Communicating spatial uncertainty matters. Here different groups of stakeholders consider usefulness of statistical information

November 12, 2022, by Lexi Earl

How should uncertainty in spatial information be communicated?

This post is written by Dr Christopher Chagumaira with contributions from Prof Murray Lark. 

Many and varied people have to make decisions about environmental management, be they farmers, policy-makers or managers, and spatial information about environmental variables (e.g., soil properties) is essential for this task. Within the Future Food Beacon, the GeoNutrition project is concerned with how soil might limit the supply of micronutrients (e.g., iron, zinc, selenium) which are essential for human health, and how interventions to address deficiencies of these can be targeted efficiently. Large datasets of soil and crop information were collected from smallholder farms in Ethiopia (2017/18) and Malawi (2018), in sub-Saharan Africa. Spatial predictions were made from the data collected from the surveys to generate the kind of information needed to support decisions on interventions, but these predictions are uncertain.

Various statistical methods can be used to combine the field survey data and other data sources (e.g., satellite data) to produce improved spatial predictions and to measure their uncertainties quantitatively. These statistical measures of uncertainty are indigestible for decision-makers. We examined this problem by eliciting stakeholder’s opinions on the accessibility of different statistical descriptions of uncertainty.

Our findings confirmed that many of the statistical descriptions of uncertainty are difficult to understand. However, different groups of stakeholders could make sense of information when the measure of uncertainty was linked directly to its interpretation, for example by presenting the probability that the concentration of a nutrient in a crop exceeded a nutritionally significant threshold. Therefore, the GeoNutrition project has used such outputs to present its data.

Presenting uncertain information so that stakeholders can make sound decisions with it is not a simple task and requires perspectives from psychology as well as statistics.  We have shown, for example, that the same information is interpreted differently if it is `framed’ in terms which draw attention to the general idea of nutritional deficiency rather than sufficiency.  Our experimental approaches provide a basis for finding ways to frame information so that stakeholders can make robust decisions which reflect their understanding of risks in transparent ways.  This is important, because information is costly, and the decisions based on it can have significant effects on people’s lives.  This is not only true of decisions about nutrition, of course, but applies across many important global issues from the impact of climate change to how we can ensure food and water security.  Lessons from the GeoNutrition project are likely to be relevant to all of these.

Posted in COP