Equivalence testing and Bayes factors are informative statistical methods for analyzing replication studies of original studies with null results, and can address the limitations of the commonly used non-significance criterion.

What can authors and reviewers do to keep common statistical mistakes out of the literature?

A project to repeat experiments from high-impact papers in cancer biology did not complete all replications, with challenges ranging from mundane reasons to unexpected methodological issues.

In mouse models of Rett Syndrome, global loss of the Mecp2 gene induces microglia to engulf excess synapses, while microglia-specific loss or gain of Mecp2 has little impact on disease.

Microbial macroecological patterns under coarse-graining are captured by a null model of ecology, a result which was leveraged to successfully predict the relationship between fine and coarse-grained diversity.

Bayesian estimation of the proportion of the population that would show an effect in an experiment provides a broadly applicable alternative to standard null hypothesis tests on the population mean with several advantages.

Although questionable research practices inflate false positive rates, they have little effect on replicability because they also increase power.

Genetic variants introgressed into modern humans from Neanderthals tend to be depleted in their contribution to heritable trait variation relative to modern human variants consistent with the action of purifying selection.

Continuous noisy learning in neural networks leads to slow, directed qualitative changes in their internal representations, without affecting their quality.

Herpes simplex ICP4 preferentially binds to and delineates the viral genome, ultimately resulting in robust viral transcription at the expense of the host.