11. Comparing Means Flashcards
What is the T-test used for?
to assess whether two group means are different
IVs and DVs in the t-test
only 1 IV
only 1 DV
t-test statistic equation
t-test = (model ÷ error)
two types of t-test
- independent means t-test
- dependent means t-test
When do you use an independent means t-test?
- 2 experimental conditions
- different participants assigned to each condition
What is the independent means t-test also called?
- independent samples
- unpaired
What must you acknowledge with an independent means t-test?
unsystematic variance
How do you use the independent means t-test score?
- compare value of t against the max value that would be expected to have by chance alone in a t-distribution with same df
What happens if the t-stat is more than the critical value in an independent means t-test?
results are due to the effect of the IV and not by chance alone
When do you use the dependent means t-test?
same group of people are exposed to different treatments or are assessed before and after an intervention
dependent means t-test and experimental conditions
- 2 experimental conditions
- same participants take part in both
dependent means t-test is also called
- dependent samples
- paired t-test
How do you use the dependent means t-test score
- calculate t-test stat and compare against the max value that would be expected to have occurred by chance alone in a t-distribution with the same df
What is the critical value?
max value that would be expected to have occurred by chance alone
If the t-stat is more than the critical value, results are due to
the effect of the IV and not by chance alone
Assumptions for independent t-tests
- normally distributed
- interval data
- homogeneity of variance
- independence
Assumptions for dependent t-tests
- normally distributed
- interval data
degrees of freedom
- number of observations that are free to vary
- 1 less than the sample size
What are your 3 options if the data isn’t normally distributed?
- use a larger sample
- transform the distribution
- use non-parametric tests