|
Study questions
- Rerun GenStat for 60-day milk yield for the first two batches and calculate the coefficient of variation (100 x standard deviation / mean). Comment on the value you get. What would you tell the researcher? If the researcher were to do another experiment what steps would you advise him/her to take in order to try and reduce the variation?
- Following analysis of the data from the first two batches, your advice to the researcher is that there seems to be little value in proceeding with the experiment and that it should be aborted. He ignores your advice, completes the experiment and brings the data back to you for analysis. Discuss your role and the consequences of agreeing or refusing to do the analysis.
- When reporting the results of a study when things do not always go quite to plan it is usually possible to present the study in the way that disguises what really happened without completely hiding the truth. Write a description of the experimental design assuming that Option 1, as carried out, was followed correctly. You will need to pretend that the treatment A + B applied in the first two batches was actually not done.
- A 2-week pre-treatment period was allowed in this experiment to allow pre-treatment milk yield to be determined for each goat. This was then used as a covariate in the analysis. This, however, required a 15-day control period at the start of lactation. Design an alternative experiment in which the experimental treatments could be applied as the goats kidded. What additional information on the goats would you require for an efficient design?
|
|
