Multiple regression

Question 1

Multiple regression

Answer 1

• How well do the predictor values predict the outcome value

- Accounts for interrelations between multiple predictors

Question 2

Types of data multiple regression can be used for

Answer 2

• Independent variables: continuous or dichotomous (one or more)
• Dependent variables: continuous
• Dichotomous DV = chi squared

Question 3

understanding output

Answer 3

• Model summary: R= multiple correlation coefficient and R2 = coefficient of multiple determination
• Coefficients table: how much each predictor contributes to an outcome: intercept (b0), b1,b2… (slope)
• ANOVA table: how much the predictors as a set contribute (change in SS = variance explained)

Question 4

Choosing predictors

Answer 4

-Choose those most significant with sound theoretical basis

Question 5

Equation

Answer 5

• Y=b0+b1X1+b2X2

- outcome = model + error

Question 6

Forced entry

Answer 6

• all predictors entered together

- examines unique relationships between predictors and outcome

Question 7

Assumptions

Answer 7

• outcome and predictor variable must be interval (continuous) or nominal with 2 levels
• must be some variance
• large sample (power)
• linear relationship (scatterplot)
• normal distribution (histogra)
• homoscediacity (cloud)
• independence of errors (durbin watson - 1.5-2.5)
• multicollinearity: overlap between IVs (check collinearity statistic in coefficients table must be below .80 but above 0.2)

Question 8

Reporting multiple regression

Answer 8

• R2
• F statement (ANOVA)
• Standardized beta values, t-statistic and significance of each predictor
• cant graph multiple regression as it is 3D

Question 9

beta values

Answer 9

• standardized: original unit, used to make predictions using regression equation, one unit increase in X = Y…
• Standardized: can be compared across predictor values, one SD increase in X = y…