Possible Errors in Quantitative Data Analyses

Abstract

The undertaking of research in the sciences or business normally requires an understanding and correct application of carefully selected statistical methods and tests. Most often than not, statistical analyses used for certain types of data collected in research are faulty and misleading. There is often the temptation to overuse statistics to produce fanciful outputs by researchers at one extreme, while others at the other extreme just apply simple descriptive statistics to the data without inferring the sample results to the population as should be. This paper attempts to examine various statistical analyses and the most likely appropriate tests that should go with them with a focus on the Likert scale controversy. The most commonly and frequently used tests were gathered through a review of scientific and business publications and distinguished into either parametric or non-parametric tests. The resulting outcome presents a quick table with a handy explanation for choosing statistical tests based on the kind of data a researcher may be dealing with.