|
4.3 Children’s Learning with Panwapa
4.3.1 Global Citizenship Pre-Test and Post-Test
Table 4.29 lists the global citizenship pre and post-test total scores achieved by each child
from a maximum score of 25. The test consisted of multiple-choice questions and the
questions are provided in Appendix C. The change in scores over the two tests ranged from
-2 to +13. One child’s score decreased by 2 and the scores for all the other children
increased by at least +3. The mean increase was 7.37 (with a standard deviation of 3.14)
representing an increase of 58.87% from pre-test to post-test.
58
Pre-Test
Post-Test
Increase
Child 1
14
24
10
Child 2
5
12
7
Child 3
11
24
13
Child 4
13
20
7
Child 5
8
15
7
Child 6
14
21
7
Child 7
12
21
9
Child 8
13
19
6
Child 9
15
21
6
Child 10
16
21
5
Child 11
12
23
11
Child 12
17
20
3
Child 13
16
23
7
Child 14
14
12
-2
Child 15
17
25
8
Child 16
7
17
10
Child 17
15
23
8
Child 18
8
17
9
Child 19
10
19
9
Child 20
9
21
12
Child 21
18
22
4
Child 22
10
18
8
Child 23
9
15
6
Child 24
13
19
6
Child 25
13
22
9
Child 26
9
20
11
Child 27
20
23
3
Mean
12.52
19.89
7.37
St. Deviation
3.67
3.46
3.14
Table 4.29
Global Citizenship Pre-Test and Post-Test Scores (out of 25)
59
Shapiro-Wilk
Statistic
df
Sig.
PreTest
.985
27
.954
PostTest
.926
27
.055
Table 4.30
Shapiro-Wilk Tests for normality of the pre and post-test scores
Shapiro-Wilk tests (Table 4.30) confirmed that both pre and post-test scores followed a
normal distribution. None of the values in the columns labelled
Sig.
is less than .05, which
would indicate a non-normal distribution. It was therefore appropriate to use a parametric
test to compare the means of both tests.
A dependent, or matched-pairs, t-test was used to compute the statistical significance of the
difference between the pre and post-test mean scores. This test compares the mean values
of two sets of scores which have been generated by the same participants, and it is used to
assess the likelihood that the difference between mean values is due to an experimental
effect rather than occurring by chance (Field and Hole 2003).
The null hypothesis, H
0
, stated that there was no statistical difference between the pre and
post-test mean scores of the participants. The experimental hypothesis, H
1
, states that the
post-test mean scores are better than the pre-test mean scores. The SPPS output for the
dependent t-test is shown in Tables 4.31, 4.32 and 4.33.
Mean
N
Std. Deviation
Std. Error Mean
PostTest
PreTest
19.8889
27
3.44555
.66310
12.5185
27
3.67288
.70685
Table 4.31
Summary statistics for the pre and post-test scores
Table 4.31 displays summary statistics including the mean and standard deviation.
60
N
Correlation
Sig.
PostTest & PreTest
27
.613
.001
Table 4.32
Pearson correlation between the Pre and Post test scores
Table 4.32 displays the correlation coefficient, or Pearson’s
r
, between the two test
conditions and the two-tailed significance value. This correlation coefficient (
r
= .613)
reveals a strong correlation, and the significance value of less than .01 indicates this is
highly statistically significant.
Paired Differences
t
df
Sig.
(1-tailed)
Mean
Std.
Deviation
Std. Error
Mean
95% Confidence
Interval of the
Difference
Lower
Upper
PostTest - PreTest 7.37037
3.13967
.60423
6.12836
8.61238
12.198
26
.000
Table 4.33
Dependent t-Test (Paired Samples Test) for Pre- and Post-test Scores
Table 4.31 contains the main SPSS output for the dependent t-Test and reveals if the
difference between the two means is large enough not to happen by chance. As the test
statistic (t = 12.198), based on 26 degrees of freedom, is beyond the 95% confidence
interval for the difference between the means and with a highly significant one tailed
probability of p=.000, the null hypothesis H
0
must be rejected. The experimental
hypothesis, H
1
, which states that the post-test mean is greater than the pre-test mean, must
therefore be accepted.
The effect size, or
r
-value, was calculated using this equation taken from Field (2009).
=
+
Using the value of
t
and
df
from Table 4.5,
r
can be computed as follows:
=
12.198
12.198 + 26 =
148.791
174.791 = .923
61
As the threshold for a large effect is .5 (Field 2009), this effect size of .923 signifies a very
large effect. Therefore, it can be concluded that the increase in test scores after the
Panwapa programme ended was both substantive and statistically significant.
Do'stlaringiz bilan baham: |
