Research Findings

A. Research Findings

1. The Setting of the Research

The research was held on SMP Negeri 1 Ulugawo that is located on Ulugawo Sub-district, Nias regency. It is around 54 kilometers from Gunungsitoli town. The total numbers of teachers were 16 persons. Besides, there were also official administration which consisted of 4 persons. Meanwhile, the total number of the class are 8 in three different grades. The seventh grade consisted of three classes, the eighth grade consisted of three classes, while the ninth grade consisted of two classes. The total numbers of the students were 238 persons.

The population of the research was the eighth grade which consisted of three classes. The total numbers of the students of population were 91 persons. In conducting the research, the researcher selected two representative samples that was selected through cluster sampling; they were; VIII-A class as experimental class, and VIII-C as control class and each of them consisted of 30 students.

2. Tried-Out Instrument

a. Validity of the Instrument

To get the data in the research, the researcher used a tool named instrument. Yet, before using that instrument, the researcher validated it first to make sure that it is appropriate to be used.

The instrument that was used on the research is essay test which is considered needs rational validity. Sugiyono (2011:123) writes that the instrument has rational or internal validity only if the criteria that is found on the instrument itself reflect to things what will be measured, so the researcher validated the research instrument internally.

In validating the instrument, the researcher consulted the instrument to three validators. Mrs. Dra. Sulasmi, as the first validator and the lecturer of English Study Program. Besides, the instrument was also consulted to Mr. Syukurman Zai, S.Pd. as the second validator and an English teacher of the SMP Negeri 1 Ulugawo. Moreover, the instrument was also consulted to Miss Ariani Harefa, S.Pd., as the third validator and an English teacher of SMP Swasta BNKP Maranatha. All validators were trusted having high qualification on language testing. Meanwhile, the result of the consulting can be seen on the analysis sheet that were checked by all validators (See Appendix 7) and shown that the instrument was appropriate to be used.

b. Reliability of the Instrument

Based on rational validity as decide before, the researcher directly decided the reliability of the instrument itself. According to Sugiyono (2011:122), a reliable instrument is not definitely valid, but a valid instrument is trusted reliable. Based on the theory, the researcher knew the reliability of the instrument only by seeking its validity that can be seen on the analysis sheet (See Appendix 7) without any statistical calculation.

3. The Data Analysis

In analyzing the data, the researcher focused on several important points namely the mean of the data, standard deviation, variances, the normality, homogeneity and hypothesis testing that are explained as follows.

a. Mean

According to the pretest data calculation, the researcher noticed the mean of students’ mark of experimental class was 45 that was classified bad and under the MCC score (See Appendix 9a); while the mean of students’ mark of control class was

47 that was also classified bad and under the MCC score (See Appendix 9b). Based on the calculation itself, the researcher analyzed that both classes had low ability in writing recount text. However, the researcher noticed also that the students’ ability of control class was higher than the students’ ability of experimental class. Fortunately, in posttest, there was progres sing of the students’ ability in writing recount text that was shown by the mean of their mark. The mean if the students’ mark of

experimental class was 78 that was classified good (See Appendix 9c); while the mean of students’ mark of control class was 71 that was classified in good enough

criterion (See Appendix 9d). Both of those classes were pass on MCC score, but the students’ ability of experimental class was higher than the students’ ability of control class. To make clear, the mean of students’ marks are stated in the next table.

Table 6

The MEAN of PRE-TEST and POST-TEST SCORES of EXPERIMENTAL and CONTROL CLASS

Mean of Pre- Mean of Class

The Number

MCC

of Students

Experimental Class

30 63 45 78 Control Class

b. Degree of Mastery

Besides looking for the mean of the students’ score, the researcher also calculated the degree of students’ ability in writing recount text. In pre-test data of experimental class, it was counted that no one have excellent, very good and good criteria, there was 6.7 % of the students had ability in good enough criterion, 23.3% of them had ability in enough and bad criteria, while 46.7 % of students had ability in

very bad criterion (See Appendix 11a). However, the students’ degree mastery had progression after treatment that can be seen on post-test data. Based on the

calculation, the post-test data shown that there was no one have excellent, very bad and bad criteria, approximately 36.7 % of students were in very good criterion, 33.3 % of them were in good criterion, 16.7 % were in good enough criterion, 13.3 % of them were in enough criterion (See Appendix 11c). To make clear, the percentage degree of students’ mastery of pre-test and post-test of experimental class in writing recount text present in the next chart.

enta 25 erc

Good enough Good

Very bad

Bad

Enough

Very good

Excellent

Criteria

Chart 1: The percentage degree of students’ mastery of pre-test and post-test of experimental class in writing recount text

Besides, the calculation of pre-test and post-test data of control class also shown the progression of the students’ ability in writing recount text. In pre-test data,

it can be seen that there was no one student had ability in excellent, very good and good criteria. It was approximately 10.0 % students who had ability in good enough criterion. In other hand, there was 23.3 % of students had ability in enough criterion,

26.7 % in bad criterion, and 40.0% of students had ability in very bad criterion (See Appendix 11b). Surprisingly, the students’ ability had progression after treatment also

like the students’ ability of experimental class that can be seen on post-test data. After calculation, the post-test data shown that no one of student had ability in excellent,

bad and very bad criteria. It was calculated 36.7 % students had ability in very good criterion, It was approximately 33.3 % of them who had ability in good criterion, 16,7 % in good enough criterion, and 13.3 % in enough criterion (See Appendix 11d). To

make clear, the percentage degree of students’ mastery of pre-test and post-test of control class in writing recount text present in the following chart.

15 erc 13.3% P

Good enough Good

Very bad

Bad

Enough

Very good

Excellent

Criteria

Chart 2: The percentage degree of students’ mastery of pre-test and post-test of control class in writing recount text.

c. Standard Deviations

Based on the data calculation, the standard deviation of the pre-test for experimental class was 15.21 (See Appendix 12a), while in post-test was 11.58 (See Appendix 12c). Besides, the standard deviation of the pre-test for control class was

14.81 (See Appendix 12b), while in post-test was 12.55 (See Appendix 12d). To make easy for understanding, the standard deviation of students’ marks are stated in

the following table.

Table 7

The STANDARD DEVIATION of PRE-TEST and POST-TEST SCORES of EXPERIMENTAL and CONTROL CLASS

Standard

Standard

The Number of

Class

Deviation of

Deviation of

Experimental Class

30 15.21 11.58 Control Class

d. Variance

According to the pre-test calculation, the variance of experimental class was 231.37 (See Appendix 12a) while the variance of control class was 219.30 (See Appendix 12b), while according to the post-test data calculation, the variance of experimental class was 134.13 (See Appendix 12c), while the variance of control class was 157.50 (See Appendix 12d). The next table is made to make the point of variance be easily for understood.

Table 8

The VARIANCE of PRE-TEST and POST-TEST SCORES of EXPERIMENTAL

and CONTROL CLASS

Variance of Class

The Number of

Variance of

134.13 Control Class

Experimental Class

e. Normality

Before treated the samples of the research, the researcher sought made sure that the selected samples were representative or not. For this reason, the researcher

sought the normality of the data when conducting the research to know the students’ prior knowledge of each class.

To test the normality, the researcher used Liliefors’ formula as suggested by Herhyanto, et al., (2014:8.17). According to the pre-test data calculation, the researcher noticed that L count of experimental class was lower than L table

(0.0995<0.1610) (See Appendix 13a). Meanwhile L count of control class was also lower than L table (0.0993<0.1610) (See Appendix 13b).

As the conclusion, both of class had normal distribution before treatment. It shown that the students has equal prior knowledge. This condition indicated that both samples were representatives’ samples and ready to be treated.

Furthermore, based on the calculation of post-test data, the researcher also noticed L count of experimental class was lower than L table (0.1263<0.1610) (See

Appendix 13c), while L count of control class was also lower than L table (0.1059<0.1610) (See Appendix 13d).

Lastly, L table was still higher that L count of both class and it shown the samples had normal distribution in post-test calculation. To make easy for understanding, the following is table of the normality of pre-test and post-test of experimental and control class.

Table 9

The NORMALITY of PRE-TEST and POST-TEST SCORES of EXPERIMENTAL

and CONTROL CLASS

L count of Class

L count of

L table

Pre-Test

Post-test

Experimental Class

0.1263 Control Class

f. Homogeneity

In the research, the researcher used formula as suggested by Irianto, et al., (2007:276) as has been explained in Chapter III (See Page 60) to know the homogeneity of the samples . According to the pre-test data calculation (See Appendix 14a), it indicated F count = 1.06, while F table = 1.86. Since F count (1.06) ≤ F table

(1.86), so it can be concluded that the samples of the research were homogeny. Besides, based on the post-test data calculation (See Appendix 14b), it was also indicated F count = 1.17, equal with F table = 1.86. Since F count (1.17) ≤ F table (1.86), so it also can be concluded that the samples of the research were homogeny. To make

easy for understanding, the next is table about the homogeneity of pre-test and post- test scores of experimental and control class.

Table 10

The HOMOGENEITY of PRE-TEST and POST-TEST SCORES of EXPERIMENTAL and CONTROL CLASS

F count of Class

F count of

F table

Pre-Test

Post-test

Experimental Class

1.06 1.86 1.17 Control Class

g. Hypotheses Testing

According to the calculation of hypotheses testing in Appendix 15, it was noted that T count = 2.245, while T table = 2.002 (It was confirmed with df = n 1 +n 2 -2

which the level of significance 0.05). Since T count (2.245 ) > T table (2.002), it can be concluded that H a was accepted and H 0 was rejected. In other words, the result of the calculation data of post-test on the research shown that there was significant effect of

Modelled Writing on the students’ ability in writing skill at the eighth grade of SMP Negeri 1 Ulugawo in 2015/2016.