Linear Regression

Question 1

What is the main purpose of linear regression?

Answer 1

Causal associations.

Focuses on a mean group comparison.

Question 2

What are the requirements for linear regression?

Answer 2

Outcome variable must be continuous.
Predictor variable can be anything.
Outcome should be approximately normally distributed.
Should be testing the linear relationship between two variables.

Question 3

What does the ordinary least squares regression minimise?

Answer 3

The sum of the squared residuals.

Question 4

What is Bo in the equation?

Answer 4

The y-intercept.

Question 5

What is Ei in the equation?

Answer 5

The residual error (individual deviation).

Question 6

What is B1 in the equation?

Answer 6

The gradient.

Question 7

What method should you use when conducting a LR in SPSS?

Answer 7

Forced-entry/enter.

Question 8

Which assumptions must you check?

Answer 8

Normally distributed errors.
Homoscedasticity.
Linearity.
Influential cases.

Question 9

What is R-squared?

Answer 9

It is the amount of variance of the outcome variable which can be explained by the predictor.

Question 10

What is the F-value?

Answer 10

There ratio of the model MS divided by the residual MS.

Question 11

What is the cut-off point for the F-value?

Answer 11

Has to be larger than 1.

Question 12

How do you check for normal distributed errors?

Answer 12

Histogram + P-P plot.

Question 13

How do you check for homoscedasticity?

Answer 13

zResidual vs zPredicted plot.

Variance around the regression line should be the same for all values of the predictor variable.

Question 14

How do you check for linearity?

Answer 14

zResidual vs zPredicted plot.
Must add Loess line!
Line should be approximate to zero and linear.

Question 15

How do you check for influential cases?

Answer 15

Cook’s distance.

Question 16

What is Cook’s cut-off point?

Answer 16

Must be smaller than 1.

Question 17

What is an influential case?

Answer 17

A case that significantly alters the value of a regression coefficient when it is deleted from the analysis.

Question 18

In the SPSS output table, what is the intercept?

Answer 18

The unstandardised B (constant) value.

Question 19

In the SPSS output table, what is the gradient?

Answer 19

The unstandardised B (predictor) value.

Question 20

How do you report unstandardised results?

Answer 20

Results show that (predictor) is …. associated with (outcome) (B = … , p = …)) which supports our hypothesis.

Question 21

How do you report standardised results?

Answer 21

Results show that one unit higher in (predictor) is significantly associated with an … of … (outcome) (p = …) which supports our hypothesis.

Question 22

Which number should not be between the LCI and UCI?

Answer 22

0.

This means results could be non-significant as the regression weight in the population could be 0.

Question 23

Name one issue with linear regression?

Answer 23

Reverse causality.

Question 24

What can you do if the outcome variable is not approximately normally distributed?

Answer 24

Bootstrap.

Question 25

When is the F-value not useful?

Answer 25

When the predictor is non-significant.