A Robust Alternative to the t - Test

Yusof, Zahayu Md; Abdullah, Shahrum; Yahaya, Sharipah Soaad Syed; Othman, Abdul Rahim

doi:10.5539/mas.v6n5p27

Cited by 2 publications

(2 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the test does not perform well when normality assumption is violated and worsens when heterogeneity of variance and sample size exist [1]. Violation from these assumption will eventually lead to inflation of Type I error rate and depression in statistical power [2][3] [4][5] [6]. As alternative, nonparametric test could be the choice when the issue of violation, however, this test seems to be less powerful and the sample size need to be large enough in order to achieve reliable results [7].…”

Section: Introductionmentioning

confidence: 99%

“…Violation from these assumption will eventually lead to inflation of Type I error rate and depression in statistical power [2][3] [4][5] [6]. As alternative, nonparametric test could be the choice when the issue of violation, however, this test seems to be less powerful and the sample size need to be large enough in order to achieve reliable results [7]. To alleviate the aforementioned problems, robust statistical methods are recommended.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

H-statistic with winsorized modified one-step M-estimator for two independent groups design

Xao¹,

Yahaya²,

Abdullah³

et al. 2014

AIP Conference Proceedings

Self Cite

View full text Add to dashboard Cite

Two-sample independent t-test is a classical method which is widely used to test the equality of two groups. However, this test is easily affected by any deviation in normality, more obvious when heterogeneity of variances and group sizes exist. It is well known that the violation in the assumption of these tests will lead to inflation in Type I error rate and depression in statistical test power. In mitigating the problem, robust methods can be used as alternatives. One such method is H-statistic. When used with modified one-step M-estimator (MOM), this test statistic (MOM-H) produce good control of Type I error even under small sample size but inconsistent across certain conditions investigated. Furthermore, power of the test is low which might be due to the trimming process. In this study, MOM is winsorized (WMOM) to sustain the original sample size. The H-statistic with WMOM as the central tendency measures (denoted as WMOM-H) showed better control of Type I error as compared to MOM-H especially under balance design regardless of the shapes of distribution investigated in the study. It also performed well under highly skewed and heavy tailed distribution for unbalanced design. In general, this study demonstrated that winsorization process (WMOM) could improve the performance of H-statistic in terms of Type I error rate control.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

H-statistic with winsorized modified one-step M-estimator for two independent groups design

Xao¹,

Yahaya²,

Abdullah³

et al. 2014

AIP Conference Proceedings

Self Cite

View full text Add to dashboard Cite

show abstract

Videos with Hands: an Analysis of Usage and Interactions of Undergraduate Science Students for Acquiring Physics Knowledge

García

Conesa²,

Pérez-Navarro³

2022

J Sci Educ Technol

View full text Add to dashboard Cite

Videos created with the hands of teachers filmed have been perceived as useful educational resource for students of Physics in undergraduate courses. In previous works, we analyzed the students’ perception about educational videos by asking them about their experiences. In this work, we analyze the same facts, but from a learning analytics perspective, by analyzing the interactions that students have with the videos during their learning experience. With this analysis, we obtain how students behave and may compare whether their behavior aligns with the perceptions obtained from previous research. The data analyzed in this work corresponds to the students’ interactions with educational videos during 5 semesters in two different courses of Physics within online degrees of Telecommunication and Computer Science. It has been found that the topic taught in the videos has influence in the way videos are used by the students. Regarding the type of content (theory or problem-solving), problem-solving videos are more used by students, although interactions with both videos are similar. This difference differs with previous results based on students’ perception. The contribution of the paper is to provide more ground and knowledge about the way the educational videos are consumed in Physics courses. The new knowledge can be used to improve the way videos are incorporated within courses and, therefore, to improve the student learning experiences.

show abstract

A Robust Alternative to the t - Test

Cited by 2 publications

References 18 publications

H-statistic with winsorized modified one-step M-estimator for two independent groups design

H-statistic with winsorized modified one-step M-estimator for two independent groups design

Videos with Hands: an Analysis of Usage and Interactions of Undergraduate Science Students for Acquiring Physics Knowledge

Contact Info

Product

Resources

About