Page 88 - Šolsko polje, XXX, 2019, št. 5-6: Civic, citizenship and rhetorical education in a rapidly changing world, eds. Janja Žmavc and Plamen Mirazchiyski
P. 88
šolsko polje, letnik xxx, številka 5–6
The relationship between the bullying at school continuous scale
and the other contextual continuous scales was tested using Pearson
product-moment correlation. The report presents only the coefficients for
those contextual variables which have a statistically significant correlation
with the scale on being bullied.
Besides testing which background and contextual variables are relat-
ed to school bullying, it is also important to test which of the contextu-
al and background characteristics are related to each other. This could in-
dicate which of these variables (1) could have a combined effect on school
bullying; and (2) this information could also identify the areas where the
educational system can intervene.
This second part of the analysis uses all the variables identified from
the first part. However, the number of variables have to be reduced for the
following reasons:
1. Some of the background and contextual variables form groups un-
der a single theme (e.g. student perception on the openness of class-
room discussion) and are expected to correlate between each other.
2. Many of the variables mentioned in the previous point are part of
complex scales and, of course, correlate with them because the com-
plex scales contain the information from these variables.
Thus, the variables which fall within the previous two points were
identified and removed in advance from the analyses of the correlations.
This way, only variables belonging to different dimensions were used in
the analyses of the correlations. It is expected that some correlations will
exist between many of the identified variables because many of them are
related with the overall context of teaching and learning. Thus, only pairs
of variables for which a correlation of at least 0.50 exists were kept in the
final results. This way, only the most highly correlated pairs of variables,
i.e. the most important ones, were kept. The results are presented in the
“Results” section.
The third part of the analysis includes multiple linear regression. The
original plan for the multivariate analysis was to use a regression mod-
el where the association between a student being bullied at school (de-
pendent variable) and student achievement as its predictor (reading or
civic knowledge) when controlling for multiple other variables like SES,
school environment security, and student behavioral issues among oth-
er background and contextual variables from the student, teacher and
school variables. However, although statistically significant, the relation-
ship between bullying and student achievement have shown to be very
weak, it is close to zero. Thus, the analysis was changed to test the associ-
86
The relationship between the bullying at school continuous scale
and the other contextual continuous scales was tested using Pearson
product-moment correlation. The report presents only the coefficients for
those contextual variables which have a statistically significant correlation
with the scale on being bullied.
Besides testing which background and contextual variables are relat-
ed to school bullying, it is also important to test which of the contextu-
al and background characteristics are related to each other. This could in-
dicate which of these variables (1) could have a combined effect on school
bullying; and (2) this information could also identify the areas where the
educational system can intervene.
This second part of the analysis uses all the variables identified from
the first part. However, the number of variables have to be reduced for the
following reasons:
1. Some of the background and contextual variables form groups un-
der a single theme (e.g. student perception on the openness of class-
room discussion) and are expected to correlate between each other.
2. Many of the variables mentioned in the previous point are part of
complex scales and, of course, correlate with them because the com-
plex scales contain the information from these variables.
Thus, the variables which fall within the previous two points were
identified and removed in advance from the analyses of the correlations.
This way, only variables belonging to different dimensions were used in
the analyses of the correlations. It is expected that some correlations will
exist between many of the identified variables because many of them are
related with the overall context of teaching and learning. Thus, only pairs
of variables for which a correlation of at least 0.50 exists were kept in the
final results. This way, only the most highly correlated pairs of variables,
i.e. the most important ones, were kept. The results are presented in the
“Results” section.
The third part of the analysis includes multiple linear regression. The
original plan for the multivariate analysis was to use a regression mod-
el where the association between a student being bullied at school (de-
pendent variable) and student achievement as its predictor (reading or
civic knowledge) when controlling for multiple other variables like SES,
school environment security, and student behavioral issues among oth-
er background and contextual variables from the student, teacher and
school variables. However, although statistically significant, the relation-
ship between bullying and student achievement have shown to be very
weak, it is close to zero. Thus, the analysis was changed to test the associ-
86
