16. Summary

1. State clearly the purpose of the study

2. Explain why you have chosen a particular animal model and why the experiments could not be done using non-animal alternatives

3. Think about the 3Rs in relation to your experiments

4. Explicitly Identify your experimental unit

5. Explain how you decided sample size (power analysis, resource equation or fixed by availability)

6. Explain how the experimental units were randomised to the treatment groups

7. Use coded samples where possible to blind yourself (and others) to which treatment group a subject belongs

8. Think about ways of reducing the variability to increase power e.g. optimum/non-stressful housing, freedom from disease

  • Use inbred strains of rodents or explicitly justify not using them

9. Choose a suitable experimental design (completely randomised, randomised block, Latin square,  etc)

10. Consider using a factorial design to explore generality of your results

11. Decide how you are going to do the statistical analysis before starting the experiment, recognising that methods may need to be modified when the results are obtained.

12. Choose a good statistical package and learn how to use it.

  • Make use of graphical methods, particularly those showing individual points, to screen and display your results
  • Consider quoting/displaying the results in standard deviation units (this will also help those doing a meta-analysis)

13. Learn some statistics (buy a good statistics textbook, take a course on statistics)

14. Learn to use the analysis of variance for these common designs:

  • One way (completely randomised design)
  • Two-way without interaction (randomised block design)
  • Two-or-more-way with interaction (factorial design)

15. Be very honest about deleting outliers, but try analysis with/without to see if they make any difference

16. Use the ARRIVE or GSP guidelines to ensure that you have not missed anything when writing your paper/thesis