multiple regression lec 9 Flashcards
what does simple regression allow us to do
predict a score on a variable y (outcome/DV) froma. known variable on X (predictor)
least squares is
line of best fit used to measure different between line and observed data
how can you predict the regression
y = a + bX
a value of intercept
b = the slope (gradient)
x = value we know
how can we measure the degree of fit
R2 with 0 being terrible fit and 1 being perfect fit
what are the limitations of simple regression
can only use one variable to predict outcome, eg wiehgt will be god predictor of height but far from perfect.
what is multiple linear regression
multiple predictors eg weight and sex and age predict heighth
higher r2 means
more accurate predictions
if you get a r2 value of 0.759, what dos this mean
model accounts for 7.50% of variance in exa performance
what does unstandardized beta tell us
how each of the individual predictors effect the study results
what are standardised beta coefficients
tells us the strength of the relationsip between the predictor and DV, meaning we can compare them differently
how to write it up
regression model used to..
r2 value
b for all
when can we use multiple regression
the continuous outcome variable (eg weight)
linear relationships
non zero variance, eg don’t use sex if just males
multicollinearity (pred variables should not correlate too much) eg lecture attendance and revision time
outliers should be identified
sample size (larger to not be affected by outliers or false negatives)
