Testing for equal treatment means
If there is no baseline treatment, it is unwieldy to analyse the data with confidence intervals for differences between pairs of treatments. Analysis should start with a single hypothesis test for whether all treatment means are equal.
If there are independent samples from treatments, we can test for differences between the treatment means with the analysis of variance test described in the previous section. For randomised block and repeated measures data, a different type of test should be used.
Grazing cattle in Uganda