A. Getting into SPSS:

Lab assignment 3: t-tests with SPSS

There are basically two kinds of t-tests. The between-groups (or "independent means") t-test is a statistical test designed to examine whether means from two different samples are significantly different from one another. The within-groups (i.e., "paired samples") t-test is designed to examine if the means of scores on two different variables for the same participants are significantly different from one another.

1. Independent means t-test (Use if you expect the average score on some variable to differ between two DIFFERENT groups of participants)

Example: Suppose you think that dogs get more ticks than cats during the Fall tick season.

You would have two variables: Type of animal (a categorical Independent variable with two levels: 1 = cat; 2 = dog) and total number of ticks (a continuous Dependent variable).

Your data would look something like this:

animal ticks

1.00 3.00

1.00 2.00

1.00 7.00

1.00 3.00

1.00 5.00

2.00 7.00

2.00 9.00

2.00 10.00

2.00 3.00

To see if the dogs have significantly more ticks than cats, you would need to conduct an independent means t-test like so:

1. Click on Analyze on toolbar.

2. Drag to Compare Means

3. Click "Independent Samples t-test"

4. Test variable is going to be the DV (ticks in this case)

5. Grouping variable is going to be the IV (animal in this case)

6. Next, click on "define groups."

7. For Group 1 type "1" for group 2 type "2"

8. Click on Continue

9. Click paste

10. Go to the .sps file and highlight the relevant commands.

11. Click on "run."

Here's what you'll see:

-----------------------------------

t-tests for Independent Samples of GENDER

Number

Variable of Cases Mean SD SE of Mean

-----------------------------------------------------------------------

ticks

ANIMAL 1 5 4.0000 2.000 .894

ANIMAL 2 5 7.6000 2.793 1.249

-----------------------------------------------------------------------

Mean Difference = -3.6000

Levene's Test for Equality of Variances: F= .361 P= .565

t-test for Equality of Means 95%

Variances t-value df 2-Tail Sig SE of Diff CI for Diff

-------------------------------------------------------------------------------

Equal -2.34 8 .047 1.536 (-7.143, -.057)

Unequal -2.34 7.25 .050 1.536 (-7.208, .008)

-------------------------------------------------------------------------------

Notice that your prediction is clearly a one-tailed hypothesis. Thus, divide p (.047) by 2 to come up with .0235 (yipes!). This number refers to the probability that your correlation was due to chance alone... pretty low odds! As long as this p value is below .05, it is considered "significant."

Here's how to report it:

Dogs had significantly more ticks (M = 7.6, SD = 2.79) than cats (M = 4.0, SD = 2.0; t(8) = -2.34, p < .05).

Note that the 8 in the parenthetical expression refers to the degrees of freedom term (above, "df," in printout).

2. Dependent means t-test (Use if you expect THE SAME participants to have significantly different scores on two different variables)

Example:

Does studying increase exam scores? (Assume you have 5 people who take one exam without studying � then they take a second exam for which they study).

your data would look something like this:

exam1 exam2

50.00 100.00

90.00 120.00

30.00 100.00

55.00 110.00

12.00 90.00

To see if studying improved exam scores, you would need to conduct a dependent means t-test like so:

1. Click on Analyze on toolbar.

2. Drag to Compare Means

3. Click "Paired samples t-test"

4. Paired variables are "exam1" and "exam2"

5. Click paste

6. Go to the .sps file and highlight the relevant commands.

7. Click on "run."

SPSS will give you output that looks something like this:

t-tests for Paired Samples

Number of 2-tail

Variable pairs Corr Sig Mean SD SE of Mean

-------------------------------------------------------------------------------

EXAM2 104.0000 11.402 5.099

5 .959 .010

EXAM2 47.4000 29.305 13.106

-------------------------------------------------------------------------------

Paired Difference Mean SD SE of Mean | t-value df 2-tail Sig

----------------------------------|--------------------------------------------

56.6000 18.649 8.340 | 6.79 4 .002

Again, notice that the default p value (under the heading "2-tail Sig") is for a two-tailed test.

You'd report your results like so:

Exam scores were significantly higher after participants studied (M = 104.00, SD = 11.40) compared with before they studied (M = 47.40, SD = 29.31; t(4) = 6.79, p < .05).

Note that the 4 in the parenthetical expression refers to the degrees of freedom term (above, "df," in printout).

_______________________________________________________________

Assignment:

1. Compute a within-groups (paired samples) t-test for the exam scores from the made up data.

Hand in:

A. A brief (no more than one page) summary outlining the question being asked by this analysis, the nature of the analysis being conducted, the results, and the implications of these results.

B. the .spo file

2. Compute a between-subjects (independent means) t-test examining sex differences in one of the four continous variables.

Hand in:

A. A brief (no more than one page) summary outlining the question being asked by this analysis, the nature of the analysis being conducted, the results, and the implications of these results.

B. the .spo file

3. Compute a between-groups t-test examining sex differences among college students in campus behavior.

As in the previous lab, you will need to collect naturalistic-behavior data from actual college students. Ultimately, you will be conducting a between-groups t-test to examine sex differences in some behavior.

Get in groups of 3 or 4.

Think of a continuous dependent variable that you can unobtrusively measure by observing male and female students on campus at this time. For instance, you could measure amount of time it takes to walk down a hallway alone, number of turns taken in a conversation within a group in a one-minute period, etc.

Compute a between-groups t-test to examine whether males and females differ significantly on this variable.

Hand in:

1. A brief (no more than one page) summary outlining the question being asked by this analysis, the nature of the analysis being conducted, the results, and the implications of these results.

2. the .spo file

Importantly, whenever you summarize the results from a t-test, you need to report the following statistics: t, df, p (or at least whether p is less than .05), M's, and SD's.

You are now done - go home and have fun!