Recent years have seen an increase in the use of secondary data in climate adaptation research. While these valuable datasets have proven to be powerful tools for studying the relationships between people and their environment, they also introduce unique oversights and forms of invisibility, which have the potential to become endemic in the climate adaptation literature. This is especially dangerous as it has the potential to introduce a double exposure where the individuals and groups most likely to be invisible to climate adaptation research using secondary datasets are also the most vulnerable to climate change. Building on significant literature on invisibility in survey data focused on hard-to-reach and under-sampled populations, we expand the idea of invisibility to all stages of the research process. We argue that invisibility goes beyond a need for more data. The production of invisibility is an active process in which vulnerable individuals and their experiences are made invisible during distinct phases of the research process and constitutes an injustice. We draw on examples from the specific subfield of environmental change and migration to show how projects using secondary data can produce novel forms of invisibility at each step of the project conception, design, and execution. In doing so, we hope to provide a framework for writing people, groups, and communities back into projects that use secondary data and help researchers and policymakers incorporate individuals into more equitable climate planning scenarios that “leave no one behind.”

Relating social inequality and vulnerability to environmental hazards is an especially challenging task in regions with a paucity of data. Researchers attempting to measure the potential environmental and human impacts of past and continuing industrial toxicity in Azerbaijan have often either questioned the reliability of environmental indicators disclosed by the state’s official statistics or found the government’s environmental and population data partial and incomplete. To contribute to a clearer description of the human impacts of toxic waste locations and to assist other researchers, we use a novel methodology. By overlaying data from Azerbaijan’s Toxic Site Identification Program (TSIP) onto national census population data – augmented with in-country interviews – we can map the inequitable distribution of infant mortality, unemployment, and toxic waste sites to better suggest some of the places and people in particular need of environmental mitigation and health, and economic intervention. This method is transferable to future research in the Caucasus, Eurasia, and other data- poor areas.