T-tests Flashcards
(17 cards)
Is a t-distribution useful in comparing group means?
Yes
The critical value tells us how likely it is that two sample means at a specified distance from one another will occur in two randomly drawn samples from the same population
The distance between sample means in a sampling distribution of means is measured in units called standard errors of the mean
What does it mean if the means of two groups are separated by a distance less than the critical t?
We can conclude that they are probably from the same population
This finding supports the null hypothesis that the population means of the two groups are equal
What does it mean if the means of two groups are separated by a distance greater than the critical t?
We can conclude that they are probably from different populations
The intervention did work
This finding supports the alternative hypothesis
What are the three types of t tests?
One sample t - only have one sample, want to test this against a hypothetical group mean
Dependent samples t - have two means, same subjects in both groups
Independent samples t - two means and two groups
What assumptions must be met for a t-test?
The means are calculated from interval or ratio data
Sample size per group should be at least 20-30
Dependent samples - data are paired by a strong commonality and the paired difference scores are approx normally distributed
Independent samples - observations are independent of one another, equal variances between groups, and data within groups are approx normally distributed
How do you calculate a t-score?
Divide the difference in means by the sample SEM (SD/square root of n)
What does it mean if the t statistic is greater than the critical value?
Reject the null hypothesis and conclude that there is a difference between groups
How is the population of possible sample means distributed?
Based on the sampling distribution of means
What are within-subject designs?
The two sets of data may come from dependent sample(s); e.g. “before treatment” and “after treatment”
What are between-subject designs?
The two sets of data may come from two independent samples (e.g. women and men, or a randomized controlled design)
What does the dependent samples t-test test?
The average of the differences between paired measurements against 0 (null value)
How do you calculate the t-score for a dependent samples t-test?
Mean difference within pairs/standard error of difference scores
How do you calculate the standard error of difference scores?
SD of difference/square root of n
How do you calculate confidence intervals for dependent samples?
Mean of the difference scores +/- t-critical(standard error of difference scores)
*if it does not cross zero, you would reject the null
What is the equation for an independent samples t-test?
Between group variability/within-group variability
What is the equation for within-group variability?
Sq root ((s1^2/n1) + (s2^2/n2))
