Regression Flashcards
What is correlation?
A measure of the relationship or covariance between variables - as one changes, what happens to another?
The analysis (including standardisation) of covariance
Not a measure of causation
What is covariance?
The underlying raw relationship
What are the two types of correlation?
Pearson’s - parametric
Spearman’s - non-parametric
What is regression?
Asks how does variable x predict variable y
Fits a “line of best fit” to the data using a linear equation (GLM)
Tests this linear model to predict values of an outcome variable from a predictor variable
What is simple regression?
When there is one predictor
What is a predictor variable?
Independent variable
What is an outcome/criterion variable?
Dependent variable
What is the overall “model fit”?
R/R^2
What does beta represent?
How steep the line is
What are the strength of predictors shown by beta values?
Positive beta = positive predictor
Negative beta = negative predictor
Similar covariance/correlations
What does R represent?
How close each point is to the line
What is multiple regression?
May be other variables that contribute
If want to make a good prediction, need to quantify how much these variables influence height and to what degree relative to each other
What are the key outputs of regression?
p-value
r^2
Unstandardised beta
Standardised beta (b)
What is the p-value?
Overall model fit
Is using your model better than using the observed outcome variable mean?
Do you have a significant model?
What is r^2?
Goodness of fit
How much of your outcome variable variance is explained by your predictors
What is the unstandardised beta?
Single regression
What’s the direction and strength of the relationship
What is the standardised beta (b)?
Multiple regression
If multiple predictors, how do they compare
Z-scores
What uses a line of best fit?
Simple regression
What uses a plane of best fit?
Multiple regression
