By Omri Shoshani
As loss and damage has grown into a central discussion during the recent Conference of the Parties under the UNFCC, and will undoubtedly play an important role during COP30, it warrants reflecting on the evidence that will be used to educate future decision making. This blog post is meant to incite the necessary criticality of any individual scientific endeavor, while embracing the greater pursuit of good science and knowledge.
Data reflects both an observer and the observed. When information is stored as data, it is collected, selected, and sampled based on the ability and “sight” of its captor. When we access data as a proxy of that original information, we are fundamentally constrained by the limitations of the data itself. Any researcher or policymaker will know that even the most meticulous, carefully planned analysis is doomed if the data used is poor. Furthermore, our analyses are never perfect and more information is always lost before a conclusion is reached. By the time true information has become data, and appears in the results of a paper, article, or policy brief, it is only a fraction of what it was.
What I mean to communicate is that one should remember in these coming days, that for all the negotiation and discussion of the impacts, losses, and destruction of a changing climate– we are limited to our data, and our data is imperfect. Exactly how much information is lost depends on what is measured. The data on CO2 emissions, global surface temperature, or cloud cover is nearly lossless, a convenience of measuring the physical and natural world. In contrast, data on social impacts, human loss, economic loss, and psychological suffering is far from complete. When information is lost, we should take care to consider which information is lost. Since data only captures what its captor could observe, we must imagine what they could not observe.
In surveys of human populations, there are two groups who are most often systematically omitted: the poorest, and the richest. The poorest are difficult to observe for many of the same reasons that make them vulnerable, such as impermanent housing or missing registration in governmental systems. We sometimes call this undercoverage bias. The richest are difficult to observe because they do not want or care to be, though the reasoning may differ depending on what is being measured. This can be called nonresponse bias.
These issues extend more abstractly beyond surveys. Data on adaptation, heat-related morbidity, losses during disasters, (smallholder) agricultural yields, spread of disease, loss of wages and industry, and many, many more impacts of climate change are all potentially subject to the same loss of information in poor and rich communities. Health records cannot include the poorest if they do not have access to healthcare. Reinsurance records cannot reflect disaster losses of the poorest if their housing was not registered. When data with undercoverage of the vulnerable is used, we can expect that any result will be an underestimate of the truth. When data with nonresponse of the rich is used, we can expect that the inequality of outcomes will be an underestimate of the truth.
I will highlight two action points that I think are important given this context. First, when estimating the socioeconomic and health costs of climate change, one should always consider the quality and coverage of the data which was used. If the vulnerable may have been left out of that collection method, recognize that costs may be greater. Second, at a time when we cannot afford to use poor data, it is of utmost importance that data is collected using the knowledge of local stakeholders, and that the voices of the surveyed or observed are acknowledged.
To face the crisis ahead of us, we must improve the quality of information used in our science, and in our discussions. In the meantime, it may be the best we can do to at least acknowledge and adapt to its shortcomings.