Experimental design/Animal Models Flashcards
What to consider when choosing an experimental design:
1) animal model
2) sample size
3) hypothesis (one or two way)
What is the classical parallel design?
two (or more) groups of animals getting each their own treatment (or act as a control)
How do you allocate animals into groups?
1) randomization
2) stratification (use of certain criteria to one or the other group, eg weight)
Different ways to collect measurements
1) single measures (one organism per timepoint, more mice but might be less stressful) (parallel)
2) repeated measures (fewer mice, can compaired with its own values)
3) Cross over design (being it´s own control) usually when animals are expensive. Better to do two way where all combinations are tested. A-B, B-A eg.
4) latin-squares = cross over put in square form.
5) multifactorial design - for every subcatagory there is a group, elimination of factors.
Mathematical calculation of sample size =
S = 2 x (significance level + power level)^2 / (((mean1-mean2)/mean1)/(SD/mean1))^2
Type one error:
there is difference but there really isnt
Degrees of freedom:
(observations-1)-(groups-1)-(blocks-1)
Meads equation:
Df need to be between 10-20.
What can you do about a lack of power?
1) use more animals
2) use animals with a smaller variance
3) have a bigger difference in means.
What is a animal model?
1) an animal where behaviour/pathological mechanism/phenomenom can be studied and is apllicable to humans.
High fidelity model:
high general similarity to what is being modeled
High discrimination model:
high specific similarity to what is being modeled
Classification of animal models:
1) spontanous = phenomenon occurs spontanously
2) induced = the phenomenon occurs after induction chemically, biologically or genetically
3) negative = penomenon never occurs
4) orphan = veterinary patients with human disease analog
