A. Comparison of means test for two unique groups B. Requires i. Interval data ii. 2 Variables (1) 1st Variable represents the Test Variable within SPSS (a) This is your dependent variable (b) This is the variable for which a group mean will be calculated for both defined groups nd (2) 2 Variable is the grouping variable (a) This is the independent variable (b) It is used to separate the cases into the 2 groups that will be compared (c) It should be a variable with only 2 values (i) Can be 1 or 2 (ii) Can also be a nominal variable recorded in text form (SPSS refers to this as a string variable) (3) The following data table demonstrates the proper format for data in SPSS Dependent/Test Variable 4 5 6 6 7 8 C.
D.
Independent/Grouping Variable 1 1 1 2 2 2
Process for conducting the test
i. From the menu: Select Analyze Compare Means Independent- Samples T-Test ii. Move dependent variable into the box labeled Test Variable iii. Move the independent variable into the box labeled Grouping Variable iv. Select the Define Groups box beneath the grouping variable (1) Enter the value for the independent variable that identifies those cases in group 1 in the first box (2) Enter the value for the independent variable that identifies those cases in group 2 in the second box (3) ***Important note: the values entered in the boxes must be identical to those recorded in the data set*** v. Select continue vi. Select the options box at the lower right corner of the T-test menu vii. Enter the appropriate level of confidence and select continue viii. Select ok to run the test Interpreting the output i. The first output table provides the mean, standard deviation and standard error for each group ii. The second output table lists several pieces of information
The value of t is provided in the 3h column of the table
(a) *Always use the top value for t (the one that assumes equal variances) (2) The degrees of freedom are listed in the 4th column of the table labeled as df (3) Analyze the value listed for Sig. (2-tailed) in column 5 of the second table (Again use the top value for Sig.) to determine level of confidence (a) If the value reported is equal to or less than .05 (at the 95% level of confidence) or .01 (at the 99% level of confidence), the difference between group means is statistically significant and the null hypothesis is rejected (b) If the value reported is greater than .05 (at the 95% level of confidence) or .01 (at the 99% level of confidence), the difference between groups is not statistically significant and the null hypothesis must be accepted (4) If the null is rejected, compare means to determine which groups scores higher than the other (5) **Extraneous Information** (a) The first 2 columns Provide data labeled Levenes Test for Equality of Variances Ignore this information. (b) The 2nd Row is labeled equal variances not assumed Ignore (c) When producing output files, it is probably wise to delete these portions of the output tables Dependent Samples T-Test A. Compares the same sample at two different points in time B. Requires: i. Interval data ii. 2 variables Each variable represents one observation of the dependent variable C. Process i. From the menu: Select Analyze Compare Means Paired- Samples T-Test ii. Select the two variables to be used for the test iii. Move them into the box labeled Paired Variables iv. Select options v. Enter the appropriate level of significance and select continue vi. Select OK D. Interpreting the output i. The first box provides the mean, standard deviation and standard error ii. The third box provides the value for t, degrees of freedom, and the level of significance (1) These values are reported in the final 3 columns of the table iii. Determining significance (1) If the value reported is equal to or less than .05 (at the 95% level of confidence) or .01 (at the 99% level of confidence), the difference between group means is statistically significant and the null hypothesis is rejected (2) If the value reported is greater than .05 (at the 95% level of confidence) or .01 (at the 99% level of confidence), the difference between groups is not statistically significant and the null hypothesis must be accepted (1)