Related Samples t statistic
Subtract the mean of sampling distribution of mean differences from the sample mean difference/by standard error
Assumptions for the related-samples t tests
1.)Observations within the group should be independent
2.)The population of difference scores should be normally distributed
Advantages for related samples from a within-groups design
1.)Requires less participants for the same amount of data
2.)Good for studying changes overtime
3.)Reduces problems that might arise from individual differences
Disadvantages for related samples from a within-groups design
1.)Time related factors (history threats)
2.)Many factors might change between the two occasions
3.)Order effects
Order Effects
One level of the IV influences responses on s subsequent level of the IV
When to use Z test?
When you are given the population standard deviation
When to use One Sample t-test?
You have to estimate the population standard deviation
When to use Related Sample t-test?
-When you have something to compare something else to
-Subtract scores
-Correlation between scores across the same people
When to use Independent-Sample t-test?
-2 sets of variables
-Different people on different days
-No reason to suspect same results
To calculate a t-statistic, what should the population distribution be?
A normal distribution where N>30
Within-Subjects
-Dependent observations
-Paired sample t-test(related)
-Testing diff variables WITHIN the person
Between-Subjects
-Independent observations
-Independent sample t-test(unrelated)
-2 separate groups are tested and not correlated
What is the right way to state the null hypothesis for a paired sample t-test?
u1=u2 or uD=0
-no difference
What is the right way to state the alternative hypothesis for a paired sample t-test?
u1 is not equal to u2 or uD is not equal to 0 (difference)
What does a two tailed test ask for?
The difference
