Spss 2
Spss 2
Spss 2
SPSS T tests
Data file used: gss.sav
In the lecture about chapter 2, only the One-Sample T test has been explained. In this handout, we
also give the SPSS methods to perform Independent Samples T tests and Paired Samples T tests, for
the sake of completeness. You might need these later in the course.
particular mean.
Compare Means
To test the null hypothesis that a sample comes from a population with a
In a one-sample T test you must select in the source variable list the variable you want to test and move it
into the Test Variable(s) List.
You can move more than one variable into the list to test all of them against the specified Test Value. For
each selected variable, SPSS calculates the t statistic and its observed significance level.
Button Options
The confidence interval level can be changed here. See following figure.
Performing a One-Sample T test on the variable age, with a test value of 40, the results are the following:
T-Test
One-Sample Statistics
Age of Respondent
N
1495
Mean
46,23
Std. Deviation
17,418
Std. Error
Mean
,450
One-Sample Test
Test Value = 40
Age of Respondent
t
13,822
df
1494
Sig. (2-tailed)
,000
Mean
Difference
6,23
95% Confidence
Interval of the
Difference
Lower
Upper
5,34
7,11
In the table One-Sample Statistics you can find the number of valid cases (N), the mean age of the
respondents (Mean), the standard deviation (Std. Deviation) and the standard error (Std. Error Mean).
In the table One-Sample Test you can see the test value (40) which is tested against the age distribution.
The t value is quite high (13,822) and reveals a significant (Sig. = ,000) difference. This indicates that the
mean (46,23) is not equal to the test value (40). The Mean Difference (46,23 40) is also given.
Then, you must select the variable whose values define the two groups and move it into the Grouping
Variable box. To define how this variable has to be split into groups 1 and 2 use Button
Define
groups. You can choose between the options use specified values and cut point, see following figure.
In this example, the values correspond to codes used in variable satjob2: 1 = very satisfied; 2 = not very
satisfied.
The cut point can be used to separate continuous numerical variables. If one group corresponds to small
values of the grouping variable and the other group to large values, select this option and enter a value that
separates the groups.
When you perform an Independent Samples T test on the variable age, grouped by the variable satjob2
(Very satisfied versus Not very satisfied), the results are the following:
T-Test
Group Statistics
Age of Respondent
Job Satisfaction
Very satisfied
Not very satisfied
Mean
43,09
39,51
489
651
Std. Deviation
13,824
12,736
Std. Error
Mean
,625
,499
Age of Respondent
Equal variances
assumed
Equal variances
not assumed
F
2,943
Sig.
,087
df
Sig. (2-tailed)
Mean
Difference
Std. Error
Difference
95% Confidence
Interval of the
Difference
Lower
Upper
4,522
1138
,000
3,58
,791
2,024
5,127
4,469
1002,625
,000
3,58
,800
2,006
5,145
In the table Group Statistics you can find, for each group (Very satisfied and Not very satisfied) the
number of valid cases (N), the mean age of the respondent (Mean), the standard deviation (Std. Deviation)
and the standard error (Std. Error Mean).
In the table Independent Samples Test you first find the result of Levenes Test for Equality of Variances.
As the name suggest, it tests the condition that the variances of both samples are equal, indicated by the
value of F. A high number of F results normally in a significant difference and you should look at the row
behind Equal variances not assumed. A low number of F results normally in a non-significant difference
and you should look at the row behind Equal variances assumed.
In this example, the variances of the ages of Very Satisfied and Not very satisfied are compared, and the
quite low value of F (2,943) reveals a not significant (Sig. = 0,087) difference. Thus, you should look at the
row behind Equal variances assumed, and you should NOT use the values in the row behind Equal
variances not assumed. You see that the t value is 4,522 and the 2-tailed significance of 0,000. Thus, there
is a significant difference in age between the groups Very Satisfied and Not very satisfied.
We performed a Paired Samples T test on the variables maeduc and paeduc, which are matched pairs of
variables.
T-Test
Paired Samples Statistics
Pair
1
Mean
Std. Deviation
Std. Error
Mean
11,10
1044
3,397
,105
11,02
1044
4,274
,132
N
1044
Correlation
,649
Sig.
,000
Pair
1
Mean
,08
Std. Deviation
Std. Error
Mean
3,310
,102
95% Confidence
Interval of the
Difference
Lower
Upper
-,12
,28
df
,785
Sig. (2-tailed)
1043
In the table Paired Samples Statistics, separate summary statistics (mean, N, standard deviation and
standard error) are given for the two matched variables.
The Correlation value in table Paired Samples Correlations indicates how strong the variables are related.
The table Paired Samples Test is most important now. A one-sample T test is performed on the
differences between the two variables. The null hypothesis is that the average difference between the two
measurements is 0, in the population. The test value of the one-sample T test is therefore 0. The mean
(0.08) is the difference between the two means standing in the table Paired Samples Statistics (11.10
11.02 = 0.08) and the standard deviations of the difference is 3.310. The 95% Confidence Interval for the
average difference is from 0.12 to 0.28. This confidence interval includes the value of 0, so the null
hypotheses is true. There is no average difference between the two variables. The Sig. of the T value also
indicates this, as 0.432 < 0.05.
,432