12 Random effects

Question 1

Hva betyr random effects(/factors)?

Answer 1

Random effects categorical variable whose levels are viewed as a sample from some larger population. In a repetition of your experiment, you would have other levels of the random effect. You also want to be able to
Extrapolate your results beyond the levels of the random effect that you used
Y = fixed effects + random effects + error
Ofte biologiske individer: folk, fugler, griser…

Question 2

Gi eksempler på variabler som ofte er random effects.

Answer 2

Individer: griser, fugler, hester, planter

Question 3

Hva er fixed effect?

Answer 3

Fixed effect you are interested in the specific levels and have decided those beforehand. If the experiment was to be repeated, you would have to use the same levels of the fixed effect again. You cannot extrapolate to other levels.
Y = fixed effects + error.

Question 4

Hvordan skilles random effect og fixed effect?

Answer 4

○ Desired inference
Hva gjelder konklusjonen for?
○ Repetition
Hva må gjentas likt i neste eksperiment for å svare på samme spørsmål?

Question 5

Hvilke faktorer er relevante for analyse av random factors?

Answer 5

1 Components of variance
2 Formulae that link components of variance and expected mean squares
3 Nesting
4 Appropriate denominators for F-tests

Question 6

Hvordan påvirker random factor F-ratioen?

Answer 6

Det legges til en ny “error term” som må tas høyde for i noen F-ratioer. Praktisk betyr dette at F-ratioer for ulike fixed effects tester vil ha ulike degrees of freedom siden de må regnes ut på en annen måte.

Question 7

Når bør en random factor tas med i modellen?

Answer 7

Feks ved repeated measurements som gjør at datapoints within units ikke lenger er independent.

Question 8

Hva er formelen for F-ratio?

Answer 8

F-ratio=FMS/EMS

Question 9

Hva vil det si når F=1

Answer 9

Da er nullhypotesen sann og x-variabelen har ingen effekt på y-variabelen