The one-tail t-test was used to compare two mean scores from the post-tests and the pre-tests, concerning listening and speaking of each group (control or experiment) to see the improvement of the students after the experimental period.
3.1.1. Control group: (see detail scores at Appendix E)
The scores of the pre-tests and post-tests on listening and speaking of the students of control group were computed respectively by running the one-tail t-test using Microsoft Excel Analysis tool.
t-Test: Paired Two Sample for Means LISTENING
Post-test Pre-test
Mean 7.15 5.833333
Variance 0.408929 0.764881
Observations 15 15
Pearson Correlation 0.902007 Hypothesized Mean Difference 0
df 14
t Stat 12.55846
P(T<=t) one-tail 2.6E-09 t Critical one-tail 1.76131 P(T<=t) two-tail 5.21E-09 t Critical two-tail 2.144787
Table 2: One-tail t-test for control group (Listening test)
As can be seen from the table above, Mean of the listening post-test is 7.15 compared to that of the listening pre-test (5.83). In addition, the obtained t-test value (tStat) is 12.56 which is much bigger than the critical t-test value (tCritical one-tail) which is 1.76 with p (one-tail) = 2.6E-09 (equal to 0.0000000026), much smaller than 0.05. Then, it is clear that the t-statistic yields a significant result, which means that the difference in mean scores between listening post-test and listening pre-test is statistically significant. In other words, students in control group
31
made progress in listening with the teacher’s conventional technique of teaching pronunciation.
t-Test: Paired Two Sample for Means SPEAKING
Post-test Pre-test
Mean 8.133333 5.95
Variance 0.963095 1.1625
Observations 15 15
Pearson Correlation 0.62696
Hypothesized Mean Difference 0
df 14
t Stat 9.461153
P(T<=t) one-tail 9.23E-08 t Critical one-tail 1.76131 P(T<=t) two-tail 1.85E-07 t Critical two-tail 2.144787
Table 3: One-tail t-test for control group (Speaking test)
It can be seen from the table that students in control group also made much progress in speaking. With Mean of the speaking post-test is 8.13 compared to that of the speaking pre-test (5.95) and tStat is much bigger than tCritical one-tail, together with p (one-tail) = 9.23E-08 (equal to 0.0000000923), it is almost 100% of confidence that the progress that the students in control group made in speaking was valid and reliable. From the mean of listening and speaking tests, it is clear that the students in control group made more progress in speaking than in listening (the difference in mean of speaking tests is 2.18 compared to that of the listening is 1.31). This seems due to the fact that the students work with Vietnamese teacher, while the listening test is in authentic voice. Another reason for this may lie in the fact that there are no tools or criteria to check the level of difficulty of the speaking and listening test.
3.1.2. Experimental group: (see detail scores at Appendix E)
32
The scores of the pre-tests and post-tests on listening and speaking of the students in experimental group were also computed by running the one-tail t-test using Microsoft Excel Analysis tool.
t-Test: Paired Two Sample for Means LISTENING
Post-test Pre-test
Mean 7.616667 5.8
Variance 0.677381 0.805357
Observations 15 15
Pearson Correlation 0.65036
Hypothesized Mean Difference 0
df 14
t Stat 9.738133
P(T<=t) one-tail 6.48E-08 t Critical one-tail 1.76131
P(T<=t) two-tail 1.3E-07
t Critical two-tail 2.144787
Table 4: One-tail t-test for experimental group (Listening test)
As is illustrated by the table above, Mean of the listening post-test is 7.6, which is much higher than that of the listening pre-test (5.8). Moreover, the tStat is 9.73, much bigger than the tcritical one-tail which is 1.76. In addition, p (one-tail) is 6.48E-08 (equal to 0.0000000648) which is much smaller than 0.05, meaning that it is almost 100% of confidence that the difference between the two means of the listening post-test and pre-test is significant. In other words, the students in experiment group made lots of progress in listening with the new techniques of teaching pronunciation.
As can be seen from the table below that the students in experimental group also made remarkable improvement in speaking. With Mean of the speaking post- test is 8.53 compared to that of the speaking pre-test (6.0) and tStat (9.9) is much bigger than tCritical one-tail (1.76), plus p (one-tail) of 5.25E-08 (equal to 0.0000000525), it is apparent that the progress that the students in experimental group made in speaking was valid and reliable. With this result, we can reject the
33
null hypothesis 1 (Ho1). From the mean of the listening and speaking tests, it is also clear that the students in experimental group also made more progress in speaking than in listening with the difference in mean scores of speaking and listening is 2.53 and 1.82 respectively.
t-Test: Paired Two Sample for Means SPEAKING
Post-test Pre-test
Mean 8.533333 6
Variance 0.989881 1.803571
Observations 15 15
Pearson Correlation 0.67825
Hypothesized Mean Difference 0
df 14
t Stat 9.906358
P(T<=t) one-tail 5.25E-08
t Critical one-tail 1.76131
P(T<=t) two-tail 1.05E-07
t Critical two-tail 2.144787
Table 5: One-tail t-test for experimental group (Speaking test) 3.2. The two-tail t-test (two-sample t-test):
The two-tail t-test was used to compare two mean scores of the post-tests, including listening and speaking between control group and experimental group or to measure the students’ progress after 2 months of experiment which is the main purpose of this study. The result was the main answer to the research question 2.
The scores of the listening and speaking post-tests of the students of control and experiment group were computed respectively by running the two-tail t-test using Microsoft Excel Analysis tool.
As the table 6 reveals, the Mean of listening post-test of the experimental group is higher than that of the control group, with 7.61 and 7.15 respectively.
However, the tStat of 1.73 is smaller than the tCritical two-tail of 2.04, and the p value two-tail of 0.094 is much bigger than 0.05, which means that the difference is not significant enough in favour of the experimental group. In other words, it
34
appears that the use of pronunciation websites does not surely help students improve their pronunciation better than the conventional technique. The limited time as well as the small number of students taking part in the study are probably the reasons that make the difference not really significant.
t-Test: Two-Sample Assuming Equal Variances LISTENING
Exp. Group Control Group
Mean 7.616667 7.15
Variance 0.677381 0.408929
Observations 15 15
Pooled Variance 0.543155
Hypothesized Mean Difference 0
df 28
t Stat 1.734106
P(T<=t) one-tail 0.046949
t Critical one-tail 1.701131
P(T<=t) two-tail 0.093898 t Critical two-tail 2.048407
Table 6: Two-tail t-test for control group and experimental group (Listening test) t-Test: Two-Sample Assuming Equal Variances
SPEAKING
Exp. Group Control Group
Mean 8.533333 8.133333
Variance 0.989881 0.963095
Observations 15 15
Pooled Variance 0.976488
Hypothesized Mean Difference 0
df 28
t Stat 1.108555
P(T<=t) one-tail 0.138527
t Critical one-tail 1.701131
P(T<=t) two-tail 0.277054 t Critical two-tail 2.048407
Table 7: Two-tail t-test for control group and experimental group (Speaking test)
35
The table above indicates that the students in experimental group also made certain positive improvement in speaking in comparison with the students in control group. However, similarly to the listening, the difference is not significant enough to come to the conclusion that using websites is more beneficial than conventional technique because the tStat of 1.10 is much smaller than the tCritical two-tail of 2.04, and the p value two-tail of 0.27 which is much bigger than 0.05. In addition, as can be seen from the mean scores of the listening and speaking post-tests of both groups, there appears to be no significant differences in students’ performance in pronunciation whichever skill, speaking or listening, is tested.