Regression

Question 1

What is regression

Answer 1

How does one thing affect the other?

Question 2

Linear model of regression

Answer 2

Yi = bo + b1X + error

Question 3

What method is used to find the line that best fist the data?

Answer 3

Method of least squares (smallest amount of residuals)

Question 4

What is bi?

Answer 4

Gradient, direction, strength (B on SPSS)

Question 5

what is bo?

Answer 5

Intercept, value of y when x = 0 (constant unstandardised B on SPSS)

Question 6

If model is the best fit how should SSm relate to SSr

Answer 6

SSm should be greater than SSr

Question 7

What does SSt mean?

Answer 7

Variability between scores and the mean

Question 8

What does SSr mean?

Answer 8

Variability between model and actual scores

Question 9

What does SSm mean?

Answer 9

SSt - SSr

Question 10

Regression calculate r2 (3 ways)

Answer 10

SSm / SSt OR (SSt - SSr) / SSt OR beta on SPSS

Question 11

What does r2 mean?

Answer 11

How well model generalises to the population

Question 12

F ratio

Answer 12

MSm / MSr

Question 13

Calculating mean from SSt/SSr/SSm

Answer 13

Divide by df

Question 14

What does F mean?

Answer 14

Whether it was done by chance

Question 15

Assumptions for regression

Answer 15

Continuous outcome, predictors must not have 0 variance, independence, linearity, homoscendasticity (variance of error term should be constant, no shape on graph), independent erros and normally distributed errors

Question 16

Regression output

Answer 16

SSm = sum of squares of regression
SSr = sum of squares of residual
SSt = sum of squares total
MSm = Mean square regression
MSr = Mean sqaure residual
Report as F (dfregression, dferror) = XX , p