Stats terminology y2 1 Flashcards
What is multiple linear regression used for?
Removing IVs from a dataset with a large amount of IVs to make it easier to handle/measure
Benefits of MLR?
Easier than calculating the relationship between each IV and the DV separately
Can remove IVs that have little effect on the DV
Limitations or MLR?
Step by step so may take a while
First step of MLR?
Make a model that includes all the variables
R script for making a linear model with many IVs?
MODELNAME=lm(DV~IV1+IV2+IV3…,OBJECTNAME)
What is the estimate value that comes from a MLR?
The estimate of the gradient for each of the variables plotted against the dv
What is the std Error that comes from a MLR?
The error of the gradient
What is the t value of a MLR?
estimate of gradient/std error
What is the p value of an MLR?
Probability of making a type 1 error–> if you think that that variable has an effect it is the probability that you are wrong
Further steps of MLR?
Keep removing the IV with the highest p value until all of the remaining IVs are significant
What does a dot on the side of an IV mean?
The IV is marginally significant–>(P<0.1)
What does a star on the side of an IV mean?
the IV is significant
Three ways of testing if a MLR is good?
Removing each until all are significant, anova between successive models or AIC on the different models
deltaAIC <2?
models are similarly good
4<deltaAIC<7?
Considerable evidence that model with lower AIC is better
