Measuring the Impact of Human Resource Practices on Organization Performance: Scale Validation Based on Pilot Study

Abstract

Validity and reliability are crucial when interpreting scores derived from psychometric instruments like questionnaires, symptom scales, educational tests, and observer ratings. These instruments find application in clinical practice, research, education, and administration. The traditional concepts of face, content, and criterion validity are being replaced by the unified concept of "construct validity," which pertains to the extent to which a score accurately represents the intended underlying construct. Validity can be envisioned as a meticulously structured argument, where evidence is compiled to either support or challenge proposed interpretations of findings. While reproducible (reliable) outcomes are essential, they alone do not suffice to form valid conclusions. This research aims to develop a scale to determine the impact of HRPs on an organization’s performance. The principal component factor analysis results reveal that the scale encompasses six distinct dimensions, each comprising four to five items. To establish the validity and reliability of the scale, various statistical analyses were conducted, including Cronbach Alpha, Average Variance Extracted, HTMT, and Fornell & Larker. This piece offers a concise introduction to the concepts of validity and reliability concerning the overall effectiveness of test instruments. It outlines the essence and understanding of content, face, criteria, and construct validity, and highlights various methods for gauging and enhancing reliability. The use of PLS-SEM in conjunction with SPSS allowed thorough analysis. When an item fails to reach the statistical testing threshold level, it is reviewed and possibly revised. Although some things were not deemed valuable at first, they were later considered useful in gathering sufficient data to complete the primary study. This study identified key errors in its chosen instrument based on preliminary testing with suggestions to review or revisit particular item/s that may impact results while undertaking subsequent data analyses.