学术文献库

A protocol deployed in a scientific study is designed to bring results to a state from which inductive inference is allowed to be made. In the nineteenth century triplication was introduced into the plant sciences, and Fisher's p<0.05 rule (1925) was incorporated to give a triple-result protocol designed to counter random and systematic errors. The aims here were: to define replication protocols (including a non-replicated protocol); assess prevalences for the plant sciences in one year (2017) in the twenty-first century for one particular science type; and to elucidate the theoretical basis for triplication. Methods involved: survey of the entire field of plant sciences to produce a protocol-prevalence report; an assessment of association versus experimental studies; and use of real-world-data proxies to show confidence-interval-width patterns with increasing replicate number. Results showed that triplication was extensively applied (in ~70% of studies) in the plant sciences, with triple-result protocols in ~15% of studies. Theoretical considerations showed that, even if systematic errors predominate, square-root rules are likely to apply to many datasets, and real-world-data proxy examples are given. Conclusions are: there are strong theoretical reasons why (1) triplication (rather than duplication or quadruplication) is the standard; and (2) a triple-result (rather than global) protocol is optimal; additionally (3) that standards in the plant sciences remain at high levels despite the immense pressures for publication.