Equality of Means (t-test) Flashcards
What are the three t-tests?
▪️One sample
▪️Independent samples
▪️Paired sampled
When would you used a one sample t-test?
To test is the mean of the population differs from a pre-specified test value?
E.g. difference between average height of your class and known average height of children
When would you used an independent t-test?
To study the difference in means between two groups
E.g. Healthy OA vs MCI
When might you used a paired sames t-test?
To study the difference in means of the same group.
E.g. before and after treatment, matched case controls
For a t-test, the variable whose mean is tested needs to be _____________________
Fairly symmetrical (bell shaped)
What can you use to check whether your data is symmetrical/normal enough?
A normality test (e.g. Kolmogorov-Smirnov test)
Non-parametric –> Legacy dialogs –> K-S
What is the null hypothesis of a normality test?
There is no difference between your data and normality
What is the main problem with normality tests?
They tend to be very conservative, especially when the sample size is large
The ____________ the sample size, the more likely the K-S test is to tell you the data is not symmetrical.
Larger
What test would you used to test whether the population mean is equal to a certain value?
One sample t-test
What assumptions are made for the one sample t-test?
▪️The observations are randomly and independently drawn
▪️Symmetrical, bell-shaped data (normal)
▪️There are no outliers
What test would you use to see whether the mean of group A is equal to the mean group B?
Independent samples t-test
What is the df of a one sample t-test?
n-1
What is the df for an independent samples t-test?
nA + nB - 2
What assumptions are made for independent samples t-test?
▪️The observations are randomly and independently drawn
▪️Symmetrical within each group
▪️No outliers within each group
What must you do before conducting an independent samples t-test?
Check the suitability of the data for each group (how normal they are)
Split the file
After splitting a file to assess the normality of eacih group, remember to….
Merge them again!
Split file –> Analyse all cases
How do we decide whether to use the equal variances assumed or not assumed value?
Levene’s test for the quality of variances
Is the variability the same?
If the p-value for Levene’s test is significant, which row do you interpret?
The one where equal variance is NOT assumed
(the variability is statistically significantly different)
What test would you use to see if the mean of a group in one condition is equal to the mean of the same (or paired) group in another condition?
Paired samples t-test
What assumptions are needed for the paired samples t-test?
▪️The paired observations are randomly and independently drawn (only occur once)
▪️The paired differences variable is symmetrical and continuous
▪️There are no outliers in the difference
What must you do before conducting a paired samples t-test?
Check how symmetrical the difference between the two are.
For this, you have to compute the variable (e.g. weight1-weight2)
For what type of t-test do you need to check the variability (equality) of the variance?
Independent samples t-test (only)
If the 95% CI of the difference includes 0, what can you conclude about the difference?
It is not statistically significant
Null hyoothes: difference = 0
If the 95% CI of the difference does not include 0, what can you conclude about the difference?
The difference is statistically significant
Alternative hypothesis: difference does not equal 0
