Information Technology Reference
In-Depth Information
2
The SEM results produced good fits to data (Latino/a group—
χ
=
2,498.05,
2
/
df
df
=
1,322,
p
=
0.000,
χ
=
1.90, CFI
=
0.92, TLI
=
0.92, RMSEA
=
0.03;
2
2
/
df
White—
χ
=
2,540.59,
df
=
1,322,
p
=
0.000,
χ
=
1.92, CFI
=
0.94, TLI
=
0.93, RMSEA
0.03).
For Latino/a students, the overall relationships among opportunities to learn,
prior level of achievement, and motivation factors were similar to those of all
students (e.g., positive effects of motivation factors—efficacy, mastery goal—on
learning strategies). However, unlike the overall SEM, the Latino subgroup did
not show any significant effect of prior knowledge on performance avoidance,
although they showed a positive effect of OTL on performance approach (
=
0.12,
p
< 0.001). We also found that male Latino students reported more academic effi-
cacy than female Latino students (
β
=
0.14,
p
< 0.001), as expected in findings
from the studies on gender and self-efficacy. Efficacy and mastery influenced learn-
ing strategies positively (
β
=
β
=
β
=
0.39 and
0.53, respectively,
p
< 0.001) but efficacy
β
=−
=
had negatively influenced help-seeking (
0.34,
p
0.001). Maladaptive help-
seeking strategies were positively associated with worry (
0.30,
p
< 0.001). The
pattern of effects on GPA (positive effects of prior knowledge, negative effects of
worry, and help-seeking) was similar to those of the overall model.
For White students, the results showed different patterns in relations of SES to
prior knowledge and OTL. White female students achieved more in high school than
their male peers (
β
=
β
=−
0.14,
p<
0.001) and students who had more OTL reported
higher self-efficacy (
0.09,
p<
0.001). Also, there was no effect of OTL on
mastery goal orientation, whereas there was a positive effect of prior knowledge on
performance avoidance (
β
=
0.19,
p<
0.001). Self-efficacy and mastery goal orien-
tation positively affected learning strategies (
β
=
0.65, respectively,
p<
0.001). Like Latino students, both maladaptive help-seeking and perfor-
mance avoidance positively affected worry (
β
=
0.33 and
β
=
0.11, respec-
tively,
p<
0.001) and worry had a negative effect on achievement (
β
=
0.24 and
β
=
0.25,
p<
0.001). Prior knowledge and learning strategies were found to be two promi-
nent predictors of academic achievement (
β
=−
β
=
0.53 and
β
=
0.21, respectively,
p<
0.001).
Multiple-group SEM: White and Latino groups
. Finally, we analyzed a multi-
group SEM for White and Latino groups to examine the extent to which parameter
estimates in our theoretical model differ between White and Latino groups (See
Fig. 7.3). We first estimated the constrained model, which fixed all common param-
eters between groups. Our constrained model resulted in a CFI of 0.91 and TLI of
0.90 with
2
2
/
df
2.19,
p
< 0.000). We then performed
chi-square difference tests several times, in order to identify which pairs of path
coefficients were significantly different in multiple groups. For example, we freed
the parameter of the path from gender to prior knowledge to test whether there was
a significant chi-square decrease. The model improved significantly (
χ
=
6,083.70 (
df
=
2,773,
χ
=
2
χ
=
82.11,
1,
p
< 0.001) which means that there was a significant difference in parameter
estimates between two groups.
In other words, although both of the standardized path coefficients were negative,
the effect of gender on prior knowledge was slightly greater for Latino students
df
=
