62
2015, the teacher downloaded all videos submitted and commented by the students previously. Further, the researcher did posttest. For control group, the
posttest was held on Thursday, 28
th
May 2015 and on Friday, 29
th
May 2015. For experimental group, the posttest was held on Thursday, 28
th
May 2015 and on Saturday, 30
th
May 2015. The posttest took 90 minutes for each term half of class member for each group.
In posttest, the researcher found different numbers of students in control class who attended the test. The numbers of students attending pretest were 45
students, however there were only 44 students attending posttest. Therefore, there researcher conducted the change to be adapted in the process of t-test.
After having posttest, the researcher then distributed individual questionnaire in the experimental class to fill by the students. Then the researcher
analyzed the data obtained from posttest to compare to the data obtained from pretest for each group.
Following are the findings of this research divided into two parts based on the research questions.
1. Effect of Video Project to Promote Students’ Speaking Skill
The first research question showed the findings on statistical tests done by the researcher.
To know whether the scores of pretest and posttest of each group have been improved or not, the researcher displays the mean and standard deviation for both
group. Table 4.1 below shows the mean differences of two groups based on the result of pretest and posttest see Appendix I
63
Table 4.1 Means and Standard Deviations of Experimental Group N
Mean Std. Deviation
Minimum Score
Maximum Score
PreEx PostEx
39 39
10.3590 15.0769
3.96357 3.19032
2.00 8.00
17.00 20.00
Table 4.2 Means and Standard Deviations of Control Group N
Mean Std.
Deviation Minimum
Score Maximum
Score PreCo
PostCo 45
44 10.8667
12.9773 3.83524
3.02308 2.00
8.00 18.00
18.00
A change was appeared between means in both groups. The maximum score of the test was 20.00 divided into five categories in which each category has score
4. Both groups obtained the same minimum score in pretest 2.00 and in posttest 8.00. The maximum score of posttest in experimental group increased from
17.00 to 20.00 while the maximum score of posttest in control group did not get any change 18.00. The improvement of pretest and posttest mean scores in both
groups can be seen in the following chart.
Figure 4.1 Mean plots on Tests between EG and CG
2 4
6 8
10 12
14 16
1 2
M e
a n
s
Tests
Chart Title
Class E Class F
64
In addition, the improvement can be seen more obviously from each category of speaking test including Pronunciation, Grammar Accuracy,
Vocabulary, Fluency, and Interactive Communication based on the rubrics of speaking test by Pandiya. The following are the tables which show the
frequencies of speaking rubrics of both groups.
Table 4.3 Frequencies of Scale Criteria of Speaking of Experimental Group
Criteria Numbers of Students
Pron GA
Voc Flue
IC Pre
Post Pre
Post Pre
Post Pre
Post Pre
Post
Very Poor Poor
Average Good
Very Good 8
14 15
2 9
24 6
1 12
15 11
10 17
12 3
5 17
13 1
1 5
18 15
1 11
16 11
15 19
5 11
14 13
1 1
10 10
18
Total numbers of
students 39
39 39
39 39
39 39
39 39
39
Table 4.4 Frequencies of Scale Criteria of Speaking of Control Group Criteria
Numbers of Students Pron
GA Voc
Flue IC
Pre Post
Pre Post
Pre Post
Pre Post
Pre Post
Very Poor Poor
Average Good
Very Good 7
17 20
1 11
25 8
3 9
16 15
2 4
19 17
4 6
16 22
1 4
16 23
1 1
10 23
10
1 4
16 24
2 10
20 10
3 1
21 18
4 Total
numbers of students
45 44
45 44
45 44
45 44
45 44
The total numbers of students of the experimental group are 39 students while the total numbers of students of the control group are different in pretest 45
students and posttest 44 students. Number 0 shows that there is no student
65 getting those categories. From the tables above, it can be seen that there was
improvement in both group related to the criteria of „very poor‟. In posttest, there
was no student who got this kind of criteria. For all criteria, except pronunciation, the experimental group had more students who got perfect score very good than
the control group. There were 6 students getting perfect score very good in Pronunciation category, 12 students in Grammar Accuracy, 15 students in
Vocabulary, 5 students in Fluency, and 18 students in Interactive Communication. Meanwhile, in the control group, there were only 8 students who got „very good‟
in Pronunciation, 4 students in Grammar Accuracy, 1 student in Vocabulary, and 4 students in Interactive Communication. There was no student of the control
group who got „very good‟ in Fluency in posttest. It can be concluded that students in experimental group had more improvement in speaking than those in
control group. For the complete data of students‟ score, see Appendix D.
Then, in gathering the data for the experimental study, the independent sample t-test was conducted by the use of gain score. The t-test allows the
examination of the difference between the mean scores relative to the spread or variability of the scores which could not be showed by the descriptive statistics
only. It was applied to examine if the means of two groups of data are significantly different from one another. The t-test itself in fact indicates the
sample differences from the population by the use of means and the distribution of sample scores around them.
Before analyzing the statistical result, the researcher conducted normality test to know whether the data were normal or not to be tested. She used Kolmogorov-
Smirnov test to show that the sample comes from particular distribution. If the
66
result is not significant P0.05, the data have normal distribution. The result of the normality test can be seen completely in Appendix I and is presented briefly in
the following table.
Table 4.5 One Sample Kolmogorov Smirnov Test
The result shows that the Komolgorov Smirnov KS Z score of class E the experimental group in pretest is 0.803 with Asymp. Sig 0.540 and of class F the
control group is 0.663 with Asymp. Sig is 0.771 while in the Komolgorov Smirnov KS Z score of class E in posttest is 0.791 with Asymp. Sig 0.559 and of
class F is 1.345 with Asymp. Sig is 0.054. All is higher than 0.05 which indicates that the test distribution is normal for both pretest and posttest. Thus, the
independent sample t-test could be conducted. Before conducting independent t- test, the researcher conducted paired sample t-test to know whether any difference
between pretest score and posttest score in each group.
Table 4.6 t test Statistics of Experimental Group Pretest and Posttest
One-Sample Kolmogorov-Smirnov Test
44 44
39 39
10.80 12.98
10.36 15.08
3.849 3.023
3.964 3.190
.100 .203
.129 .127
.062 .153
.091 .115
-.100 -.203
-.129 -.127
.663 1.345
.803 .791
.771 .054
.540 .559
N Mean
Std. Dev iat ion Normal Parameters
a,b
Absolute Positiv e
Negativ e Most Extreme
Dif f erences Kolmogorov -Smirnov Z
Asy mp. Sig. 2-tailed Contrrol
Pre Test Control
Post Test Experiment
Pre Test Experiment
Post Test
Test distribution is Normal. a.
Calculated f rom data. b.
Paired Samples Test
-4.718 2.645
.424 -5.575
-3.860 -11.138
38 .000
Experiment Pre Test - Experiment Post Test
Pair 1
Mean Std.
Dev iation Std. Error
Mean Lower
Upper 95 Conf idence
Interv al of the Dif f erence
Paired Dif f erences t
df Sig.
2-tailed
67 From the table, it is shown that H0 is rejected if the
–t value t table or t value t table. Because t value is 11.138 t table which is 2,024, H0 can be rejected. It
means that meaning that there is any difference between pretest result and posttest result of experimental group see Appendix I.
Table 4.7 t Test Statistics of Control Group Pretest and Posttest
The table shows that the t = -8.24. H0 is rejected if the –t value -t table or t
value t table. Because t value is 8.24 t table 2,017, H0 can be rejected, meaning that there is any difference between pretest result and posttest result of
control group. From the result of t-test for each group, it can be seen that both control
group and experimental group have significant difference in pretest and posttest while control group did not get any treatment. The numbers of students getting
lower score in posttest in experimental group were the same as those in control group. However, the numbers of students getting higher score in posttest in
experimental group were bigger than those in control group. Then, after knowing the paired sample correlations for each group, the
researcher conducted independent samples t-test to know whether any significant difference between both groups see Appendix I.
The following tables show the result of the gain scores and the statistical analysis of independent sample t-test.
Paired Samples Test
-2.182 1.756
.265 -2.716
-1.648 -8.24
43 .000
Contrrol Pre Test - Control Post Test
Pair 1
Mean Std.
Dev iation Std. Error
Mean Lower
Upper 95 Conf idence
Interv al of the Dif f erence
Paired Dif f erences
t df
Sig. 2-tailed
68
Table 4.8 Group Statistics of Independent Sample t-test
The variable of those statistical analysis were the gain scores from the experimental group and from the control group. From the analysis, it was found
out the significant difference between both groups. The output of means and standard deviation is followed by the value of mean difference. In the
experimental group consisting of 39 students, the mean is 15.08 sd = 3.190 while for the control group consisting of 44 students, the mean is 12.98
sd=3.023. The mean difference between both groups is 0.167.
Table 4.9 The Result of the Independent Samples Test
The Lavenes Test for Equality of Variances is to represent a test of the hypothesis that the population from which the groups were sampled have equal
variances. It aims to prove the sample homogeneity of both experimental and control groups to claim that they are from the same population.
The t-test results labeled equal variances which is assumed gives the t-value t =3.077 and the degree of freedom df = 81. Ha is accepted if
–t value - t table or t value t table. The t table is at 1.989 3.077 1.989 which indicates that the
Group Statistics
44 12.98
3.023 .456
39 15.08
3.190 .511
Group Control Post Test
Experiment PostTest Gain Scores
N Mean
Std. Dev iat ion Std. Error
Mean
Independent Samples Test
.003 .960
-3.077 81
.003 -2.100
.682 -3.457
-.742 -3.067
78.57 .003
-2.100 .685
-3.462 -.737
Equal v ariances assumed
Equal v ariances not assumed
Gain Scores F
Sig. Lev enes Test f or
Equality of Variances
t df
Sig. 2-tailed
Mean Dif f erence
Std. Error Dif f erence
Lower Upper
95 Conf idence Interv al of the
Dif f erence t-test f or Equality of Means
69 difference is significant because t-test result is higher than t table. Here, the result
while in the table the result is presented in the form of 2-tailed significance, the researcher tested by using 1-tailed test because most calculators and computers
give the exact p-value for 2-tailed tests. The researcher used the interpretation of 1-tailed test by using t table.
As the t-test result shows that the difference between the two groups is significance, the null hypothesis Ho is rejected. In other words, the first alternative
hypothesis Ha is acceptable. This condition, indeed, shows that there is a better achievement in speaking in Experimental Group than in Control Group. The
students achievement based on the post-test result is indeed better than their scores in pretest which indicates that the treatment has positive effect and it
promotes the students speaking skill including five categories of pronunciation, grammar accuracy, vocabulary, fluency, and interactive communication seen from
the score difference.
2. Video Project to Promote Students’ Learner Autonomy