Page 157 - Maša Vidmar, Vedenjske težave in učna uspešnost. Ljubljana: Pedagoški inštitut, 2017. Digitalna knjižnica, Dissertationes, 30
P. 157
summary
Results and Discussion: Concurrent and Longitudinal
Relations between Academic Achievement and Problem
Behaviour and Significant Predictors
Descriptive statistics and the correlations among the observed variables
were calculated (they are not included in the summary). Confirmatory fac-
tor analyses (CFA) and structural modelling (SEM) were carried out in or-
der to test whether the constructs (and the observed variables) were related
as predicted in the hypotheses.
The fit of CFA and SEM models was interpreted with different fit indi-
ces: CFI > 0.95, 0.05 within confidence interval (CI) of RMSEA and SRMR
< 0.05 was interpreted as good model fit; while CFI values > 0.90, 0.08 with-
in CI of RMSEA and SRMR < 0.08, were deemed satisfactory fit (Hu and
Bentler1998; Hair et al. 1998).
The following paragraphs summarise key findings in relation to the
hypotheses.
Constructs
In relation to Hypothesis 1, the anticipated structure was supported for all
examined constructs—the constructs used in structural modelling were, in
general, adequately measured by indicators. A slightly worse fit was found
for academic achievement, while one of the indicators (the puzzle box task)
was dropped due to nonsignificant loading on self-regulation.
The Relation between Problem Behaviour and Academic
Achievement and Predictors of Problem Behaviour
and Academic Achievement
Internalizing Behaviour, Academic Achievement
and other Chosen Predictors
In relation to Hypothesis 2, the SEM model showed that an absence of in-
ternalizing behaviour in first grade students significantly and positively
predicts academic achievement in the second grade (the relation was weak).
However, academic achievement in first grade students does not predict a
less frequent internalizing behaviour a year later. Model fit was acceptable
(CFI = 0.95, RMSEA (CI) = 0.08 (0.07–0.09), SRMR = 0.07). The value χ2 (84)
= 243.58, p < 0.001 was statistically significant. The results partially support
Hypothesis 2.
157
Results and Discussion: Concurrent and Longitudinal
Relations between Academic Achievement and Problem
Behaviour and Significant Predictors
Descriptive statistics and the correlations among the observed variables
were calculated (they are not included in the summary). Confirmatory fac-
tor analyses (CFA) and structural modelling (SEM) were carried out in or-
der to test whether the constructs (and the observed variables) were related
as predicted in the hypotheses.
The fit of CFA and SEM models was interpreted with different fit indi-
ces: CFI > 0.95, 0.05 within confidence interval (CI) of RMSEA and SRMR
< 0.05 was interpreted as good model fit; while CFI values > 0.90, 0.08 with-
in CI of RMSEA and SRMR < 0.08, were deemed satisfactory fit (Hu and
Bentler1998; Hair et al. 1998).
The following paragraphs summarise key findings in relation to the
hypotheses.
Constructs
In relation to Hypothesis 1, the anticipated structure was supported for all
examined constructs—the constructs used in structural modelling were, in
general, adequately measured by indicators. A slightly worse fit was found
for academic achievement, while one of the indicators (the puzzle box task)
was dropped due to nonsignificant loading on self-regulation.
The Relation between Problem Behaviour and Academic
Achievement and Predictors of Problem Behaviour
and Academic Achievement
Internalizing Behaviour, Academic Achievement
and other Chosen Predictors
In relation to Hypothesis 2, the SEM model showed that an absence of in-
ternalizing behaviour in first grade students significantly and positively
predicts academic achievement in the second grade (the relation was weak).
However, academic achievement in first grade students does not predict a
less frequent internalizing behaviour a year later. Model fit was acceptable
(CFI = 0.95, RMSEA (CI) = 0.08 (0.07–0.09), SRMR = 0.07). The value χ2 (84)
= 243.58, p < 0.001 was statistically significant. The results partially support
Hypothesis 2.
157
