This study focuses on the non-academic measures of educational effectiveness. It is intended to find out how educational effectiveness can be measured with non-academic scales.
The data used in this article is a subset of the data collected from 8500 students focussing on schools participating in an initiative known as the Primary Social and Emotional Aspects of Learning (SEAL) programmes (DfES 2005; DfES 2006). The data used in the article relate to children in school years 1 to 6, who are aged 5 to 11.
The studens from whom that the data used in this research are from the schools that had been participating in the Primary Social and Emotional Aspects of Learning programmes. The elements of the questionnaire used to gather data for the survey were taken from a pre-existing questionnaire called ‘About me and my school’ that was used as a part of an evaluation of Primary SEAL which is a government funded evaluation.
The data were collected by the aid of an online questionnaire tool with groups of children usually completing the survey as a class group in their school compute rooms. The youngest children and others who may have had difficulty accessing the wording of the questionnaire items were supported in completing the questionnaire by teaching assistants. The guidance of the teaching assistants shows that the data used in the research is relaible because children were provided with help to complete the survey to get valid results.
The process used during data collection is known as confirmatory factor analysis (CFA) which proposes a model to that links the student responses to groups of items to a common underlying dimension, such as self-image. It then tests this model against the data (in our case from a range of schools that have just started SEAL to provide a baseline) and reports how well the proposed model fits the data from the students in those pre-SEAL schools. According to the CFA model, the fit for a model of complexity is only acceptable whtn CFI>0.95 and RMSEA<0.05 (Hu, Bentler 1999). The CFI value for the survey is 0.958 and RMSEA is 0.036.
So, the data used in the research can be considered as acceptable according to this criterion. According to the Raykov’s scale reliability coefficient (Raykov 2001) the generally accepted threshold value for a scale reliability is the same, namely greater than a value of 0.7. Five of the dimesions of the survey is greater than 0.7 and only two are less than 0.7 and they both have values of 0.68. So, the data of the survey may be considered reliable.
In this article, central tendency will be used as descriptive statistics. (Wert, Neidt & Charles 1954) says in their book “Since the human mind is unable to comprehend readily the significance of unorganized masses of data, it is helpful to characterize or to describe the data in terms of one or more summary values which stand in place of the numerous individual values represented by the data”. The reason of choosing central tendency as descriptive statistics is that the survey consists of lots of numerical data and I aim to come up with a statistical measure that has the ability to describe the center of the measurements and that can represent the whole distribution of data.
To sum up, central tendency is used in this study to identify a single value as a representative of whole data set. Apart from central tendency, I plan to use Ordinary Least Squares Regression Analysis in this study. “Regression analysis is a statistical technique for investigating and modelling the relationship between variables” (Douglas et al. 2012). The SEAL survey’s aim is to identify the relationship between variables, so regression analysis will fit the aim of the survey.
There are 4 assumptions for Ornidary Least Squares Regression. The first one is that the OLS regression errors should be equal to zero on average, which is satisfied in the SEAL survey. Secondly, the observations should not be chosen systematically in a biased way and that is also satisfied as there is no systematical choice of data. Thirdly, there should not be any large outliers as in the data that we are using in this study, in SEAL survey items that aren’t included in the summary model are those representing extremes of behaviour. Lastly, there should not be any perfect multicollinearity and in our study there is no multicollinearity since none of the independent variables is a linear combination of the others.
Findings About Educational Effectiveness
The data gathered by the SEAL student survey is analysed and interpreted by using MS Excel 2016. The data of the SEAL student survey has 7 dimensions in total. The first dimension. There are three dimensions potentially linked to the SEAL themes of self-awareness and managing feelings and these dimensions are Self Image (SI), Managing Feelings (MF), Managing Behavior (in Class) (MB). There are two Dimensions potentially linked to the SEAL theme of motivation and these dimensions are Independence (In Class Work) (Ind) and Resilience (Both work and relationship components) (Resil). Lastly, there are two dimensions potentially linked to the SEAL themes of empathy and social skills and these are Friendship (Friend) and Attitudes to Teachers and School (AttT&S).
Firstly, in this study I am going to examine the descriptive statistics that was indicated as central tendency in the earlier parts of the study. The first central tendency measure that will be examined is mean. Mean is the “Simple or arithmetic average of a range of values or quantities, computed by dividing the total of all values by the number of values” (Dixon et al. 1957).
The means of all the dimensions are going to be examined one by one. The mean of Self Image is 80%, mean of Managing Feelings is 67%, mean of Managing Behavior is 70%, mean of Independence is 76%, the mean of Resiliance is 43%, mean of Friendship is 76% and lastly the mean of Attitudes to Teachers and School is 74%. These values of mean imply that on average for the 80% of the students, self image is a measure of educational effectivenes, for 67%, managing feelings is a measure of education, for 70% managing behavior is a measure, for 76% independence, for 43% resiliance, for 76% friendship and lastly for 74% attitudes to Teachers and School is a measure of educational effectiveness.
Another part of central tendency is median. Median is known as the middle number in a set of data. The median of Self Image, Managing Feelings, Managing Behavior, Independence, Resiliance, Friendship and lastly Attitudes towards Teachers and School are 88%, 69%, 72%, 75%, 42%, 76% and 76% respectively. These values imply that the centers of the data related to each of these dimensions are as shown. For instance, center for the Self Image dimension is 88% whereas the center for Managing Feeelings dimension is 69%.
The centers of different dimensions differ highly in our data set. The highest centered dimension is Self Image, whereas the lowest centered dataset is Resiliance. Even we have calculated the median scores for the data of the SEAL student survey, median is not as useful as mean as an element of central tendency because our dataset is not very skewed, which means our data set do not have too much extreme values as stated in the earlier parts of the study. So, as a result mean is a better central tendency measure for our study.
Lastly, another measure of central tendency is mode. Mode is the mostly occuring score in the data set. In our study mode is considered as a measure that is supplemental and since the mode will not give us any crutial information about our data set, it is not preferred to deal with mode in this study.
As stated earlier, apart from central tendency, this study is going to analyze the data from the SEAL student survey by the aid of linear regression model. “Regression is the study of dependence. It is used to answer questions such as does changing class size affect the success of the students?” (Weisberg 2005). In this study, linear regression model will be applied to all the dimensions seperately. As a result of computations, it is found out that the graph of the affect of Self Image on education has a y-intercept of 22.3.
This is the average value of Self Image when the dependent variable is 0. The variable coefficient for SI is 1.5, at this coefficient implies that increasing the self image by 1 unit increases its effect on education by 1.5 units. So, this gives us a clue that self image of the students have a respectable effect on education of the students. When we apply the same procedure on menaging behavior, we find a greater value of variable coefficient and this implies that managing behavior has a greater effect on the dependent variable when it is increased by one unit compared to increasing self image by one unit and its effect on the dependent variable.
When all the variables, so called ‘dimensions’, are analyzed one by one, we see that the variable coeffecient of resilience is the greatest. By using this information, we can conclude that increasing resiliance of the students by one unit has the greatest impact on education compared to increasing the other variables by one unit and their corresponding effect on the dependent variable. As an overall result, we can say that resiliance has a greater effect on education compared to other 6 dimensions that we examined by using the data from the SEAL student survey.
Discussion on Educational Awareness
To conclude, this study used a portion of the data collected from 8500 students which focuses on schools participating in an initiative known as the Primary Social and Emotional Aspects of Learning (SEAL) programmes (DfES 2005; DfES 2006). The data used is gathered from students who are between first graders to six graders and those students are aged from 5 to 11 years old. An online tool was used in order to obtain data from the primary school students. This is the most suitable way in order to get as much data as possible and in order to reach the greatest number of students.
In this way, the data is accepted to give better results as more students would give more accurate results. The accuracy and the reliability of the data were ensured by using different scientific ways. The first way used to ensure the acceptability of the data is CFA, which stands for confirmatory factor analysis which links the student responses to groups of items to a common underlying dimension and then tests the model against the data. According to this criterion, the data used in this study is in the expected CFI and RMSEA range so the data may be considered reliable.
After testing the reliability of the data, the data was then analyzed by using central tendency. Mean and median for central tendency were considered as measures of analysis and mode was not used because of the reasons explained above. All the means and medians of the dimensions of the study were measured and averages and the center values were found as a result of these computations. After analyzing the central tendency of the data, the study continued with the data analysis by using linear regression model. By the linear regression model, the aim was to see the relationship between dependent variable and the independent variables.
All the independent variables were used seperately in order to come up with seperate linear regression equations and in order to examine all the dimensions of the data seperately to see the effects of these dimensions on education seperately. After comparing all the data, it is seen that increasing the resiliance of the students by one unit has the greatest effect on education compared to increasing all the other six independent variables, dimensions, by one unit.
DfES, 2005. Primary National Strategy Excellence and Enjoyment: social and emotional aspects of learning guidance. London: Department for Education and Skills.
DfES, 2006. Primary national strategy Excellence and Enjoyment: social and emotional aspects of learning Family SEAL.
Hu, L., 1999. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives.
Raykov, T., 2001. Estimation of congeneric scale reliability using covariance structure analysis with nonlinear constraints. British Journal of Mathematical and Statistical Psychology.
Weisberg, S., 2005. Applied Linear Regression, Third Edition. John Wiley & Sons Inc..
Wert, J. E. N. C. O., 1954. Statistical methods in educational and psychological research. East Norwalk: Appleton-Century-Crofts.