The benefits and pitfalls of using satellite data for causal inference
Meha Jain
The benefits and pitfalls of using satellite data for causal inference - USA : Oxford University Press, 2020.
Peer review
There has been growing interest in using satellite data in environmental economics research. This is because satellite data are available for any region across the globe, provide frequent data over time, are becoming available at lower cost, and are becoming easier to process. While satellite data have the potential to be a powerful resource, these data have their own sources of biases and error, which could lead to biased inference, even if analyses are otherwise well-identified. This article discusses the potential benefits and pitfalls of using satellite data for causal inference, focusing on the more technical aspects of using satellite data. In particular, I discuss why it is critical for researchers to understand the error distribution of a given satellite data product and how these errors may result in biased inference. I provide examples of some common types of error, including nonrandom misclassification, saturation effects, atmospheric effects, and cloud cover. If researchers recognize and account for these potential errors and biases, satellite data can be a powerful resource, allowing for large-scale analyses that would otherwise not be possible.
Text in English
1750-6816 1750-6824 (Online)
https://doi.org/10.1093/reep/rez023
Satellite observation
Data
Research
Environmental Economics
The benefits and pitfalls of using satellite data for causal inference - USA : Oxford University Press, 2020.
Peer review
There has been growing interest in using satellite data in environmental economics research. This is because satellite data are available for any region across the globe, provide frequent data over time, are becoming available at lower cost, and are becoming easier to process. While satellite data have the potential to be a powerful resource, these data have their own sources of biases and error, which could lead to biased inference, even if analyses are otherwise well-identified. This article discusses the potential benefits and pitfalls of using satellite data for causal inference, focusing on the more technical aspects of using satellite data. In particular, I discuss why it is critical for researchers to understand the error distribution of a given satellite data product and how these errors may result in biased inference. I provide examples of some common types of error, including nonrandom misclassification, saturation effects, atmospheric effects, and cloud cover. If researchers recognize and account for these potential errors and biases, satellite data can be a powerful resource, allowing for large-scale analyses that would otherwise not be possible.
Text in English
1750-6816 1750-6824 (Online)
https://doi.org/10.1093/reep/rez023
Satellite observation
Data
Research
Environmental Economics