Small values in big data: The continuing need for appropriate metadata

摘要: Abstract Compiling data from disparate sources to address pressing ecological issues is increasingly common. Many datasets contain left-censored – observations below an analytical detection limit. Studies single and typically small show that common approaches for handling censored — e.g., deletion or substituting fixed values result in systematic biases. However, no studies have explored the degree which documentation presence of influence outcomes large, multi-sourced datasets. We describe a lake water quality database assembled 74 illustrate challenges dealing with big data, including limits are absent, range widely, trends over time. substitutions can also bias analyses using ‘big data’ datasets, be effectively handled modern quantitative approaches, but such rely on accurate metadata treatment each source.