Statistical analysis on quantitative data Issues with missing data Descriptive Statistics

84 accompanied with the results of the repeated measures MANOVA on the effects of CBIT on the level of teachers‘ efficacy beliefs. The third part was the supporting information on the level of teachers‘ work engagement and the relation between teachers‘ efficacy and work engagement. This information provides additional informat ion concerning the level of teachers‘ self-efficacy beliefs. The second category was in the form of transcriptions of recorded interviews with the four teachers who were observed in the classroom teaching. Interviews were done in English except for one teacher who registered as having low English efficacy. This particular participant was interviewed in Bahasa Indonesia. This interview was later translated by the researcher and was checked by two people who spoke fluent English and Indonesian. The third category of materials was materials collected from the classroom observations of the four participants‘ classroom teaching practices. These observations were used to develop vignettes of the teachers and their teaching practices in the classroom. Observations were initially recorded in the form of an observation schedule and then together with the researcher‘s field notes, the case study vignettes were written.

3.10.2 Statistical analysis on quantitative data

In analyzing the data the researcher classified the data into two classifications, the quantitative and qualitative data. These classifications were due to the 85 different nature of the data and the different treatments needed. Quantitative data in this research were analyzed using the SPSS package of data analysis. Some statistical forms of analysis were used in analyzing the research data. Those analyses were descriptive analysis, the General Linear Models consisting of Multivariate Analysis of Variance and the Repeated Measures Multivariate Analysis of Variance, and the Spearman Rho correlations.

3.10.3 Issues with missing data

A number of non-response answers appeared in the data. Although some alternatives are offered in the literature, the researcher chose the random assignment within groups as the method to handle the missing data. This was done by firstly dividing the sample into subgroups on the basis of background variable most likely related to the variable with missing data. Secondly, the missing data in the cases in a particular variable were substituted with valid values of the closest cases within the same variable De Vaus, 2002. This method was chosen based on the consideration that deletion of cases with missing data and pair wise deletion would severely affect the number of cases due to the random nature of the missing data. Secondly, this method was chosen as an effort to minimize the effect of substituting the missing data thus reducing the variability on the variable. 86

3.10.4 Descriptive Statistics

Descriptive statistical analysis used in this research consisted of two analyses, the descriptive and the frequency. The descriptive summary covered the means, standard deviations, median and modus, while the frequency dealt with the number of cases in the data. In doing the descriptive analysis, the researcher also sought to identify the nature of the data, particularly the degree of skewness of the data and the kurtosis.

3.10.5 General Linear Model MANOVA and Repeated Measures MANOVA