Page 196 - Šolsko polje, XXVIII, 2017, no. 5-6: Znanje, motivacija in pogoji učenja v luči mednarodnih primerjav TIMSS in PISA, ur. Barbara Japelj Pavešić in Klaudija Šterman Ivančič
P. 196
šolsko polje, letnik xxviii, številka 5–6
which enabled calculations of population estimates and standard errors
with the use of suitable sample weights and all ten plausible values of liter-
acy achievement in the PISA database.
Results with discussion
Analyses were conducted using least-squares regression on four sets of
models of predictors for reading, mathematical and scientific literacy
achievement. The predictors in models are grouped according to their
content: psychological, social, physical wellbeing and school experiences.
Socio-economic and cultural status was added to all models in order to
control for its well-known high impact on achievement. Sets of predictors
in the models are presented together with the results of analyses. Testing
of significance is carried out at 0.05 level.
The first set of predictors in the regression models for reading, math-
ematics and science achievement comprised of the constructs of persever-
ance, engagement, empathy, happiness, optimism, pro-social behavior and
sadness or worrying.
With these predictors, 15 percent of variance in reading, 13 percent
in mathematical and 16 percent of variance in scientific literacy was ex-
plained. In all three models, the strongest predictor shows to be socio-
economic and cultural status. It is well known that this is an important
predictor of achievement in literacy domains (OECD, 2017). For reading
literacy, the regression coefficient of this predictor is 29.07 scale-points.
This indicates that between two groups of students, having a one stan-
dard deviation difference of socio-economic and cultural status and being
similar on other predictors, there is an average difference of 29.07 scale-
points in reading literacy achievement; the group with higher socio-eco-
nomic and cultural status having higher achievement in literacy domains.
Furthermore, the value of standardized regression coefficient, beta, shows
that this difference is approximately a third of the standard deviation in
the reading literacy scores in the student population.
Further on empathy was shown to be a significant and positive pre-
dictor for achievement in all three literacy domains. Since the values of
predictors were standardized, the value of regression coefficient 15.90 for
empathy in the model for reading literacy means that for students, whose
empathy scale values are one standard deviation apart while having sim-
ilar values of other predictors, reading literacy achievements, on aver-
age, differ 15.90 scale-points; students with higher empathy scale-values
also having, on average, higher achievement. The value of standardized
regression coefficient shows that this accounts for approximately 18 per-
cent of the standard deviation in reading literacy scores in the student
194
which enabled calculations of population estimates and standard errors
with the use of suitable sample weights and all ten plausible values of liter-
acy achievement in the PISA database.
Results with discussion
Analyses were conducted using least-squares regression on four sets of
models of predictors for reading, mathematical and scientific literacy
achievement. The predictors in models are grouped according to their
content: psychological, social, physical wellbeing and school experiences.
Socio-economic and cultural status was added to all models in order to
control for its well-known high impact on achievement. Sets of predictors
in the models are presented together with the results of analyses. Testing
of significance is carried out at 0.05 level.
The first set of predictors in the regression models for reading, math-
ematics and science achievement comprised of the constructs of persever-
ance, engagement, empathy, happiness, optimism, pro-social behavior and
sadness or worrying.
With these predictors, 15 percent of variance in reading, 13 percent
in mathematical and 16 percent of variance in scientific literacy was ex-
plained. In all three models, the strongest predictor shows to be socio-
economic and cultural status. It is well known that this is an important
predictor of achievement in literacy domains (OECD, 2017). For reading
literacy, the regression coefficient of this predictor is 29.07 scale-points.
This indicates that between two groups of students, having a one stan-
dard deviation difference of socio-economic and cultural status and being
similar on other predictors, there is an average difference of 29.07 scale-
points in reading literacy achievement; the group with higher socio-eco-
nomic and cultural status having higher achievement in literacy domains.
Furthermore, the value of standardized regression coefficient, beta, shows
that this difference is approximately a third of the standard deviation in
the reading literacy scores in the student population.
Further on empathy was shown to be a significant and positive pre-
dictor for achievement in all three literacy domains. Since the values of
predictors were standardized, the value of regression coefficient 15.90 for
empathy in the model for reading literacy means that for students, whose
empathy scale values are one standard deviation apart while having sim-
ilar values of other predictors, reading literacy achievements, on aver-
age, differ 15.90 scale-points; students with higher empathy scale-values
also having, on average, higher achievement. The value of standardized
regression coefficient shows that this accounts for approximately 18 per-
cent of the standard deviation in reading literacy scores in the student
194
