Page 88 - Ana Kozina and Nora Wiium, eds. ▪︎ Positive Youth Development in Contexts. Ljubljana: Educational Research Institute, 2021. Digital Library, Dissertationes (Scientific Monographs), 42.
P. 88
positive youth development in contexts
constructs were examined in each group. If partial measurement invari-
ance was not achieved, some constraints were removed. A change in CFI
(equal to or less than .01) was used as an indicator of measurement invari-
ance since chi-square difference tests depend on the sample size (Cheung
& Rensvold, 2002).
Finally, MANOVAs with a Bonferroni correction were used to com-
pare groups or levels across gender and school. MANOVA was applied in-
stead of univariate ANOVAs to avoid Type I error since only one instead of
several tests are being conducted, meaning that MANOVA takes account of
the relationship between the combinations of dependent variables, which is
not possible in the case of a series of univariate ANOVAs, and MANOVA
holds greater power for detecting an effect. Since only two participants de-
fined themselves as non-binary, they were excluded from further gender
analysis.
Results
After the descriptive statistics, the reliability analyses are outlined. The re-
sults of CFA and ESEM for each PYD measure together with measurement
invariance across gender and school are then presented. At the end, differ-
ences across gender and school are described.
Descriptive statistics
Descriptive statistics and correlations for the whole sample are presented to
provide an insight into the data (see Table 1). The great majority of the vari-
ables are positively correlated, except for Confidence and Caring, which are
not correlated with each other at all.
Exploratory Structural Equation Model (ESEM) and Confirmatory
Factor Analysis (CFA)
In this section, ESEM results for the DA and CFA results for the 5Cs are
presented.
The ESEM results showed an adequate fit for DA with 8 latent constructs
(Support, Empowerment, Boundaries and expectations, Constructive
use of time, Commitment to learning, Positive values, Social competen-
cies, Positive identity): χ2(1420) = 4701.37, p <. 001, CFI = .922, RMSEA =
.034, 90% CI [.033, .035], SRMR = .023. For Support, the majority of target
loadings were fair (ps < .001), except for items that refer to support from
88
constructs were examined in each group. If partial measurement invari-
ance was not achieved, some constraints were removed. A change in CFI
(equal to or less than .01) was used as an indicator of measurement invari-
ance since chi-square difference tests depend on the sample size (Cheung
& Rensvold, 2002).
Finally, MANOVAs with a Bonferroni correction were used to com-
pare groups or levels across gender and school. MANOVA was applied in-
stead of univariate ANOVAs to avoid Type I error since only one instead of
several tests are being conducted, meaning that MANOVA takes account of
the relationship between the combinations of dependent variables, which is
not possible in the case of a series of univariate ANOVAs, and MANOVA
holds greater power for detecting an effect. Since only two participants de-
fined themselves as non-binary, they were excluded from further gender
analysis.
Results
After the descriptive statistics, the reliability analyses are outlined. The re-
sults of CFA and ESEM for each PYD measure together with measurement
invariance across gender and school are then presented. At the end, differ-
ences across gender and school are described.
Descriptive statistics
Descriptive statistics and correlations for the whole sample are presented to
provide an insight into the data (see Table 1). The great majority of the vari-
ables are positively correlated, except for Confidence and Caring, which are
not correlated with each other at all.
Exploratory Structural Equation Model (ESEM) and Confirmatory
Factor Analysis (CFA)
In this section, ESEM results for the DA and CFA results for the 5Cs are
presented.
The ESEM results showed an adequate fit for DA with 8 latent constructs
(Support, Empowerment, Boundaries and expectations, Constructive
use of time, Commitment to learning, Positive values, Social competen-
cies, Positive identity): χ2(1420) = 4701.37, p <. 001, CFI = .922, RMSEA =
.034, 90% CI [.033, .035], SRMR = .023. For Support, the majority of target
loadings were fair (ps < .001), except for items that refer to support from
88
