The median and pseudo-sigma measures are robust for cases of non-normality
The effect of non-normality
on F-test robustness has, since the 1930s, been extensively studied under a wide variety of conditions.
Due to non-normality
and the small sample size, we used non-parametric tests like Mann-Whitney U test to compare variables between fertile and infertile groups, and Spearman's correlation to explore the possible relation among the variables.
The Jarque and Bera (1980) statistics for all variables in this study are significant at the 1 % level, indicating just such non-normality
For the explanatory analyses, the first step was the evaluation of normality of the distribution of the main dependent variable (KAP Scale), and after confirmation of non-normality
, the model of 'Generalizing Estimation Equation' (GEE) was applied.
The Kolmogorov-Smirnov normality test indicated non-normality
in both group (intervened group [I-G] and non-intervened group [NI-G], K-S test I-G = .
ANOVA using ranked data (Conover & Iman 1981) generated results identical to those obtained using raw data, indicating that the parametric analyses of raw data were sufficiently robust to accommodate the degree of non-normality
present in a few cases.
However, tests of normality using the method proposed by Looney and Gulledge (1985) indicate no evidence of non-normality
for either the low deterrence or high deterrence samples; [alpha] = .
In sections on fundamental issues, classic and emerging approaches, and applied problems they consider such topics as estimating the latent density in uni-dimensional IRT to permit non-normality
, assessing person fit in typical-response measures, an illustration of the two-tier item factor analysis model, developing item banks for patient-reported health outcomes, and using IRT to evaluate measurement invariance in health-related measures.
In the case of normal distribution of the collected data, parametric tests can be used to test the hypotheses and in case of non-normality
of the collected data nonparametric tests can be used.
The colloquium witnessed specialised brainstorming all in the social, economic and technological developmental perspective from Data to Big Data, Big Data concepts, applications, issues and tools, properties of a mixture probability model, Big data innovations and conjoint analysis, exploring Big data in social networking, process capability indices under non-normality
Key words: One-way ANOVA Outlier Non-normality
Homogeneity of variance Type-I error.