Figure (2) shows the funnel plot for precision. The Egger’s regression
test was not significant, b = -0.59, SE = 0.67, p =
.18. In addition, the Begg and Mazumdar’s correlation test was not
significant, τ = -0.02, zτ = 0.41, p = .34. These
results show that publication bias did not affect the results.
Insert Figure 2 around here
The effect size values ranged between -0.21 to 0.51. To estimate the
mean effect size, results from 20 studies (k = 105; N =
4,227) indicated that, overall, there is a significant positive
correlation between EI and academic success, r = 0.13, 95% CI
[0.08–0.27], p < .01. As expected, a high
heterogeneity was observed, Q(105) = 375.48, p <
.001, I2 = 72.04. Moderator analysis showed
that the mean effect size significantly varied by EI test, Q(3) =
42.93, p < .001, and EI subscale, Q(3) = 18.87,p = .04, while EI task nature [Q(1) = .71, p =
.40], country [Q(1)
= 3.08, p = .08] and academic performance criterion
[Q(1) = .38, p = .54] did not significantly explain
variability in the mean effect (see Table 3). Together, the EI test and
EI subscale explained 34% of variability in the mean effect. Age was
treated as a continuous variable, and the results showed that age did
not significantly explain variability in the mean effect, b =
0.011, SE = 0.007, p = .17 (see Figure 2).
Insert Table 3 and Figure 3 around here
As Table (3) shows, the EQ-i test was highly correlated with academic
success compared with other EI tests, and the perceiving emotions
subscale was highly associated with academic performance compared with
other EI subscales. Finally, the three-level multiple regression
analyses showed that Level 3 (between-studies variance) explained 29.5%
of variability in the mean effect, while Level 2 (within -studies
variance) explained 33.5%. Together, Levels 2 and 3 explained 63% of
variability in the mean effect.