| Summary: | Introduction: In structural equation modeling (SEM), among the desirable requirements in the measurement model are the reliability and validity of the constructs. The average variance extracted (AVE) provides a numerical measure of the overall validity of each construct in the model. Meanwhile, composite reliability (CR) reflects the internal consistency reliability of the items under each construct. Materials and methods: In this study, the existing estimator in SEM namely unweighted least squares (ULS) has been used for nonnormal data in SEM. However, the method is seen less efficient as the method leads to improper solutions like unique variance which introduces some level of bias, hence affecting the reliability and validity of the constructs. Therefore, the regularized unweighted least squares (ULS), a new approach of regularization is introduced in this study. Utilizing 300 samples of breast cancer awareness data, the analysis was carried out using “lavaan” package in R programming Environment. Results: Regularized ULS consistently yields higher CR and AVE values, enhancing the reliability and validity of measurement instruments. Conclusion: Besides assisting researchers in achieving the reliability and validity of a construct, the findings of this study can aid survey-based researchers to generate a more reliable model. The findings indicate that employing regularized ULS estimation allows for the retention of a greater number of items or questions within the respective construct in the Malay Version of the Breast Cancer Awareness instrument. This proves to be invaluable in validating the factor structure through confirmatory factor analysis. © 2024 Universiti Putra Malaysia Press. All rights reserved.
|
|---|