Comparative robusity of Analysis of Variance and Regression Statistics using the Assumption of Sample Size, Homogeneity of Variance and Normality of the Distribution

Aniyom Asuquo Obambon & Dr Isaac Ofem Ubi


The violation of statistical assumptions in testing hypotheses has been a problem confronting most researchers in different fields of endeavour. This study was principally carried out to compare the robustness of Analysis of Variance and Regression Statistics using the assumptions of sample size, homogeneity of variance, and normality of the distribution. The study area was Southern Cross River State, Nigeria. Cluster and simple random sampling techniques were adopted in selecting a sample of 640 students from a student population of 4,265. Two instruments were used for data collection. The first was a questionnaire called Students’ Attitude and Test Anxiety Questionnaire (SAATAQ), while the second was scores from past examination scripts of WAEC May/June SSSCE for 2015/2016 (multiple choice test only). Cronbach Apha reliability estimate method was adopted to test the reliability of the instruments. The researchers tried, as much as they could, to meet the three assumptions of Analysis of Variance and Regression Statistics chosen for the study. Results of the study indicate that the F-ratios of both test statistics were robust as their p-values were .000 each. The study recommends strict adherence to the three assumptions by researchers when Analysis of variance and Regression statistics are proposed for a study.

Read Full Article

Scroll to Top