What we are testing, how classrooms are sampled, what the instrument measures, and where the design is weak. Written for researchers deciding whether to collaborate.
In the two founding classrooms beside Kasungu National Park, Malawi, almost every child said protecting wild animals mattered to them. Asked to draw the future, roughly one in nine drew a world with no animals in it at all. That gap between what children say they value and what they appear to expect is the finding this project extends.
The open question is whether it is local or general. Is the emptied-out future something children picture where wildlife is visibly declining around them, or is it what children picture everywhere? A single site cannot answer that. It needs classrooms across contrasting settings, running an identical protocol.
Each child produces two things that can be analysed against one another. This pairing is the point of the design: a questionnaire records what a child will state about wildlife, and a drawing records what they picture, which is far less susceptible to social desirability. Divergence between the two is the primary object of study.
| Instrument | What it captures |
|---|---|
| Drawing | Coded centrally on a common scheme: wildlife presence, human–wildlife interaction, environmental threats, conservation activity, emotional tone, and a separate flag for a future containing no wildlife at all. |
| Questionnaire | Four 5-point Likert items carried over verbatim from the founding study (conservation importance, self-rated knowledge, affect toward wildlife, affect about extinction); a local species-naming task with a perceived-decline marker; three open items. |
| Class context sheet | One per classroom, completed by the teacher: conservation teaching hours, visits to protected areas, everyday wildlife contact, recent lessons, school resourcing, teacher training and confidence, and the teacher's prediction of what the class has drawn. |
The prompt is fixed and read verbatim, the questionnaire always precedes the drawing, and no wildlife teaching may occur beforehand. The annual art prize applies only to a second, separate drawing made after the follow-up lesson, so that no incentive touches the drawing being analysed.
Teachers opt in as a group, so children are nested within classes that share a teacher, a curriculum and a place. Treating children as independent observations would understate the standard errors and inflate any apparent country difference. Analysis therefore uses multilevel models with class as a random effect, and the number of participating classrooms, not the number of children, is what drives statistical power.
Classrooms are sought on the dimensions the hypothesis concerns, rather than wherever volunteers happen to appear. We are seeking coverage across:
Any school is welcome to run the activity. But we actively recruit into the gaps, because an open call left to itself clusters in well-connected, English-medium schools, which is close to the opposite of the distribution the question needs.
Stated plainly, because a collaborator should know before committing and a reviewer will find them anyway.
Schools opt in. No claim can be made about children in a country as a population, and cross-country contrasts are suggestive rather than population-level. Findings are reported as holding across participating classrooms. Securing a defensible sampling frame in even two or three countries, for example all upper-primary classes in one district, would give the study a much stronger spine, and we would welcome collaborators positioned to do that.
A drawing coded "threat" in Malawi and one coded "threat" in Finland must mean the same thing or the comparison is an artefact. Inter-rater reliability in the founding study was established within a single cultural context. The extension requires reliability established across coders from different regions, with a double-coded subset and reliability reported by region. This is the open problem we most want help with.
Acquiescence and social desirability affect Likert responses, and differentially across cultures. The drawing is the partial defence, being much harder to answer strategically, which is another reason the drawing–questionnaire divergence rather than either measure alone is the primary object of interest.
The class context sheet allows findings to be related to teaching exposure and wildlife contact, but the design is cross-sectional and cannot establish that education changes what children expect. A second time-point in a subset of classrooms would be the natural next step.
The coding scheme and primary hypotheses will be pre-registered before data collection. The dataset will be published openly alongside the resulting papers.
Co-authorship on the resulting publications for substantive contributions: bringing classrooms in, contributing to coding or design, or leading a country's data collection. The dataset is published open, and no funder or collaborator influences how drawings are coded.
Write to cmccar27@jh.edu with where you are and which of the three fits. For country leads we will arrange a short call.
This extends work published as Beyond Pictures: children's drawings and conservation attitudes around Kasungu National Park, Human Dimensions of Wildlife 31(3), 2026. The instrument used here preserves that study's item wording so results remain directly comparable against the original baseline.
Download the full protocol and materials · See participating classrooms