Findings, implications and lessons learned

• This design could be improved further by taking into account the positions of the different surrounds within the poultry house and grouping them according to different environmental conditions apparent within the house, e.g. proximity to an extractor fan or to an exposed area. Let us suppose that three blocks can be defined to describe three areas within the poultry house, each containing four surrounds, each with one of the diets. This would lead to this randomised block analysis of variance:

Should blocking have a positive effect this would reduce the size of the residual variance:

• Finally there was a reduction in mortality from 8% in the control group to an average of 2% in the spent yeast groups. In view of the comparatively small numbers of chickens this was not statistically significant. Study question 5 illustrates how one can evaluate ahead of time the size of sample that might be needed to achieve statistical significance.