Statistical testing

When attempting to identify significant correlations between variables, statistical testing is essential. The majority of the time, decision-makers in real-world situations are motivated to make choices based on the demonstrated impacts of an explanatory variable on response variables. Experience can be very beneficial in these situations, but it's crucial to base decisions on the realities of the moment because things change with time; therefore, what might have been the main cause in the past may not be the case today. Without a doubt, statistical analysis enables researchers to thoroughly explore the issue before making such crucial conclusions. However, choosing the correct statistical testing procedure can be challenging given that most of these researchers are amateurs to statistical analysis, that is, their fields of professionalism are not statistics related. The current paper presents an application of t-test family (either paired or independent t-test) to a real life problem.

The t-test family comprises of one sample, paired samples, and independent samples t-test. The most challenging thing about using t-tests is deciding whether to analyze research question using the paired or independent samples t-test. “The Effect of Gender on Language Learning” by Masoud Zoghi, Seyyed Ali Kazemi and Ali Kalani is a study that illustrates the application of t-tests in solving real life problems. Zoghi, Kazemi & Kalani (2013) identifies that gender disparity is a growing issue in pedagogical context especially in language learning. Pooling from an extensive review of the literature the study that most of the researched topics including motivation, teaching materials, classroom interaction, and teacher perceptions have reported the significant gender disparities. The research question in the study was determining whether student’s gender influences are learning English as a foreign language.

The target population was students from guidance school who were taking English at Iran Language Institute (ILI) in 2013. The students comprised those in RUN2 and RUN3 levels and samples of 100 students (50 males and 50 females) was selected at random all aged between 12 and 14 years. The sample has one-year experience of learning the English language. The English achievement test results of semester two were used for comparison purposes. The four parts tested included vocabulary, sentence function, grammar, and reading comprehension. Using the SPSS, the researcher computed the descriptive statistics (means and standard deviations) to explore the results of both tests in the two groups. The descriptive statistics were presented in different tables for RUN2 scores and RUN3 scores where each test comprised 50 participants equally divided by gender. Independent t-tests were performed at alpha = 0.05 to compare mean differences in both test results by gender. The p-values from the both results were less than 0.05 indicating significant gender differences in learning English as a foreign language.

The independent t-test was the most appropriate for analyzing the test score among the students. Some of the reasons that make the independent t-test the best choice for Zoghi, Kazemi, & Kalani (2013) include (1) the independent variable (gender) have a dichotomous nominal level of measurement; (2) the two groups are unrelated; (3) the dependent variable (English test score) has an interval/ratio level of measurement; and (4) there is independence of observations. However, the study abstract comes somehow mislead a reader since it states that a paired t-test was used contradicting what is stated in the methodology section. The study does not support why the independent t-test assumed equal variances. In summary, the choice of variables, the use descriptive and test statistics tables justifies why the researcher chose independent t-test as analysis procedure despite the few weightless mistakes.

Reference

Zoghi, A., Kazemi, S. A., & Kalani, A. (2013). The Effect of Gender in Language Learning. Journal of Novel Applied Sciences, 1124-1128.