STATS 4 MULTIPLE REGRESSION

Question 1

What statistics are used for describing the overall regression model?

Answer 1

Variance of Y observed (variance of Y predicted + error)

Question 2

What statistics are used for describing the stat. significance of the overall model in MR?

Answer 2

F-test

Question 3

What statistics are used for describing the stat. significance of the individual predictors in MR?

Answer 3

T-test

Question 4

What are the assumptions in regression?

Answer 4

Normality, linearity, homoscedasticity, independent sampling

Question 5

Where do we look to test the assumptions in regression?

Answer 5

The residuals in the model, not the IVs or DVs?

Question 6

How do we test assumptions using SPSS?

Answer 6

Most assumptions can be checked using the “plots” menu

Question 7

How do we check the assumption of linearity?

Answer 7

Check the residual scatterplot, there should be no non-linear patterns

Question 8

How do we check the assumption of independence?

Answer 8

Use the Durbin-Watson statistic. Value = 2 indicates independence. Below 1 or above 3 is problematic

Question 9

How do we check the normality assumption?

Answer 9

Using the Histogram and Normal Probability plot. Check histogram for bell-shaped curve. Check Normal Probability plot for deviation of plots from the trend/line

Question 10

How do we check the homoscedasticity assumption?

Answer 10

Using the residual scatterplot, the variability of the residuals should be the same for all values of Ypred (there should be no funnelling of residuals)

Question 11

How do we measure the distance of an outlier?

Answer 11

The distance between Yobs and Ypred

Question 12

When would we say an outlier has leverage?

Answer 12

When it has an unusual value on predictor

Question 13

How do you tell if an outlier has influential potential?

Answer 13

Standardised residuals or predictors in excess of +/- 3.29 (p < .001)