Using Children’s Drawings in Program and Policy Evaluation
I want to devote this post to a new tool that I have begun to use with colleagues in program and policy evaluation, the digital coding of children’s drawings. An example appears in our newly forthcoming paper in the Journal of Human Resources, “Developing Hope Among Impoverished Children: Using Child Self-Portraits to Measure Poverty Program Impacts.” One of my graduate students, Rafael Panlilio, is also using the technique to understand the impacts of re-settlement using a sample of over 1,000 Syrian refugee children from the Za’atari camp to urban areas in Jordan.
Children’s drawings have been used for decades in clinical practice to help psychologists diagnose psychological disorders in children. Until recently, this had been done subjectively in the context of psychotherapy sessions, where a therapist asks a child to draw a self-portrait, or sometimes to free draw, to support diagnosis. When studying the well-being of children, drawings have a number of advantages over answers to survey questions. Children sometimes have difficulty communicating and expressing feelings; drawings often reveal more than spoken words. Drawings represent observed rather than reported behavior (important for economists) that can reveal complex facets of mental process and emotion. We have found that children’s drawings can also be digitally coded and studied objectively rather than just subjectively. From digital coding via a set of 1/0 values representing the presence or absence of a characteristic, it is possible to create indices of psychological phenomena like hope, depression, self-efficacy, stress disorders, and anxiety for use in evaluating differences in children’s psychological health. This can help us to understand how different types of humanitarian disasters, policies, and interventions affect the internal well-being of children.
A long empirical literature in clinical psychology has established correlations between characteristics of children’s self-portraits and different facets of psychological health and disorder. For example, choice of dark over light colors has been consistently been found to be correlated with depression and anxiety, a tiny figure indicates low self-esteem, and a monster figure with aggression. (See references for these in our main paper.) Many of these empirical correlations are fascinating, and summarized in the table here.
A few years ago a Bolivian psychologist, Marcela Berazaín, introduced a group of us (Paul Glewwe, Phillip Ross, Laine Rutledge, and myself) who were working on an evaluation of international child sponsorship programs to the technique. She suggested that as part of our impact evaluation we should ask Bolivian children in both treatment and control groups to “draw a picture of yourself in the rain.” She explained to us that the rain was important because it represented a potential discomfort, an obstacle, a challenge that needs to be dealt with. Does the child see herself as a victim? Does she see herself able to cope with the rain by seeking shelter or using an umbrella? Does the he optimistically draw a sun behind the clouds–or throw in lighting and wind (not called for by the instructions) along with the rain? We piloted this exercise among a smaller number of Bolivian children and found some compelling correlations, but our main study took place among internationally sponsored children through Compassion in Indonesia.
In our main study on the impact of child sponsorship, we analyzed 540 drawings by children in the slums of Jakarta. We use the natural experiment of an age-eligibility rule for identification of causal effects to find that sponsorship significantly elevates hope (0.66 standard deviations), happiness (0.42 s.d.), and self-efficacy (0.29 s.d.) among sponsored children. 
Children ages 5-16 were given a box of 24 colored pencils to do their self-portrait. Figure 2 gives examples of children’s drawings that show variation in the three factors. Panel A illustrates drawings of two boys of roughly the same age ranking in the 17th and 92nd percentiles of Happiness, respectively, principally due to differences in facial expression, body language, and inclusion of a sun. Panel B shows children’s drawings ranking in the 8th percentile and 94th percentile in Optimism/Self-Efficacy. Note in B1 the use of a single color, the presence of lightning, and poor integration of body parts. These stand in contrast to the multiple light colors used in B2, the presence of a sun above the clouds, and the inclusion of an umbrella. Panel C illustrates differences in Hopefulness, where the drawing in C1 (7th percentile) was by a teenage girl and the one in C2 (85th percentile) by a primary school age boy. Note the missing facial features and hidden limbs in the girl’s self-portrait on the left, which are correlated with hopelessness and depression. In contrast the bright colors used by the boy on the right, facial expression, full illustration of facial features and limbs, and use of the umbrella are features correlated with hopefulness, despite the lower quality of the artwork.
In other exercises, children can be asked to free draw. In Rafael’s study on Syrian refugee children, this was done along with a self-portrait during a survey session with the parents. When children are given a blank piece of paper and asked to draw a picture of anything they wish, one is able to learn about what a child is thinking and feeling in a non-threatening environment.
Preliminary findings from his thesis find that children in the Za’atari refugee camp are significantly more likely in their free drawings to depict scenes that included the types of images correlated with post-traumatic stress syndrome (PTSD). Two of these drawings can be seen here. These are tough images for anyone to see, revealing the kinds of violence to which Syrian children have been exposed. 
The images may be more present among children in the camps perhaps because living among other refugees may cause them to more easily recall traumatic and violent experiences from the war. Assimilation of refugees may help children to psychologically move past the trauma at least to some degree, but in no way does this imply that they do not have damaging and lasting effects from it. However, it does offer some evidence against the policies of countries (such as ours now, sadly) that would keep children and families in refugee camps rather allowing them asylum.
- The Behavioral Economics of Why Hyperopic People Like Me Prefer Gift Cards over Cash
- How Does Hope Work?