Chapter 14: Correlation and Linear Regression

-1-

.. FDR_L

-l/, {/ e_o-

c_is ,__'/ (55> ") (55y) _ -1 '.

__ ss > < 0-'

to

-

/ e, s'c_n"l -

OthC

.. sc.." S _ O,/

O/ Jł-ŁC

Łc

, / L

I.£ and O -'= __straw

-

0LVL_C.. OMC. '

L b_ox__'/ c Ł e c.__/,.. J.n/ c./ -.. _ 13_0c_, 13 2 x 2, o. FŁo, HL _., Ł -+

- > <

categorical variable

has separate, indivisible categories; no values can exist between 2 neighboring categories

continuous variable

have an infinite number of possible values between any 2 observed values

correlation coefficient

measures and describes the relationship b/w 2 continuous variables

characteristics of the correlation coefficient

direction (negative or positive), strength/consistency (varies from -1 to 1), form (linear is most common)

pearson correlation

measures the degree and the direction of the linear relationship b/w 2 variables

Sum of Products (SP)

measures amount of covariability b/w 2 variables

steps to compute pearson correlation

1. compute SP; 2. compute SSx and SSy

simple linear regression

uses data to draw a best fit line - creates a corresponding equation for prediction; only uses 2 measured variables (predictor variable = x and outcome variable = y)

multiple regression

adds more predictors to regression equation; see which predictor variable is the best (explains the most variance in the outcome variable); helps with the third variable problem