Journal of Al-Quds Open University for Educational & Psychological Research & Studies
Abstract
The study aimed at identifying outlier values in Jordanian eighth graders scores in TIMSS international tests (TIMSS, 2011) and the effect of its treatment method on statistical results. For this purpose, a sample of 100 students was selected for each of math and science subjects. The Box plot Method detected two outlier values in math and seven values in science. The study investigated the effect of outliers treatment method on statistical results of ANOVA, independent samples t-test and analysis of regression. The results revealed that ANOVA results do not vary due to outliers treatment method. Practical significance and test power were higher in the case with outliers than in the case of deleting or replacing them. On the other hand, the t-test results differ depending on outliers treatment method. Practical significance and test power were higher in the case of deleting or replacing outliers than in the case of keeping them. With regression analysis, it was shown that the simple and multivariate linear regression models are different due to outliers treatment method. In case of keeping outliers, chemistry score was the main predictor of mathematics score. While biology score was the main predictor of mathematics score, in the case of deleting outliers or replacing them.
Recommended Citation
Qwaider, Omar Qasim and Al Sawalmeh, Yousuf Mohammed
(2017)
"Outliers in Jordanian Students Performance on (TIMSS) Test in Mathematics and Science and the Effect of its Treatment method on Statistical Results,"
Journal of Al-Quds Open University for Educational & Psychological Research & Studies: Vol. 6:
No.
20, Article 2.
Available at:
https://digitalcommons.aaru.edu.jo/jaqou_edpsych/vol6/iss20/2