Psy 242

Lab exercise

Hypothesis Testing: Comparing 2 means with t-tests in SPSS

In this lab session, you will practice the steps for comparing two means in SPSS using t-tests. You will also learn how to import a text file for use as data in SPSS, and how to check for outliers in the data before testing your hypothesis.

The data file you will use for this lab is 242-lab-data-ttest.sav. This file contains data from an experiment on Spatial Cuing from Coglab. The hypothesis you will test is that reaction times to detect a light are faster when a valid cue is given for the location where the light will appear than when an invalid (misleading) cue is given for the location. (For more information on the experiment, see http://coglab.wadsworth.com/experiments/SpatialCueing/index.html.) Begin by right-clicking on the link to the data file in the preceding sentence, and saving the file on your hard drive. Then double-click the file you saved to start SPSS and open the file.

1. Look for outliers

Before doing hypothesis testing, you should always do some exploratory data analysis to check for errors and outliers. Because the mean is sensitive to outliers, one absurdly slow reaction time could completely change the results of an experiment like this one.

To check for outliers in SPSS, make the following selections on the menus:

Analyze -> Descriptive Statistics -> Explore

Move the labels for both columns that contain the dependent variable data (the columns valid and invalid in this case) into the box labeled Dependent List. Then click OK.

Look at the stem-and-leaf plots and the box plots to see if SPSS identified any outliers. If so, remove them before doing further analyses. One way to remove outliers is to just delete the individual data points that are outliers by hand. Another is to use a filter in SPSS to exclude them. Here is how:

• Look at the exploratory data analysis and determine a value for each condition that will exclude the outliers but not exclude any non-outlying data points.
• Bring the SPSS data window back to the foreground (click on its icon on the taskbar)
• In the menu select Data -> Select Cases
• Click on a condition that has outliers you wish to exclude
• In the Select box, click the If Condition is Satisfied button
• Click the If box just below that button
• Type the rule that will exclude the outliers into the box in the upper right of the screen. For example, if you are excluding times greater than 600 milliseconds from the variable Valid you should type in: valid <= 600
• Click Continue, then click OK

2. Follow steps in the previous hypothesis testing handout or lecture notes for doing a t-test.

Name ______________________ Date _________________ Psy 242 Lab 1

3. In complete sentences, report the results below. Be sure to identify the data that you will analyze, state how many data points were excluded as outliers and what criteria you used to exclude them, restate the hypothesis that you were testing, state whether the results confirmed the hypothesis or not, state the means and standard deviations of each condition, and correctly state the results of the t-test. Turn in this page to receive credit for this lab.

Here is an example of how a t-test could be reported (for a different experiment):

Mean exam scores were analyzed in an independent samples t-test in which testing condition was the independent variable. Three outliers (defined as scores of less than 20%, identified as outliers by an SPSS stem and leaf plot) were excluded from analysis. I had predicted higher scores in the caffeine testing condition than in the no caffeine condition. This prediction was not confirmed. The mean exam score with caffeine was 87% (SD = 2.3%), which was not significantly greater than the mean exam score of 80% (SD = 3.4%) without caffeine, t (67) = 1.34, n. s.

4. Re-do the analysis without excluding outliers. Remove the filter (Data -> Select Cases; select All Cases). Now re-do the t-test without excluding any outliers. Report the results of this analysis as you did for #3.

5. From this exercise, what would you conclude about the effects of outliers?