Many systems and processes in ecology cannot be experimentally controlled, either because the temporal and spatial scales are too broad, or because it would be unethical. Examples include large wildfires, alternative conservation strategies, removal of top predators, or the introduction of invasive species. Unfortunately, many of these phenomena also do not occur randomly in time or space, and this can lead to different biases (selection bias, unobserved variable bias) in statistical analyses. Economics has evolved largely without experiments, and developed statistical approaches to study “quasi-experiments”, i.e., situations were changes in time or space reveal relationships even in the absence of a controlled experiment. The goal of our paper was to compare and evaluate four quasi-experimental statistical approaches commonly used in economics, (1) matching, (2) regression discontinuity design, (3) difference-in-differences models, and (4) instrumental variables, in terms of their relevance for ecological research. We contrast the strengths and weaknesses of each approach and provide a detailed tutorial to demonstrate these approaches. We suggest that quasi-experimental methods offer great potential for investigating many phenomena and processes in ecological and coupled human-natural systems. Furthermore, quasi-experimental methods are common in environmental policy research and policy recommendations by ecologists may be more valuable when based on these methods.