Mt = the total average
St = the total score of standard deviation
P = proportion subject with true answer
q = proportion subject with false answer
11
An items is considered valid if the score of validity is higher than 0.444. It was based on the table of product moment where the r-critical for 20 subjects, df = N- nr = 20-
2= 18 with alpha 0.05 was 0.444. The result of validity analysis shows that there were 15 items considered invalid.
They were 1,3,5,11,13,16,20,23,24,25,26,28,31,33,36, and 39 items number. Finally, the total valid items in try out were 25 items. see Appendix 13
J. Reliability of the Test
Arikunto says that reliability shows that instrument can be believed to be used as a tool of data collecting technique is good enough.
12
If the data are true based on the fact, no matter how many data are taken the result is same. Reliability shows the
degree of mainstays about something. Reliability means the data can be delivered, so it can be relied on.
To get reliability of using adjective, the writer used spearman Brown formula as
11
Anas Sudijono, Pengantar Statistik Pendidikan, Jakarta: Rajawali Press, 2014, p. 258
12
Suharsimi Arikunto,Op.Cit.,p.221
follows: r
xx
Notes: r
xx
= estimated reliability of the entire test r
1122
= pearson r correlation between the two halves.
13
see Appendix 17
To get reliability of the descriptive writing ability test, the write used inter rater ability.
It was done by two raters who examine the students’ writing test with the intention of knowing the reliability of the test. see Appendix 18 Then, the writer
used Rank Order formula as follows: ρ = 1 -
∑
Notes: ρ = the number of rank order correlation
6 and 1 = constant number D = difference of rank order correlation D= R1
– R2 N = the number of students
After that, the writer consulted the result to the criteria of the reliability as follows. 1. A very low reliability ranges from 00.0 to 0.19
2. A low reliability ranges from 0.20 to 0.39
13
Donald Ary,Op.Cit.,p.244
3. An average reliability ranges from 0.40 to 0.59 4. A high reliability ranges from 0.60 to 0.79
5. A very high reliability ranges from 0.80 to 1.00.
14
Based on the data see Appendix 17, the reliability of adjective mastery test was 0.907. Based on the category, the reliability of adjective mastery was very high
reliability since it amount 0.80-1.00.
From the data gained see appendix 18, the reliability of writing descriptive text test was 0.77. Based on the category, the reliability of writing descriptive text was high
reliability since it amount 0.60- 0.79.
K. Readability of Test
Readability tests are indicators that measure how easy a document is to read and understand. For evaluators, readability statistics can be solid predictors of the
language difficulty level of particular documents. The essential information in an evaluation document should be easily understandable.
15
To know readability of the students’ writing ability of descriptive text test instrument,
the writer followed Kouamé’s research. The participants asked to evaluate
14
Sugiyono, Statistika untuk Penelitian, 17
th
Edition, Bandung, Alfabeta,2010,p.231
15
Julien B. Kouamé, Journal of Multi Disciplinary Evaluation Vol. VI No. 14 August 2010: Using Readability Tests to Improve the Accuracy of Evaluation Documents Intended for Low-Literate
Participants, Western Michigan University, Michigan, p.133.