Summary
Highlights
The video starts by introducing the paired-samples t-test for comparing two related means, often used in pre-test/post-test designs or with matched pairs like twins or spouses. The data must be structured with pre- and post-measurements in separate columns.
To perform the test, navigate to 'Analyze', then 'Compare Means', and select 'Paired-Samples T-Test'. Users must input the pre- and post-measurement variables. The order of input can affect the sign of the t-value but not the overall significance.
The output includes descriptive statistics (means, standard deviations, standard error of the mean) for both measurements. Crucially, a significant correlation between the paired samples (pre and post) is essential for the validity of this test. The t-value, degrees of freedom (N-1), and significance level (p-value) are reported. A p-value less than 0.05 indicates a statistically significant difference between the means. The mean difference and confidence interval for the difference are also discussed, with the confidence interval's signs indicating significance (both positive/negative means significance, mixed signs mean no significance).
Reversing the order of input variables (e.g., post-test then pre-test) changes the sign of the t-value and the confidence interval, but the p-value and conclusion of significance remain the same. This highlights that the overall interpretation of results is unaffected by the input order in SPSS.
The one-sample t-test is used to compare the mean of a single sample to a known or hypothesized population mean (a 'test value' or 'critical score'). The video uses an example of comparing a sample's depression score to a clinical cutoff point to determine if the sample, as a whole, exhibits depression.
To perform this test, go to 'Analyze', then 'Compare Means', and select 'One-Sample T-Test'. Users select the variable to test and input the 'Test Value' (the population mean or critical score). As with the paired-samples test, a pre-test (the initial state, e.g., high anxiety) might not differ significantly from the cutoff, but a post-test (after intervention) might show a significant decrease, indicating effective treatment.
The output provides the descriptive statistics for the sample mean and the t-value, degrees of freedom, and significance level. If the p-value is greater than 0.05, there is no statistically significant difference between the sample mean and the test value. If it's less than 0.05, a significant difference exists. The confidence interval is again useful: if it contains zero (one value negative, one positive), there's no significant difference; if both values have the same sign, there is a significant difference.
It's possible to test both pre- and post-measurements against the same critical value in a single one-sample t-test. The video demonstrates this, showing that the pre-test mean might not be significantly different from the cutoff, while the post-test mean might be significantly lower, indicating a positive treatment effect. This dual comparison provides a comprehensive view of the change relative to a benchmark.