PSYC1040 Week 10 + 11

Question 1

Effect size catch up (Cohen’s d)

Answer 1

• when comparing two means, a simple measure of ‘effect size’ is to state the difference between means in SD’s.

Question 2

d versus t (for a single sample or repeated measures situation)

Answer 2

• d is a difference expressed in SD’s while t is a difference expressed in SE’s
• therefore, d won’t be affected by SS but t will be larger in larger samples
• d can be extreme even in a small sample, but t might still be non-significant
• you can obtain a very large mean difference that is not statistically reliable
• any difference will be statistically significant if the SS is large enough

Question 3

The independent groups t-test: logic, conditions, steps and example: in a repeated measures study

Answer 3

• each participant is measured in two situations, so has two raw scores
• do the difference scores have a mean further from 0 than we would usually expect if we drew a sample from a population of different scores with m=0?
• we run a repeated measures t-test (a single sample test on difference scores)

Question 4

The independent groups t-test: logic, conditions, steps and example: in an independent groups study

Answer 4

• some participants are measured in one situation and other participants are measured in another situation
• is the difference between group means larger than we would usually expect if we just drew two samples from the same population?
• we run an independent groups t-test
• we want to find out how likely a given difference between two sample means would be, if both samples were drawn from the same population
• if two samples are drawn randomly from the same population, the two sample means will usually be different from each other
• to test the Null Hypothesis, we need an idea of how often a difference between means as large as the one we observed would occur just by chance

Question 5

confidence intervals

Answer 5

• the critical t-value tells us how many SE’s apart two values need to be statistically significant 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