Measuring and modeling the influence of co-existing positive forces at the macro level hold promise to help explain the relatively weak and sometimes surprising effects of tension and threat on the association between exposure to sectarian violence and emotional security in the community. As a thought experiment, imagine having two additional measures in the authors’ models: one at the micro level (exposure to nonsectarian violence) and one at the macro level (community-level social cohesion). Among the many new questions enabled by the addition of these two variables, two come to mind: (1) what are the separate, combined, and unique effects of sectarian and nonsectarian violence on youth development? (2) can positive features of the sociopolitical climate (e.g., social cohesion) buffer youth from the negative features of the sociopolitical climate and from exposure to both sectarian and non-sectarian violence?

Cross-disciplinary perspectives. As I suggested earlier, other research disciplines (e.g., sociology) and other theoretical perspectives within those disciplines (e.g., social disorganization theory) developed largely in the US could be drawn upon by developmentalists who study the effects of political violence on children and youth in other countries. Indeed, studies of the impact of community-level poverty and violence on children and youth (“neighborhood effects”) in the US come from a rich and long tradition (e.g., Brooks-Gunn, Duncan, & Aber, 1997a, 1997b). These literatures have potential implications for both theory and empirical research on the effects of political violence. For instance, the work by Robert Sampson, Felton Earls and their colleagues from the Project on Human Development and Chicago Neighborhoods (PHDCN) have identified the critical role of collective efficacy (social cohesion among neighbors combined with their willingness to intervene on behalf of the common good) in explaining variation in levels of community violence. Poor communities with high levels of collective efficacy appear to protect children and youth from violence better than do equally poor communities with low collective efficacy (Sampson, Raudenbush, & Earls, 1997). Another methodological innovation emerged out of the PHDCN. This is Patrick Sharkey’s work on estimation of the causal impact of specific discrete incidents of community violence on child and youth development. Sharkey compared children who were assessed for cognitive performance (reading and vocabulary measures) either within the week following a homicide incident in the immediate neighborhood or within the week prior to the homicide incident. Because the cognitive assessment dates had been scheduled prior to the homicide incidents, this study’s design constitutes a form of a natural experiment. By establishing a clear counterfactual that controls for all confounds (as in a well-designed and well-executed true experiment), it was possible to conclude with certainty that exposure to violent incidents causes decrements in children’s cognitive performance. As is broadly appreciated, identifying causal processes can be invaluable for the design and evaluation of prevention and intervention efforts.

Harnessing developmental-contextual research on the influence of political violence to inform action. By the time Edward Zigler invited Urie Bronfenbrenner to Yale in the 1970s, they had already spent much of the prior decade designing and evaluating the Head Start program which was part of the War on Poverty. Head Start was one of the first major systematic attempts to use scientific knowledge of child development to design a social program. Soon the field of Child Development and Social Policy began to blossom (Aber et al., 2007) and developmental science was increasingly brought to bear on understanding and addressing social problems. Since then, there has been an explosion of activity in numerous sectors which fuses developmental science with prevention and intervention science.

I know that Townsend and colleagues aspire for their work to be put to practical use. And it should…but how? There is a growing field of research and action on how children’s learning and development may be supported through conflict- and crisis-affected contexts like those of Northern Ireland (Aber, et al., 2020, in press). Such work is growing in Africa, the Middle East, Latin America, and Southeast Asia, all areas in which political violence and armed conflict rage. One type of high priority activity suggests itself in contemporary times. We could bring together researchers who study development in violent contexts (e.g., the monograph’s authors) and interventionists who try to protect children from exposure to violence or try to mitigate the negative effects of such exposure (e.g., those working with NGOs). Cross-national groups of researchers and practitioners could meet regularly to (a) identify causal mechanisms by which exposure to violence leads to unwanted consequences and (b) brainstorm to design and test interventions to disrupt dysfunctional trajectories and thereby improve child and youth outcomes.

As one small example of this approach, my colleagues and I at NYU, in collaboration with the International Rescue Committee (IRC) and the Ministry of Education on the Democratic Republic of Congo (DRC), evaluated the impact of an intervention designed by IRC in the eastern DRC via a cluster-randomized trial. The eastern DRC had been the epicenter of what has been called “Africa’s World War” involving political violence and armed conflict in seven nations. The intervention, “Learning in Healing Classrooms” (LHC), infuses social-emotional learning principles into literacy and numeracy curricula and trains and supports teachers in the implementation of the curriculum. In two independent cohorts of schools, we found LHC improved conflict-affected children’s perceptions of their schools as safe, and their teachers as supportive. These changed perceptions were, in turn, associated with improvements in children’s literacy and numeracy skills, although not with their social-emotional well-being (Aber et al., 2017).

The provision of safety and support is central to fostering felt security. Could Townsend and colleagues begin discussions with civic organizations in Northern Ireland to draw on their exquisite research to help design and evaluate intervention strategies? Such strategies could target improvements in youths’ sense of emotional security, safety, and supports in their micro contexts. I believe such interventions would then serve as buffers against the impact of political violence on children and youth.


Aber, J. L., Bishop-Josef, S. J., Jones, S. M., McLearn, K. T., & Phillips, D. L. (Eds.). (2007). Child development and social policy: Knowledge for action. Washington, DC: American Psychological Association.

Aber, J. L., Tubbs, C., Torrente, C., Halpin, P. F., Johnston, B., Starkey, L., Shivshanker, A., Annan, J., Seidman, E., & Wolf, S. (2017). Promoting children’s learning and development in conflict-affected countries: Testing change process in the Democratic Republic of the Congo. Development and Psychopathology, 29, 53-67.

Aber, J. L., Tubbs Dolan, C., Kim, H., & Brown, L. (2020, in press). Children’s learning and development in conflict- and crisis-affected countries: Building a science for action. Development & Psychopathology, Special Issue – Legacy of Edward Zigler.

Brooks-Gunn, J., Duncan, G. & Aber, J. L. (Eds.). (1997a). Neighborhood poverty I. Context and consequences for children. New York: Russell Sage.

Brooks-Gunn, J., Duncan, G. & Aber, J. L. (Eds.). (1997b). Neighborhood poverty II: Policy implications for studying neighborhoods. New York: Russell Sage.

Bronfenbrenner, U. (Ed.) (1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press.

Duncan, G. J., Magnuson, K. A., & Ludwig, J. (2011). The endogeneity problem in developmental studies. Research in Human Development, 1(1-2), 59-80.

Eck, K. (2012). In data we trust? A comparison of UCDP GED and ACLED conflict events datasets. Cooperation and Conflict, 47(1), 124-141.

Macksoud, M. S., & Aber, J. L. (1996). The war experiences and psychosocial development of children in Lebanon. Child Development, 67(1), 70-88.

Richters, J. E. & Martinez, P. E. (1993). The NIMH community violence project: I. Children as victims of and witnesses to violence. Psychiatry, Interpersonal & Biological Processes,56(1), 7–21.

Richters, J. E., & Saltzman, W. (1990). Survey of exposure to community violence: Self report version. Rockville, MD: National Institute of Mental Health.

Sampson, R. J., Raudenbush, S. W., & Earls, F. (1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277(5328), 918-924.

Sharkey, P. (2010). The acute effect of local homicides on children’s cognitive performance. Proceedings of the National Academy of Sciences107(26), 11733-11738.

Townsend, D., Taylor, L. K., Merrilees, C. E., Furey, A., Goeke-Morey, M. C., Shirlow, P., & Cummings, E. M. (2020). Youth in Northern Ireland: Linking violence exposure, emotional insecurity, and the political macrosystem. Monographs of the Society for Research in Child Development, 85(4).

Aber, J. L. (2020). Lessons from Youth in Northern Ireland: Building Measures, Methods, and Interventions. [Peer commentary on the article “Youth in Northern Ireland: Linking Violence Exposure, Emotional Insecurity, and the Political Macrosystem” by D. Townsend, L. K. Taylor, C. E. Merrilees, A. Furey, M. C. Goeke-Morey, P. Shirlow, and E. M. Cummings]. Monograph Matters. Retrieved from