ABSTRACT
Pseudoreplication is a widely discussed topic in the scientific community. Its principal critique concerns the lack of independence in flawed experimental designs and the use of inferential statistics to test the hypothesis of such experiments. Thirty years after its appearance, it remains misunderstood by many researchers, including ecotoxicologists. In the present study, we try to clarify some of its concepts by filling in what seems to be a gap in the terminology of manipulative experiments. We propose the term "experimental medium" to refer strictly to the relevant spatial scale of the experiment to preserve the specificity of the experimental and observational units and to display pseudoreplication as a kind of misinterpretation and/or misanalysis of inferential statistics. A classification of the types of experimental designs in ecotoxicology is offered, and the problems in using inferential statistics in suboptimal designs are discussed. We hope to shed some light on such a classic topic for ecotoxicologists. Environ Toxicol Chem 2017;36:2868-2874. © 2017 SETAC.
