Lecture 9 - ANCOVAs

Question 1

What is an ANCOVA?

Answer 1

Combination between ANOVA and linear regression, that shows the effect of an IV on a DV while partialling out the effect of another variable (covariate), which also has an effect on the DV.

Question 2

What are the two clues that ANOVAs are a form of linear regression?

Answer 2

• Significance of a regression model is tested by ANOVA.

- ANOVA in SPSS is carried out by the General Linear Model.

Question 3

What is dummy coding?

Answer 3

It is the process of coding a categorical variable into dichotomous variables, generating a matrix showing which group an individual belongs to.

Question 4

The means in each group for the regression aspect of the output are calculated in terms of what?

Answer 4

They are based on the mean of the constant.

E.g. if the mean of the constant is 20.3, and the mean of the next condition is 60.3 on the ANOVA, the regression output would display it as 40.0.

Question 5

What do adjusted means allow?

Answer 5

Meaningful comparisons that take into account differences in a covariate.

Question 6

What are the benefits of using ANCOVAs? (rather than just using ANOVA or regression)

Answer 6

• Remove influence from confounding factors.
• Pre and post test
• Non equivalent/intact groups (naturally occurring differences might hide a confounding variable).

Question 7

How might we remove the influence of pre testing?

Answer 7

Treat the pre-test score as a co-variable, so that it can be partialled out.

Question 8

What is the problem with non-equivalent/intact groups?

Answer 8

There may be naturally occurring differences between groups (covariates).

Question 9

How can we account for problems caused by intact groups?

Answer 9

Make them equivalent by partialling out a covariate that may be interfering.

Question 10

An uncontrolled covariate is basically what?

Answer 10

A confounding variable.

Question 11

What are the assumptions of ANCOVAs?

Answer 11

• ANOVA assumptions
• Linear relationship between DV and covariate
• Homogeneity of regression –> parallel lines.
• Covariate must be reliably measured.

Question 12

What are the assumptions of ANOVAs?

Answer 12

• Interval/ratio data
• Scores normally distributed
• Homogeneity of variance

Question 13

What happens if the DV and the covariate to be partialled out are not correlated?

Answer 13

The ANCOVA cannot be carried out.

Question 14

What does it mean to say that the regression lines of groups to be compared are homogeneous?

Answer 14

It means that the regression lines of the groups are parallel. (which is an assumption of ANCOVAs)