Pseudo-Replication — When Logic Wears a Disguise
Pseudo-replication occurs when non-independent observations are treated as if they were statistically independent, artificially inflating sample size and deflating standard errors. This is common when multiple measurements are taken from the same individual, or multiple individuals come from the same cage or ecological plot. The result is vastly overconfident statistical tests.
Also known as: Hurlbert's pseudo-replication, Clustering error
How It Works
The additional measurements within units feel like additional data and statistically look like additional degrees of freedom. Researchers are often unaware that the independence assumption is violated by their data structure.
A Classic Example
A neuroscience study records spike activity from 50 neurons across 5 mice (10 neurons per mouse). If the analysis treats the 50 neurons as 50 independent observations, it commits pseudo-replication. Neurons from the same mouse are correlated. The true independent sample size is 5, not 50.
More Examples
An educational researcher tests a new teaching method in one classroom of 30 students and a traditional method in a second classroom of 30 students. Analyzing the 60 students as 60 independent observations ignores that students within the same classroom share a teacher, classroom environment, and group dynamics. The true sample size for comparing methods is two classrooms, not sixty students.
A food scientist tests whether a new preservative extends shelf life by placing 20 samples from the same loaf of bread in treated bags and 20 samples from the same loaf in control bags. Treating these as 40 independent observations commits pseudo-replication — all treated samples share the properties of one loaf, and all controls share the properties of another.
Where You See This in the Wild
Pseudo-replication is endemic in animal research (multiple readings per animal treated as independent), genomics (correlated features within genes), and ecological studies.
How to Spot and Counter It
Identify the true unit of randomization and replication. Use multilevel models or generalized estimating equations that account for clustering. Confirm that the reported n corresponds to independent experimental units.
The Takeaway
The Pseudo-Replication is one of those reasoning errors that sounds perfectly logical at first glance. That's what makes it dangerous — it wears the costume of valid reasoning while smuggling in a broken conclusion. The best defense? Slow down and ask: does this conclusion actually follow from these premises, or am I just connecting dots that happen to be near each other?
Next time someone presents you with an argument that "just makes sense," check the structure. The feeling of logic is not the same as logic itself.