Building Statistical Capacity of Undergraduates in Experimental Design and Analysis with R
64th ISI World Statistics Congress - Ottawa, Canada
Format: IPS Abstract
Session: IPS 182 - Impact of Statistical Computing Techniques on Statistical Capacity Building in Africa
Monday 17 July 10 a.m. - noon (Canada/Eastern)
Building Statistical Capacity in Experimental Design and Analysis in Undergraduates with R
Authors: 1) Mbachu Hope I. firstname.lastname@example.org +2348034732597 Department of Statistics, Imo State University Owerri Imo State Nigeria; 2) Adenomon Monday O. email@example.com
The ability to design and analyze experiments is a critical skill for undergraduate students in many fields, from science to social science. However, many students lack sufficient statistical knowledge to properly conduct and interpret statistical analyses. To address this issue, this paper proposes an approach to building the statistical capacity of undergraduate students in experimental design and analysis using the statistical software R. The proposed approach included performance of post-hoc comparison tests ; Tukey test, Duncan range test, and Least Significant Difference test (LSD) in R. We first load a dataset and performed ANOVA using the aov() function in each case. Then, we performed Tukey's test using the TukeyHSD() function. It showed the difference in means between each pair of groups, along with the p-values and confidence intervals. We also performed Duncan's range test using the duncan.test() function. The output showed the means of each group and the range between the highest and lowest means. The LSD test was done using the lsd.test() function from agricolae package. It showed the means of each group and the LSD value. These tests in R showed how to determine if there are significant differences between groups and which groups are significantly different. In one-way, two-way factorial design, we first load a dataset and performed a one-way ANOVA using the aov() function. It showed the F-value, degrees of freedom, and p-value, to determine if there is a significant difference between the groups. Two-way factorial design: we loaded a dataset and created a linear model using the lm() function. A two-way ANOVA using the Anova() function from the car package was performed. The output showed main effects of each factor, the interaction effect, and the F-value, degrees of freedom, and p-value for each effect. The results of these analyses showed if there were significant differences between the groups and an interaction effect between the factors. In Levene's test for homogeneity of variance in R, the leveneTest() function from the car package was used which showed the test statistic, degree of freedom and p-value. The null hypothesis of Levene's test is that the variances are equal across all groups. Otherwise, we reject the null hypothesis and conclude that there is evidence of unequal variances. The result of Levene's test can help determine whether to use a parametric or non-parametric test. If the variances are equal, a parametric test such as ANOVA can be used. If the variances are unequal, a non-parametric test such as the Kruskal-Wallis test can be used. The results suggest that the approach was successful in improving students' statistical knowledge and skills, as demonstrated by their performance on pre- and post-course assessments and their ability to design and analyze experiments using R. KEYWORDS: Design, analysis, ANOVA,test, Software Package R.