Session 7b - Comparison of means, non-independent

Question 1

Examples of samples being not-independent

Answer 1

Measuring a variable before and after an event in the same subject
Matched pairs eg parent / child

Question 2

Implications of variables not being independent

Answer 2

Observations are correlated - no longer independent
No between-subject differences, therefore smaller standard error of group differences
Larger observed t-values, because smaller standard error

Question 3

Degrees of freedom for t-distribution for non independent variables

Answer 3

Only one group, not two (n-1)

Question 4

Assumptions for paired t-test

Answer 4

Pairs of observations are independent
PAIRED DIFFERENCES are normally distributed (check on histogram)