Page 60 - Azerbaijan State University of Economics

THE JOURNAL OF ECONOMIC SCIENCES: THEORY AND PRACTICE, V.76, # 1, 2019, pp. 48-69

Table 7: Rosenthal’s Fail Safe N for publication partiality of the studies whose
effect sizes are calculated based on the relationship between school effectiveness
and leadership and school climate

Independent Number of studies required for
variable k Z p Α Zα missing studies to reach alpha
value
Leadership 11 43.70 0.000 0.05 1.96 5458
School climate
6 19.89 0.000 0.05 1.96 612

Rosenthal’s Fail Safe N analysis data show that for the effect size calculated for the
study to go back to the alpha level, 5458 studies which have zero effect size are
needed. According to this analysis, in order to be able to make an evaluation
expressing that there is no publication partiality the number of missing studies
should be 10 times more than 5 times the number of total number of studies (Ren,
Lu and Fu, 2016). According to this, since the number of missing studies is much
higher than 65 (5x11+5=65) in the leadership independent variable and much higher
than 40 (5x6+10) in the school climate independent variable, it can be stated that
there is no publication partiality according to the Rosenthal’s Fail Safe N analysis.

Table 8: Orwin's Fail Safe N for Publication Partiality Analysis of the Studies
for which Effect Sizes have been calculated based on the relationship between
school effectiveness and leadership and school climate

Independent Variable
Criteria Leadership School
Climate
The Fisher's Z Value for the observed studies 0.631 0.495
Fisher’s Z for unimportant Criteria 0.05 0.05
Fisher’s Z Average in Missing Studies 0.000 0.000
Number of studies needed to bring Fisher’s Z Value 128 54
below 0.05

Fisher’s Z value for studies observed in line with the results of Orwin’s Fail Safe N
analysis was determined as 0.631 for the leadership independent variable and as
0.495 for the school climate independent variable. The results show that, fort he
calculated effect size value to reach the 0.00 level, 128 studies leadership and 54
studies with an effect size of 0.05 for are needed. In this respect, according to
Orwin’s Fail Safe N analysis results, it has been decided that there is no publication
partiality since N>5k+10.
60