RESEARCH SETTING AND RESPONDENTS DATA ANALYSIS PROCESS

71 question of the interview invited the students to reflect on what they obtained from the test and evaluate themselves from their experiences . Asking students‟ perceptions on the test based on their perspectives includes learning priorities, learning difficulties, self-grading, and self-evaluation. In order to confirm the result of the questionnaire, ten participants were interviewed. The participants selected for the interview were those who had taken the TKBI test. Five participants of the interview had passed the test and five participants had to retake the test. The consideration to choose these participants was because they could give more complete information about how the test affected their attitudes during the test preparation and test taking processes and how their test performance was affected by their attitudes.

D. RESEARCH SETTING AND RESPONDENTS

The study was conducted in Sanata Dharma University, from October to December 2014. The respondents were 130 students from five departments in the university, including 30 students from Mathematic Department, 25 students from Elementary Teacher Education Department, 30 students from Indonesian Letters Department, 30 students from Psychology Department, and 35 students from Accounting Department. The questionnaires were distributed in the classroom, so all students in the classroom were invited to fill out the questionnaires and return them to the researcher. The sampling technique was cluster random sampling. The respondents were the representatives from each faculty or the subpopulation. The process of 72 this random sampling started after the subpopulations were determined. The respondents were randomly selected from the five faculties by distributing the questionnaires to the students from a class in each faculty. The classes were determined in random, meaning that students from all classes in each faculty had the equal chance to be the samples. The respondents of this research were from their fifth to ninth semester in academic year 20142015. They came from various parts of Indonesia and they had different level of English proficiency, level of confidence, and level of motivation to learn English. These students were preparing themselves for taking TKBI test. Almost all of the students planned to take the test in the following semester. Some of the students had taken TKBI test, but some of them still needed to retake the test.

E. DATA ANALYSIS PROCESS

Creswell mentions some steps of data analysis process 2012, pp. 238-239. The step begins by preparing and organizing the data. The data are organized by type into questionnaire and interview results. The questionnaire results are organized in table and counted into frequency per item. The results of the interview are transcribed, explored and coded to describe the partici pants‟ opinions and ideas. Then, the data are analyzed and interpreted to obtain findings. After the findings are summarized, the researcher can use the statistical methods to interpret the findings obtained from the data gathering. The last step of data analysis is reporting the findings. 73 Survey research mostly uses descriptive statistics to describe the characteristics of the population by counting the mean, median, and mode of the data from the sample. This measurement is known as the measure of central tendency. It describes and summarizes the distribution of the data and represents the value of the individual in the entire population Gravetter Wallnau, 2006, pp. 71-72. By using descriptive statistics, the findings of the research can be generalized to portray the real situation of the population. After the mean, median, and mode of the data were calculated, the findings could be used to support Likert scale analysis in interpreting the answer to the research problem. Mi + 1.5SDi Very positive Mi + 0.5SDi Positive Mi – 0.5SDi Negative Mi – 1.5SDi Very negative Table 3.1 The Likert Score Criteria Sudijono, 2009, p. 175 Criteria of Answer Score Very positive 3.4 – 4.0 Positive 2.8 – 3.3 Fair 2.5 – 2.7 Negative 2.2 – 2.4 Very negative 1.4 – 2.1 Mi = ½ maximum score + minimum score = ½ 4 + 1 = 2.5 SDi = 15 maximum score – minimum score = 15 4 – 1 = 0.6 Mi ideal mean SDi ideal standard deviation 74 Therefore, the first data analysis was conducted to infer information from the quantitative data. In order to analyze the research findings and answer the research question, the questionnaire results were calculated using Likert scale. Since the questionnaire did not use mid point, the responses were transferred into numerical values of 1, 2, 3, and 4. For totally disagree, the items were transformed into 1, disagree items worth value 2, agree items obtained 3, and value 4 was given for totally agree items. After all responses from the students were transcribed into numerical values, the number of answers in each category was counted into frequency and percentage. The researcher counted how many students answered strongly disagree, disagree, agree, and strongly agree. After counting the frequency of answers, the values of Likert scale could be summed up and calculated to get the average score of each questionnaire item. Then, the Likert score was interpreted for the analysis process. The interpretation used criteria proposed by Sudijono 2009, p. 175. The calculation had to find the ideal mean Mi and ideal standard deviation SDi to develop the score criteria using the formulas. The interpretation of the score itself was derived based on the criteria which can be seen in Table 3.1. Referring to the criteria, the researcher could identify whether the students gave positive or negative responses to forty statements about the language, language learning, and the test, as the main focus of the study. When the average of students ‟ response was 2.5 to 2.7, they gave fair response to the statement, meaning that the number of positive response and negative response were not significantly different. Negative response could be identified from average response of 2.4 and below 2.4 75 which means that almost all of the students disagreed with the statements. When the average score was 2.8 or above 2.8, almost all of the students agreed with the statements, so the response was positive. After finding the characteristics of the students ‟ attitudes towards the test, the relationship between the components of attitudes could be identified from the result of Spearman ‟s rho analysis. The questionnaire data were analyzed using SPSS to correlate the components. After the researcher inputted and processed the data in SPSS, the result could show the correlation and test the hypothesis. If the score was below 0.05, the null hypothesis that there was no correlation between the components of attitudes was rejected. That score confirmed that the components were correlated. The data analysis could not support causal and effect relationship because of the limitation of the data due to confidentiality. The results of quantitative data analysis could be supported with qualitative data from the questionnaire. Although the nature of the data could not support inferential statistic for cause and effect relationship, it was still possible to infer correlational relationship between the constructs. The data could be calculated by using Spearman‟s rho to identify the correlation between the components of attitudes. Furthermore, the result could ensure the validity and reliability of the instrument and the analysis. In order to support the quantitative analysis, the second analysis process was conducted on the interview result. As mentioned before, the interview was transcribed first. Before the coding, the researcher needed to list the important themes that appeared in s tudents‟ answers. The themes could be referred to the 76 theoretical constructs. Then, the researcher could identify the themes found in the interview. The data analysis based on the themes was beneficial to describe students‟ attitudes and perceptions about language, language learning, and the test. To report the interview result, the researcher adapted Saldana ‟s coding system 2009, pp. 16-30. In the way of adapting the system, the researcher mentioned the theme, the number of participant, page and response into codes like theme.Pnrnpn where Pn was the number of participant, rn was the response number, and pn was the page number. This coding system could help the researcher to refer to the students ‟ responses and attitudes. Mentioning the theme in the system was beneficial for tracking the data because one response could contain more than one theme.

F. TRUSTWORTHINESS