The empirical study
5.2 Data analysis methods
This section explains all statistical techniques and tests that will be used via SPSS and
Smart-PLS in (Chapter 6) for analysing the collected data and for extracting the significant findings.
1. SPSS Statistics is defined a statistical programme manufactured by IBM, Inc. IBM SPSS is aimed to perform a variety of statistical procedures, explain how to choose the appropriate statistics and present the results in a usable form (Cronk 2017). These techniques can be exposed as follows:
a. Cronbach's Alpha coefficient test for measuring the reliability: Reliability measures the skill to obtain consistent scores from the participants’ answers (Treiman et al. 2009). Reliability can be tested through the internal–consistency of questionnaire (Ibid.). Cronbach's Alpha coefficient is a test and technique for
measuring the reliability of the research instruments eventually when the value is above 0.7 they are considered acceptable but when values are above 0.8 they are preferable and suggesting a good internal consistency reliability for the scale with this sample (Bryman & Cramer 2011;Pallant 2007;Pallant 2010;Tavakol and Dennick 2011). This test has been used in Chapter 6 (data analysis Chapter) for measuring the reliability of the current study constructs.
b. KMO Kaiser-Meyer-Olkin test and Barlett’s test of Sphericity for measuring the validity: This test ensures and measures the validity of research instruments (sub-scales and items) (Comrey and Lee 1992;Sekaran & Bougie 2010). KMO and Barlett’s test of Sphericity measure of sampling adequacy (Pallant 2007;Pallant 2010). The KMO index ranges from 0 to 1 with 0.6 suggested the minimum value for a good factor analysis (Field 2009; Pallant 2007; Pallant 2010). Bartlett’s test of Sphericity should be significant (the value of Sig should be less than .05) (Ibid.). It is suggested that the Correlation matrix for evidence of coefficients is greater than 0.3 (Pallant 2010;Tabchnick and Fidell 2007). If the researcher does not find so many values more than 0.3 he/she should reconsider the use of factor analysis. This test has been used in Chapter 6 (data analysis Chapter) for measuring the validity of the current study constructs.
c. Skewness & Kurtosis tests for testing the normality: Normality is used for identifying the sample distribution and the distribution of scores on the dependent variable if it is normal i.e. normality test is used to describe a symmetrical, well- shaped curve which has the greatest frequency of scores in the middle, with smaller frequencies towards the extremes (Gravetter and Wallnau 2004;Pallant 2010). It is important for researchers to understand the normality because if this assumption is not met then the logic behind hypothesis testing is flawed (Field 2009). The Skewness provides an indication of the symmetry of distribution and if Skewness values are positive that means the scores are clustered to the left at the low values and if the values are negative that means the scores are clustered to the right at the high values (Pallant 2010; Pallant 2007). But Kurtosis provides information about the ‘Peakedness’ of the distribution if kurtosis values are positive that means the distribution is rather peaked (clustered in the centre with
long thin tails) and if the values are negative that means the distribution is flat (too many cases in the extremes) (Ibid.). If the distribution is normal the Skewness and Kurtosis value must be 0 (Ibid.). This test has been used in Chapter 6 (data analysis Chapter) for measuring the sample distribution and the distribution of scores (normality test) in the current study.
d. Kolmogorov- Smirnov test: This test measures the normality if the results are not significant (Sig. Value > .05) that means the value indicates normality. If the values are Significant (Sig. Value < .05 or less) that means the distributions of scores are not normal (Pallant 2016). This test has been used in Chapter 6 (data analysis Chapter) for identifying the nature of distribution (normality test) in the current study. The aim of using this test is to explore the normality test to identify the kind of tests supposed to be used statically for analysing the data (Ibid.).
e. The difference between the Original Mean and 5% Trimmed Mean for accepting or refusing the outliers: Outliers are cases with scores that are quite different from the remainder of the sample, either much higher or much lower (Pallant 2010; Pallant 2016). Sometimes these values should be removed and sometimes can be retained (Ibid.). According to Pallant if the difference between the Original Mean and 5% Trimmed Mean is too little so there are no extreme scores having a big influence on the Mean, eventually, the data does not need further investigation and it is appropriate for more statistical analysis and the values are not so different from the remaining distribution therefore the extreme cases can be retained (Ibid.). This test has been used in Chapter 6 (data analysis Chapter).
f. The Mean is the measure of central tendency that you are most likely to have heard of because it is simply the average score. To calculate the mean we simply add up all of the scores and then divide by the total number of scores we have (Field 2009). We can write this in equation form as:
∑ represents the summation X represents scores N represents number of scores This test has been used in Chapter 6 (data analysis Chapter).
g. The Median is a way to quantify the centre of a distribution and to look at the middle score when scores are ranked in order of values (Field 2009). The median is relatively not affected by extreme scores at either end of the distribution and it is also relatively unaffected by skewed distributions and can be used with interval, ordinal and ratio data (it cannot be used with nominal data because these data have no numerical order) (Ibid.). If the total number of numbers (n) is an odd number, then the formula is given below:
(TutorVista.com Website 2013). If the total number of the numbers (n) is an even number, then the formula is
given below:
(Ibid.).
This test has been used in Chapter 6 (data analysis Chapter).
h. The Correlation (Spearman correlation) is used when the scores are not normally distributed and when you want to explore the strength of the relationship between two continuous variables (Pallant 2016). This gives indication of both the direction (positive or negative) and the strength of the relationship (Ibid.). A positive correlation indicates that as one variable increases, so does the other (Ibid.). A negative correlation indicates that as one variable increases, the other decreases (Ibid.). Additionaly, the effect sizes are interpreted according to
large effect (Weast-Knapp et al. 2015). This technique has been used in the next Chapter (Chapter 6).
2. Smart-PLS 3 is an objective statistical program, easy to use, and can be used for latent variable modelling. It uses the art methods to design the graphical user interface (Hair et al. 2014). Additionally, it is good software function for Partial Least Squares Structural Equation Modelling (PLS-SEM) (Wong Kwong-kay 2013). This programme and its techniques have been used in Chapter 6. The data analysis methods can be defined as follows:
a. Structural Equation Modelling (SEM) method helps researchers to integrate unobservable variables measured indirectly by indicator variables and to assist computing the measurement error in observed variables (Hair et al. 2014).
b. Partial Least Squares-Structural Equation Modelling PLS-SEM (PLS path modelling): is used for exploring the variance in the dependent variables through examining the model (Hair et al. 2014).
5.3Summary
This Chapter has been about the empirical part of teaching which involves delivering a lesson via Online. The Chapter has explained the teaching methods, the selected unit of learning, the participants, the context of the study, the formative assessment, the role of the instruction sheets, and the course aims and objectives have been defined. After the teaching course three research methods are used (a- written test for collecting quantitative data), (b- questionnaire for collecting quantitative data) and (c- semi-structured interview for collecting qualitative data). The semi-structured interview and the questionnaire were designed for understanding KS3 learners’ attitudes. But the written test was designed for measuring KS3 learners’ achievement. In this Chapter the data analysis methods were defined and explained and these methods will be used statistically in Chapter 6. The
Chapters of (research methodology, research design, and the empirical study) can be summed as shown in Table 22.
Table 22: Outline of the Chapters of 3,4, &5
The point Description
Research questions a. How does BBC Bitesize affect the KS3 learners’ attitudes towards Online learning methods?
b. How does BBC Bitesize (Online learning source) affect the KS3 learners’ achievement?
c. How do the KS3 learners’ attitudes affect their achievement?
The philosophy of research Pragmatism paradigm (Positivism & Interpretivism)
Data Collection methods a. Semi-structured interview. b. A questionnaire.
c. A written test.
Data Analysis methods a- Quantitative data will be analysed by SPSS (program) & Smart-PLS program and some data analysis methods will be used such as:
Cronbach's Alpha, KMO, Skewness & Kurtosis, Kolmogorov- Smirnov test, the outliers’ tests, the Mean & Median, Spearman correlation, Cohen’s criteria, and Fisher’s exact test.
b- The findings will be developed via using Smart-PLS statistical software and some methods will be used such as:
PLS-SEM (Partial Least Squares- Structural Equation Modelling).
c- Qualitative data will be analysed and interpreted by the researcher.