Page 65 - Štremfel, Urška, ed., 2016. Student (Under)achievement: Perspectives, Approaches, Challenges. Ljubljana: Pedagoški inštitut. Digital Library, Documenta 11.
P. 65
low, basic, and higher-achievement groups. Since standardised values are 65
the basis, the means of all indices in the entire target population equal 0, and
comparisons of means of indices by groups show differences between these
groups. The significance of differences was tested by means of the t-test be-
tween individual pairs.
The second approach is a correlation analysis between indices and read-
ing achievement for the total population by gender, and for individual groups
by gender and by achievement. The correlation indicates the direction and
strength of the association between the factor and achievement, which is
used to determine whether a factor is significant for reading development.
However, caution is needed in making interpretations, as this is not neces-
sarily a direct causal relationship; for instance, a strong correlation may orig-
inate from a third factor, which, in the background is correlated with both the
discussed index and achievement, or this may be a case of reverse causality
where achievement impacts the factor. At the same time, correlations are bi-
variate, which neither gives a picture about the correlations between the fac-
tors themselves, nor consequently about partial correlations between individ-
ual factors and achievement in reading with others being controlled. Some of
the factors discussed are relatively highly correlated (e.g. the bivariate corre-
lation between the index of usefulness of strategies for writing a summary and
the index of usefulness of strategies for understanding and memorising a text is
0.46), which means it is not possible to make a clear distinction when it comes
to the correlations between an individual factor and reading achievement. An
analysis of the correlation, or the effect, of individual factors and control of
others could be performed by means of a regression analysis or use of struc-
tured models. However, stability of various models would have to be verified
by accounting for all, or a smaller number of, factors with a different sequence
of the analysis of the effect of one factor and control of previous ones. How-
ever, this would be too demanding for the scope of this paper. The correlation
analysis will, in spite of the aforementioned limitations, suffice for discussion
on the question about possible differences in the associations between factors
and achievement in reading by gender and achievement groups.
It should also be pointed out, and taken into consideration when inter-
preting the results, that the indices have been developed from students’ an-
swers to questions in the questionnaire and not from any independent ob-
servations or any other types of measurements. This means that the answers
and thereby also observations about correlations with achievement depend
on the way students understand and then answer the questions.
low reading achievement in pisa 2009
the basis, the means of all indices in the entire target population equal 0, and
comparisons of means of indices by groups show differences between these
groups. The significance of differences was tested by means of the t-test be-
tween individual pairs.
The second approach is a correlation analysis between indices and read-
ing achievement for the total population by gender, and for individual groups
by gender and by achievement. The correlation indicates the direction and
strength of the association between the factor and achievement, which is
used to determine whether a factor is significant for reading development.
However, caution is needed in making interpretations, as this is not neces-
sarily a direct causal relationship; for instance, a strong correlation may orig-
inate from a third factor, which, in the background is correlated with both the
discussed index and achievement, or this may be a case of reverse causality
where achievement impacts the factor. At the same time, correlations are bi-
variate, which neither gives a picture about the correlations between the fac-
tors themselves, nor consequently about partial correlations between individ-
ual factors and achievement in reading with others being controlled. Some of
the factors discussed are relatively highly correlated (e.g. the bivariate corre-
lation between the index of usefulness of strategies for writing a summary and
the index of usefulness of strategies for understanding and memorising a text is
0.46), which means it is not possible to make a clear distinction when it comes
to the correlations between an individual factor and reading achievement. An
analysis of the correlation, or the effect, of individual factors and control of
others could be performed by means of a regression analysis or use of struc-
tured models. However, stability of various models would have to be verified
by accounting for all, or a smaller number of, factors with a different sequence
of the analysis of the effect of one factor and control of previous ones. How-
ever, this would be too demanding for the scope of this paper. The correlation
analysis will, in spite of the aforementioned limitations, suffice for discussion
on the question about possible differences in the associations between factors
and achievement in reading by gender and achievement groups.
It should also be pointed out, and taken into consideration when inter-
preting the results, that the indices have been developed from students’ an-
swers to questions in the questionnaire and not from any independent ob-
servations or any other types of measurements. This means that the answers
and thereby also observations about correlations with achievement depend
on the way students understand and then answer the questions.
low reading achievement in pisa 2009
