T-Tests Flashcards
What are the 2 types of T-tests?
Unpaired / independent/ unrelated samples t-tests
Paired / dependent / related samples t-tests
Are t-tests parametric or non-parametric?
Parametric tests
used when data is interval or ratio
When receiving results from a t-test, what do you look at?
how far apart means are from each other
in the context of the variance of the data
Is there a significant difference if there’s a small mean difference but a large difference?
unlikely to have a significant difference
What is the t-test used for a between subjects design?
Unpaired / independent / unrelated samples t-test
comparing 2 different groups
Who created the independent samples t-test?
William Gosset
What is the independent samples t-test assumptions?
- data must be continuous (interval or ratio)
- normally distributed data
- data must have homogeneity of variance
How do we know if we have homogeneity of variance (HoV)?
using the Levene’s test
If the Levene’s test is non-significant (p-value above 0.05), what does that mean for HoV?
means we have homogeneity of variance
If Levene’s test is significant, what does that mean for HoV?
means homogeneity of variance is not found and violated
What is the Levene’s test?
tests the null hypothesis that the population variances are equal
increases our likelihood of finding a type 1 error (false positive)
What are effect sizes and what is one example of one?
effect sizes are how big the difference between the two groups mean are
example = Cohen’s d
How are effect sizes interpreted?
small = 0.2
medium = 0.5
large = 0.8
What are confidence intervals?
intervals which are expected to contain the parameter being estimated
give 95% CIs as a rule
What kind of estimate does a narrow confidence interval give?
precise estimate
What kind of estimate does a wide confidence interval give?
imprecise estimate
What is a Welch test?
an alternative to the independent samples t-test
does not assume homogeneity of variance
df in a welch test always has decimals
What t-test is used for a within subject design?
Paired / dependent / related samples t-test
used when comparing one group at 2 diff times or conditions
What is the dependent samples t-test assumptions?
- data must be continuous (interval /ratio)
- normally distributed