Pseudoreplication and Experimental Units in Preclinical Ageing Studies
What counts as an experimental unit?
The experimental unit is the entity independently assigned to an intervention. The observational unit is what is measured; these need not coincide. When treatment is assigned to individual mice, tissue sections and cells sampled from each mouse are observations within that animal. When a diet is assigned to whole cages, the cage is the experimental unit for that comparison.
Pseudoreplication occurs when an analysis treats dependent observations as independent replicates. It can underestimate uncertainty and increase false-positive findings. The experimental unit depends on treatment allocation, rather than simply on whether a measurement comes from an animal, a cell or a culture well. ARRIVE: experimental units.
Illustrative ageing-study example
Suppose eight older mice are independently randomised, four per group, and researchers measure a tissue marker in ten microscope fields per mouse. There are 80 field measurements, but four experimental units per group. Treating the 40 fields in each group as independent mice would be pseudoreplication.
One approach is a prespecified animal-level summary; another is a suitable hierarchical analysis retaining the nested measurements. Neither creates additional independent animals. If instead one treated cage is compared with one control cage, additional mice cannot separate treatment from cage effects. These are hypothetical examples applying the principles discussed by Lazic and colleagues, not reported experimental findings.
Culture experiments and the scope of inference
Independently treated wells can be experimental units, while cells measured within each well are subsamples. However, wells from one donor do not establish reproducibility across donors. Describe donor, culture, treatment and measurement levels explicitly instead of relying only on labels such as “biological triplicate”. Statistical handling of nested data cannot repair missing replication at the level needed for the scientific claim. Lazic and colleagues.
What readers should look for
For each experiment, look for an explicit definition of the experimental unit and exact counts allocated to each group. Distinguish those starting counts from the numbers included in each analysis. Differences may reflect attrition or exclusions and require explanation. When cages or litters are the units, the total animal count should also be reported. These details help readers assess whether the stated sample size matches the design. ARRIVE: sample-size reporting.
A useful question for a figure legend is: does each plotted point represent an independently treated unit, a subsample or a repeated measurement? If this is unclear, the figure alone cannot establish independent replication.
Implications for longevity claims
These are methodological principles, not evidence that any intervention slows ageing. Keep laboratory findings, animal outcomes and human evidence separate. A culture experiment using human-derived cells remains laboratory evidence; it is not a clinical demonstration of improved human healthspan. Correct replication supports interpretation within the studied system, while broader claims require evidence appropriate to their scope.
Return to the research-literacy collection.