Discriminant analysis in export research: an imperative for methodological rigor

Abstract

Purpose The purpose of this paper is to appraise methodological rigor in the application of discriminant analysis (DA) in export-focused research and to offer guidelines for future studies. Design/methodology/approach The sample includes 89 empirical peer-reviewed studies, comprising 102 models published over the period 1979-2014. Content analysis and vote counting are used to evaluate each of these studies. Findings This review highlights major flaws in the application of DA in export research. The shortcomings are self-evident particularly concerning suitability of DA for research context, completeness in the reporting of descriptive results, and validity and reliability of predictive results. Practical implications The study takes the position that the lack of methodological rigor may be undermining the eminence of knowledge in exporting, and this has extensive implications for both researchers and practitioners. Originality/value This review outlines steps to assess methodological rigor associated with DA and offers guidelines for scholars seeking to enhance rigor in future research.

No Justice, No Peace? Protest Participation for People with Criminal Legal Contact

“…Jackknife estimates guard against overfitting the data by leaving one observation out of each estimation to ensure that it does not bias the output (Stata 2020). It reduces the bias from certain cases that might otherwise be seen as outliers and thus assesses the validity and stability of an analysis without necessitating a large sample size (Fenwick 1979; Kahiya 2017; Powers et al 2000). Using this model, I determined that the results presented here are not influenced by outliers, suggesting that a larger sample size would likely support my results.…”

Surface Defect Identification and Grouping of Intermittent Leather Images using Linear Discriminant Model

“…Hence, it is necessary to improve the efficiency of the process led to the use of image processing tools to handle the problem that can achieve uniformity and accordingly this work focused primarily based on the GLCM [6] features. Linear discriminant analysis (LDA) [7] is used as a classifier to understand the relationship of the features.…”