PSYC1040 Week 12

Question 1

Confidence intervals

Answer 1

• the critical t-values tells us how many standard errors apart two values need to be statistically significantly different
• therefore, to find our ‘confidence interval’ we just need to find the raw values that are that far away on either side of our obtained value

Question 2

Effect size and statistical power

Answer 2

• d (a standardised mean difference) indicates the strength of association between a two-level IV and a continuous DV
  ~ how precisely you could guess the DV’s state if you knew the IV’s state
  ~ average probability of correctly guessing the IV’s state if you know the DV’s state
  ~ equally, d indicates overlap of the raw distributions
• if N is also known, overlap of the SAMPLING DISTRIBUTIONS can be calculated
• then, for a given A, the probability of an observed mean difference being statistically significant can be calculated
• effect size, SS, alpha and power are mathematically related
  ~ if you know three of these, you can calculate the fourth

Question 3

Statistically assumptions of t-tests: assumptions of independent group t-tests

Answer 3

• non-parametric tests make fewer assumptions
• normality of sampling distributions
• homogeneity of variance between conditions
• independence of observations (separate people in each group)

Question 4

Statistically assumptions of t-tests: assumptions of repeated measures t-test

Answer 4

• non-parametric tests make fewer assumptions
• normality of sampling distributions
• homogeneity of variance between conditions
• difference scores come from a justified pairing of raw scores (two scores per participant)