Experimental Design and Statistics in Preclinical Research...
... THE GOOD, THE BAD AND THE UGLY
A one-day course addressing some of the fundamental issues that are behind designing good (preclinical) experiments, the bedrock of reproducible science.
WHAT: The course consists of a series of interactive lectures and scenarios with Manuel Berdoy covering the following: the variables that determine significance, focussing particularly on power calculation, pseudo-replication, the influence of variation and how to minimise it (e.g. randomised block designs) and what the numbers in an ANOVA actually mean (see learning outcomes for more details). It will also address statistical issues relevant to animal research depending on the composition of the audience.
WHY: Arguably, all good experiments start with a good experimental design. Experimental design in the Biomedical Sciences is mostly about logic, common sense and the systematic application of relatively simple techniques to produce un-biased experimental results and reduce variation. This is good news because this should be relatively straightforward. Yet this is where the biggest blunders continue to be made, including by experienced biologists, with demonstrably drastic consequences on results.
This course aims to address these issues, their links with statistical analyses generally and some of the traps that we have all fallen into, all in a day session where the words “enjoyment” and “statistics” can hopefully belong in the same sentence.
FOR WHOM: The course is open to all students in the Medical Science Division as part of Skills Training, and all new Graduate students are encouraged to attend. This is particularly relevant to those before they embark on experiments. Students and post-docs who are further along in their studies have reported to find it useful, and are also welcome to attend. If you are not sure whether it is relevant to you feel free to contact courses@medsci.ox.ac.uk
WHEN: The course will be repeated on each term on a first-come first- serve basis (see register).