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 men should score higher on the Emotional Sensitivity Scale (ESS) than women.
You would have two variables: GENDER (a categorical Independent variable with two levels: 1 = Female; 2 = Male) and total scores on the ESS for each participant (a continuous Dependent variable).
Your Data would look something like this:
Gender ESS
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 9.00
2.00 10.00
2.00 3.00
To see if the males’ scores are significantly higher than the females’ scores, 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 (ESS in this case
5. Grouping variable is going to be the IV (gender 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-Test
| VAR00001 | N | Mean | Std. Deviation | Std. Error Mean | |
|---|---|---|---|---|---|
| ESS | 1.00 | 5 | 4.0000 | 2.0000 | .8944 |
| 2.00 | 5 | 7.6000 | 2.7928 | 1.2490 |
| Levene’s Test for Equality of Variances |
t-test for Equality of Means | |||||
|---|---|---|---|---|---|---|
| F | Sig. | t | df | Sig. (2-tailed) | ||
| ESS | Equal variances assumed | .361 | .565 | -2.343 | 8 | .047 |
| Equal variances not assumed | -2.343 | 7.248 | .050 | |||
| Mean Difference | Std. Error Difference | 95% Confidence Interval of the Difference |
|
|---|---|---|---|
| Lower | Upper | ||
| -3.6000 | 1.5362 | -7.1426 | -5.7449E-02 |
| -3.6000 | 1.5362 | -7.2075 | 7.529E-03 |
________________
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:
Males scored significantly higher (M = 7.6, SD = 2.79) than females (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 eating grapefruit raise people’s IQs? (Assume you are using the Schmedley IQ test; you have 10 people that you measure before they eat a bunch of grapefruits; then you measure the same participants later).
your data would look something like this:
pre post
50.00 100.00
90.00 120.00
30.00 100.00
55.00 110.00
12.00 90.00
To see if eating grapefruits made people smarter, 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 post and pre
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-TEST
Mean |
N |
Std. Deviation |
Std. Error Mean |
||
|---|---|---|---|---|---|
| Pair 1 | PRE | 47.4000 | 5 | 29.3053 | 13.1057 |
| POST | 104.0000 |
5 | 11.4018 | 5.0990 | |
N |
Correlation |
Sig. |
||
|---|---|---|---|---|
| Pair 1 | PRE & POST |
5 | .959 | .010 |
| Paired Differences | ||||||
|---|---|---|---|---|---|---|
| Mean | Std. Deviation | Std. Error Mean | ||||
| Pair 1 | PRE – POST | -56.6000 | 18.6494 | 8.3403 | ||
| t | df | Sig. (2-tailed) | ||||
|---|---|---|---|---|---|---|
| 95% Confidence Interval of the Difference |
||||||
| Lower | Upper | |||||
| -79.7563 | -33.4437 | -6.786 | 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:
IQ scores were significantly higher after participants ate grapefruits (M = 104.00, SD = 11.40) compared with IQ scores before participants ate grapefruits (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 mood scores from the jealousy data.
Recall that for this data set, you mood was measured at three times. The first mood measurements comprised a baseline measure. The second measurements represent mood after imagining partners engaging in sexual infidelity. The third measurements represent mood after imagining partners engaging in emotional infidelity. Recall that higher scores on this measure indicate more psychological discomfort.
Conduct ONE within-subjects t-test using two of these three composite mood variables.
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 .spv file
3. Compute a between-subjects (independent means) t-test examine sex differences in variables from Monica’s thesis data (you can obtain her data set here).
In her thesis, Monica examined sex differences in variables pertaining to perceptions of body image. In her data, the variable gender refers to the gender of participants; a value of 1 stands for females, 2 stands for males. The variable in the rightmost column, bes_t refers to body esteem scale total scores. Higher scores here indicate more positive perceptions of ones body.
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 .spv 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 .spv 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), Ms, and SDs.
You are now done – go home and have fun!