Linear regression Flashcards
what equation do we get from the scatterplot? (in relation to linear regression)
the linear regression equation
y = B1 x1 + B0
B1 =
B0 =
slope
Y int
what values can B1 take
any
describe simple linear regression
looks at one x variable and one y variable
what 3 factors must be satisfied
linear XY relationship
normally distributed y
equality of variances (of y over x)
how to determine whether XY relationship is linear
look at scatterplot
how to determine whether y is normally distributed
look at normal probability plot - ends shouldn’t be skwerd
how to determine whether there’s equality of variance
do residual plot
what’s a residual
Y observed - Y expected (by linear regression equation)
null hyp?
alt hyp?
null: B1 = 0
alt: B1 =/= 0
how to evaluate
get t score
use pairs - 2 df’s and two tailed distribution to find p value
find 95%CI, where B1 is the coefficient
if the X value is binary, what are it’s values
X=1 for experimental/what’s specified
X = 0 for control/what’s not specificed
define multiple linear regression
looking at multiple x variables
in general, how do we make equation?
Y = X1B1 + X2B2 … + constant
forwards stepwise linear regression?
add in most significant variable first
then enter the rest
all the statistics are cumulative
if you were wanting to look at the contribution of one X variable to the Y, what would you look at?
r^2 DIFFERENCE
if you were wanting to look at conribution of all X variables to Y, what would you look at?
cumulative R^2 of all the variables
define: backwards stepwise linear regression
start with all, remove least significant variables first