The motivation for ISO 9000 certification: a multivariate predictive approach

• Main Author: Texeira Fernandes Justino, Maria Do Rosário
• Publication Date: 2022
• Other Authors: Almaça, José Figueiredo, Texeira Quirós, Joaquín, Antunes, Marina Godinho, Mucharreira, Pedro Ribeiro
• Format: Article
• Language: eng
• Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Download full: http://hdl.handle.net/10400.21/15517
• topic: Quality; ISO9000; TQM; Logistic regression; Decision tree

Our empirical study compares the responses provided by certified and non-certified companies using multivariate predictive models. Conceptually, there is a subtle difference between comparing questionnaire responses with T-tests and multivariate predictive models. The former should be employed when we conjecture that certification conducts to different responses to the questionnaire; the latter should be used when we believe that the respondent’s perception on the intensity with which quality practices are implemented in their firms is a predictor of certification. That is, rather than asking whether certified companies provide responses that are statistically different from those provided by non-certified companies, we try to understand which of these responses predict, or are associated to the likelihood of the firm being certified. Using a multivariate approach, we identify the set of questionnaire items that are statistically associated to this likelihood, after we control for other items. We consider two types of multivariate models. First, we implement a parametric logistic regression model. Then, because an incorrect specification of the expected value of the dependent variable in the logistic model may lead to incorrect inferences, we also implement a nonparametric decision tree model. It is shown that companies in which the respondent manifests greater concern with respect to the increased competition due to globalization, the motivation of employees, and strategic planning have higher likelihood of being certified. Using a questionnaire answered by a sample of certified companies and a control sample of companies which have not been certified, we show that T-tests and multivariate models provide different insights. According to the T-tests, most responses are significantly different whether they are provided by certified or non-certified companies. On the other hand, the logistic regression model suggests that only three questionnaire items are statistically significant in predicting if a firm is certified or not. The decision tree model indicates that a fourth item also explains the likelihood of certification.