L7 - Differences: One Way Between-Subjects Designs for 3+ Groups

Question 1

What is a balanced design?

Answer 1

When every group has the same number of subjects.

Question 2

what is dummy coding?

Answer 2

When we use values of 1 and 0 for two-category variables in a linear regression context.

can still due this for 3+ category variables, but it needs to be done in a more systematic/meaningful way, for results to be interpretable.

Question 3

What happens if we arbitrarily code categories??? (3+ category variables)

Answer 3

Observed R squared value, and sample regression coefficients will depend on the ordering of the arbitrary coding.

So it’s basically meaningless.

Question 4

What should we do with categorical variables of 3+ variables?

Answer 4

Create two new dichotomous dummy coded variables to represent the three categories.

The coding will have values of 0 or 1

Question 5

What is referred to as the reference category?

Answer 5

This is the category that has 0 across all dummy variables.

The intercept of the regression line will show the mean of people in this group!!!

Question 6

Where can we derive the group means from in regression?

Answer 6

In the coefficients table, we can get the means.

The unstandardised coefficients (B), represents the difference between mean of the particular category and the reference category.

The intercept is the mean for the reference category

Question 7

How many dummy coded dichotomous variables?

Answer 7

number of categories (k) -1

k-1

Question 8

What is the formula to find unique pairings?

Answer 8

(k(k-1))/2

Question 9

How many unique differences can be derived from an 6 unique pairings?

Answer 9

3.
(k would have equalled 4 to have 6 unique pairings)

k-1

Question 10

What is the aim of ANOVA?

Answer 10

to investigate differences in means between two or more groups.

Question 11

What is total variation among all people irrespective of their group made up of?

Answer 11

• variation between groups (SSbetween)

- variation within groups (SSwithin)

Question 12

What is an IV referred to as in ANOVA?

Answer 12

a factor

eg. diagnosis

Question 13

What are levels of a factor?

Answer 13

groups within a factor

eg. diagnosis –> sz, sa, bp

Question 14

What is a between-subjects factor?

Answer 14

This is a factor in which person’s score on the DV is located in only ONE LEVEL of the factor.

Question 15

What are the 6 ways we can investigate mean differences among 3 + groups?

Answer 15

• FOCUSED PLANNED COMPARISONS/planned contrasts
• unfocused omnibus test
• post hoc testing
• omnibus f test, followed by planned comparisons
• omnibus f test, followed by post hoc testing

Question 16

What is an omnibus F test?

Answer 16

This investigates differences in 3 or more levels of a factor, by assuming a null hypothesis that all group means are equal at the population level.

Alternate hyp: that at least 1 group differs.

It is identified by the numerator degrees of freedom

Question 17

What is the test statistic equation for omnibus F test?

Answer 17

Tobs = MSbetween/MSwithin

where df1 = k-1

and df2 = N-k

Question 18

What is the limitation of the omnibus f test?

Answer 18

It’s really vague, because it doesn’t tell us where the statistically significant difference(S) lie!!!!!

Question 19

What must contrast weights add up to?

Answer 19

0

Question 20

What is the ‘value of contrast’ (Mcontrast)?

Answer 20

It is a value given in the contrast tests table..

It’s calculated by multiplying the contrast weight with its corresponding mean value, across all categories.