6533b828fe1ef96bd1288cc9
RESEARCH PRODUCT
Individual measurements and nested designs in aquaculture experiments: a simulation study
Kari Ruohonensubject
Set (abstract data type)Power analysisAlternative hypothesisStatisticsStatistical modelReplicateAquatic ScienceBiologyStatistical powerType I and type II errorsNested set modeldescription
Simple and nested models for analysis of variance (ANOVA) in aquaculture experiments were compared with the help of computer simulations. Simple models for analysing variables that are based on tank means, such as final weight and growth rate, were found to be sensitive to differences in the number of individual observations in each tank. In comparison to nested models that take into account individual measurements, the simple models were found to overestimate the F ratio and increase the risk of committing type I error, i.e., accepting a false alternative hypothesis. Further, nested models permit greater flexibility in experimental design, and allow more economical solutions within a given set of constraints and required statistical power. It is concluded that individual measurements do not remove the necessity of having replicate tanks but maintaining both levels of variation (tank and individual) in the statistical analysis could reduce the number of replicates needed.
| year | journal | country | edition | language |
|---|---|---|---|---|
| 1998-06-01 | Aquaculture |