3.1 Research Design

This research is conducted in Ex Post Facto design. According to Furchan (1982:50), ex post facto is a systematic empiric research where the researcher can’t set up the independent variable directly because it happened, or because of the independent variable can’t be manipulated.

Researcher does an observation without any manipulation; just choose the class that still uses traditional assessment and the class that uses portfolio assessment. The observation progresses at SMP 4 Jambi.

The design of this research can be seen at the table bellow:

Table 3.1 Ex Post Facto Design

Group Independent Variable Dependent Variable

1st

2nd (X)

- Y1

Y2

Source: Furchan (1982:404)

Explanation:

(X) : The using of portfolio assessment that have been run;

Y1 and Y2 : The students’ learning achievement

3.2 Subject of the Research

3.2.1 Population

According to Arikunto (2002:108), population is all of the subjects who are connected to the research. Gay (1987:102) gives clearer definition of population that is the group of interest to the researcher, the group to which she or he would like the results of the study to be generalizable. Population at this research is the 1st class of students of SMP 4 Jambi.

Table 3.2 Numbers of Students at 1st Class of SMP 4 Jambi

2007/2008 Academic Years

CLASS STUDENTS

MEN WOMEN TOTAL

1 A 24 18 42

1 B 25 20 45

1 C 20 20 40

1 D 21 20 41

1 E 22 18 40

1 F 23 19 42

1 G 21 20 41

1 H 23 20 43

1 I 17 25 42

TOTAL 196 180 376

Source: TU SMP 4 Jambi

3.2.2 Sample

Sample is a number of individuals for a study in such a way that the individuals represent the larger group from which they were selected (Arikunto, 2002:109). In this research, the sample is selecting by using clustered random sampling.

The steps to select the sample are arranged as bellow:

(1) Choose the class which use portfolio assessment as the first group.

Class 1A is the only one class that uses portfolio assessment. Researcher decides Class 1A as the first group.

(2) Choose other classes, which still use traditional assessment. To step up the credibility of ex post facto design should be used criteria to choose the sample (in Furchan, 1982:393) as the following:

a. Have the same English teacher with the first group;

b. Given the same material with the first group;

c. Given the same process of learning except the process of assessment;

d. Find that the classes are normally distributed, have homogeny variances, and have the same class average of English ability.

(To get this data, researcher asked the document of first English daily test score to the teacher then found mean and standard deviation of each class).

(3) The last step is choosing one class that uses traditional assessment by using roll of papers randomly.

3.2.2.1 Normality Test

To find the normal distribution of the classes, researcher uses Lilliefors Test. According to Sudjana (in Syafyendri’s thesis, 2002: 30), normality test is done to find whether the subject is normally distributed or not. The steps are arranged as bellow:

a. Arrange the score of summative test in a table by ordering the score from lower score to higher score;

b. Find standard scores from by using formula:

Explanation: = mean

s = standard deviation

c. By using list of standard normal distribution, find the chance F (Zi) = P (Zi), {F(Zi) = score of Zi that is finding in F table for z distribution subtract (-) with 0.5};

d. Find the proportion score of standard scores which is less then or the same with score Zi. It is denoted by S(Zi), S(Zi) = sum of Xi divide with n;

e. Calculate margin of F(Zi) – S(Zi), then decide the absolute score;

f. Choose the highest absolute score, which is denoted by Lo. If Lo < Ltable so the average score of students’ learning achievement is normally distributed.

3.2.2.2 Homogeneity Test

To find the homogeneity of the variances of each class, researcher uses Bartlett Test because the populations are more than two classes (in Elina’s Thesis, 2005:27). The steps are arranged as follow:

a. Calculate the values needed in Bartlett test as listed in table 3.3 bellow:

Table 3.3 List of Bartlett Test

Subject Df 1/df Si2 Log Si2 (df) Log Si2

1

2

k n1 – 1

n2 – 1

nk – 1 1/ n1 – 1

1/ n2 – 1

1/ nk – 1 S12

S22

Sk2 Log S12

Log S22

Log Sk2 (n1 – 1) Log S12

(n2 – 1) Log S22

(nk – 1) Log Sk2

Total ∑ (ni-1)

- - ∑{(ni-1)Log Si2}

b. Find total variance of all subject

c. Find value of B by using formula:

B = (Log S2) ∑ (ni-1)

d. Bartlett test by using Chi-Square statistic

2 = (Ln 10) (B – ∑{(ni-1)Log Si2})

It requires significance level () at the 0.05 level. If ratio2 < table2 so the result is: all of population variance is homogeny.

3.2.2.3 Analysis of Variance

According to Sudjana (in Elina’s thesis, 2005:29) to test that there is no difference of English ability among the populations is used analysis of Variance in one way classification. The steps are:

a. Calculate the mean of squares

b. Calculate a between-group sum of square

c. Calculate the sum of squares from all of groups

∑ X2 = X12 + X22 + X32

d. Calculate a within-group sum of square

Dy = ∑ X2 – Ry – Ay

e. Make a table of analysis of variance (ANOVA)

Table 3.4 Analysis of Variance (ANOVA)

Source of Variance Df Jk KT F

Means

Between Groups

Within Groups 1

k – 1

∑ (ni – 1) Ry

Ay

Dy R = Ry/1

A = Ay/(k-1)

D = Dy/∑(ni¬¬¬-1) A/D

Total ∑ni - - -

f. The criterion to make a decision is through the following statement:

If Fratio < Ftable with df for numerator k -1 and df for denominator ∑ (ni – 1) for significance at the level 0.05 (), it means that the English ability of the populations is not different.

3.3 Data Collection Method

3.3.1 Kind of the Data

The data in this research are from observation and documentation. The documentation involve the data about number of the students, English score at first daily test to know the average ability, and the score of final test to find the difference of students’ learning achievement between the two classes.

3.3.2 Source of the Data

Sources of the data in this research are:

1) TU (Tata Usaha) of SMP 4 Jambi, to get the data about numbers of the students of 1st class at SMP 4 Jambi in 2007/2008 academic years.

2) English teacher of 1st class at SMP 4 Jambi, to get document of the first English daily test (formative test) score at first semester in 2007/2008 academic years. Formative test was teacher made test.

3) English teacher of 1st class at SMP 4 Jambi, to get document of the final test (summative test) score at first semester in 2007/2008 academic years. Final test was teacher made test.

4) Observation of 1st class at SMP 4 Jambi, to know how teaching and learning process held, to find the differences of assessment process at the two classes.

a) The Class with Portfolio Assessment

The first step, teacher gave information to the students about the using of portfolio that would be conduct in this semester, then explained what portfolio assessment is. Each student was given peaces of paper, which contained the indicators of portfolio assessment. Those papers are put in a folder. Before teacher gave material, teacher did warming up by asking the students everything that was related to the material and then explained teaching learning goals to the students.

The second step, teacher explained the material. Then teacher gave some exercises to the students. While the students did the exercises, teacher assessed the students’ attitude. After students were finished, teacher and students discussed the exercises together.

The last step, teacher concluded the material and then gave a homework for the next meeting.

In the next meeting, teacher collected the homework and corrected it at home. It was returned to the students at the next meeting. Students, who did errors in their work, must make a revision. Score of the work were recorded on format of structured task in portfolio. All of students’ works were recorded in a folder, so both teacher and students could see the difference in finishing previous work and the revision.

b) The Class with Traditional Assessment

The first, teacher did warming up by asking the students everything that was related to the material and then explained teaching learning goals to the students.

Next, teacher explained the material. Then teacher gave some exercises to the students. After students were finished, teacher and students discussed the exercises together.

The last, teacher concluded the material and then gave a homework for the next meeting. Teacher did not ask the students to make revision of their works.

3.4 Technique of Data Analysis

In this research, researcher would conduct the data analysis through the following steps:

a. Doing tabulation to the first English daily test (formative test) score at first semester in 2007/2008 academic years.

b. Classifying the sample based on the result of the tabulation.

c. Analyzing the final test score. Goal of data analyze is to find whether there is a difference of the students’ learning achievement between the two class, by using t-test. According to Faisal (in Yasril, 1998: 47), there are some conditions must be shared by the data before doing t-test: (1) they must be interval data; (2) they have normal distribution; and (3) they have the same variance.

d. Interpreting the data that have been analyzed.

3.5 Significance of the Difference between Two Means

According to Ardhana (in Yasril, 1998: 50), to test difference between two means is used t-test formula:

Explanation:

= mean of the first group

= mean of the second group

= estimate value of population variance

= number of students at the first group

= number of students at the second group

In calculating S2 (estimate value of population variance) is using formula:

The degrees of freedom is n1 + n2 – 2 at the level of significance () 0.05. It will be said that there is a significant difference between the two classes if tratio > ttable.