Lecture 2 Flashcards
Regression I
Video 1
What is the general formula of all statistical models?
outcome = model + error
Video 1
Why does the regression model use the predictor?
To fit the linear association model.
Video 1
What is different about the linear regression model?
It uses two continuous variables.
Video 1
What is the sampling theory (theory of distribution)?
That when you have a lot of samples from a population, they will give different means. You will find the right mean when you look at the mean that is most common.
Video 1
What does the SE let you know?
How much the SD deviates from the mean, how much you expect to be wrong. Calculated by SD: squareroot of n
Video 1
When can you expect a normal distribution of a sample?
When the n is bigger than 30
Video 1
How is the df calculated?
number of observations - number of parameters estimated.
Video 2
What does the regression line look like in a formula?
y=a+bx+e (error).
a= y at x=0
b= slope
e = difference between the true and the predicted value
Video 2
When doe you use a small ‘i’ in the formula?
When the value used is predicted.
Video 2
What is the formula for the regression coefficient?
bylx = r*Sy/Sx
Video 2
When is a correlation symmetric?
When you standardize the predictor.
Video 2
What is used in regression?
A predictor and the outcome, is scalable
Video 2
Which three types of causality can you name?
Reverse causality, ommited variable, confounder (other factor)
Video 3
How can the variance be explained?
By R^2. It tells you how well the dependent variable is explained by the independent variable.
Video 3
What do you test during a hypothesis test?
Whether the regression coefficient is significant (with a t-test). And whether the model is a good fit for the data ( with F-test).
Video 3
Between which values does the R value range?
Between -1 and 1
Video 3
With which formula do you get the R^2?
The variation of y from the regression line/ total variation of y
Video 3
What is the use of an adjusted R^2?
It corrects the R^2 for the population instead of for the sample.
Video 3
What is the SSt?
The sum of squares, shows the total error from y
Video 3
What is the SSr?
The residual sum of squares, shows what is not explained by the regression line.
Video 3
What is the SSm?
The model sum of squares, it shows what is explained by the regression line. SSm = SSt-SSr
Video 3
In terms of SS…, what does the formula of R^2 look like?
SSm/SSt
Video 3
If the slope of the null hypothesis is at b=0 is it one sided or two sided?
Then it would be two sided
Video 3
If the slope of the null hypothesis is at b<=0 is it one sided or two sided?
It would be one sided then.
