|
Estimation of effect size
in a meta-analysis of series of validity studies on
matriculation examinations in Nigeria
E. O. Adeyemo* and E. R. I. Afolabi
Department of Educational Foundations and Counseling,
Faculty of Education, Obafemi Awolowo University, Ile-Ife,
Osun State, Nigeria.
*Corresponding author. E-mail:
seyiadeyemo2007@yahoo.com.
Accepted
24 November, 2009 |
|
The study formulated an algebraic path from the series of
studies meta-analyzed on the validity of the University
Matriculation Examination (UME) in Nigeria; in order to
obtain a unique and common metrics with a view of making the
results to convey the same interpretation. The study
determined both the individual and overall effect sizes of
30 empirical studies. It also established the significant
difference, in the probability levels and the effect sizes
of the selected empirical studies. The study design is
descriptive and involved the integration of correlation
co-efficient between two variables. The measure of effect
size estimate of each of the studies was based on Pearson’s
product moment indicator using the conversion process
extracted from the works of Rosenthal (1984), Glass and
Stanley (1970). The results revealed a high effect size of
0.78 and low effect size of 0.03. The empirical studies were
not significantly different in terms of their probability
levels (x2 = 2.680, p > 0.05) but were
significantly different in terms of their effect sizes (x2
= 1444.97, p < 0.05). This study concluded that there was no
statistical linear trend in terms of effect sizes across
this set of studies and that the heterogeneity of the effect
sizes referred to fluctuation from the average of the group.
The calculated effect size did not represent adequately the
outcome of all independent study, hence an indication of
moderator variables operating within the studies.
Key
words:
Meta-analysis, empirical, effect size, validity,
matriculation, examinations. |
