Global Journal of Management and Business Research, A: Administration and Management, Volume 22 Issue 9

Table 4.3: Descriptive Analysis of the Constructs Construct N Mean Standard Deviation Minimum Maximum Compensation 583 3.91 0.553 2.60 4.90 Job involvement 583 3.91 0.609 1.60 5.00 Feedback 583 3.90 0.479 2.14 4.86 Coaching 583 3.93 0.597 2.07 5.00 Employee Productivity 583 3.85 0.713 2.08 5.00 Source: Field survey The descriptive statistics for the constructs revealed that the statistical mean value of 3.91 for compensation was similar to the statistical mean value of 3.91 for job involvement. Moreover, the descriptive analysis for the construct of feedback shows that it has the statistical mean value of 3.90 as compared with the coaching, which has the statistical mean value of 3.93. The statistical mean score of employee productivity of 3.85 is relatively lower than the mean score of all other three remaining variables. a) Measurement Scale of the Research Variables The analysis of the measurement scale as well as the types of research variables used in the current study is of paramount importance. Thus, scale is viewed as a necessary tool used in quantitative research for distinguishing respondents (individual) on the basis of how they differ from one another in using some selected variables (Gelaidan, Ahmad, 2013). In effect, the four major types of measurement scales that have been identified and are widely used in recent research are ordinal, nominal, interval, and ratio. Table 4.4: Descriptive Analysis of the Measurement Scale Types Variables Types of Scale Demographic Nominal Compensation Ordinal Job involvement Ordinal Feedback Ordinal Coaching Ordinal Employee Productivity Ordinal b) Validity and Reliability of the Constructs Convergence and discriminant validity seeks to establish a level of agreement between the concept of the theory and a particular research instrument by ensuring that its attributes in the measurement scales are well represented (Cohen, Manion, & Morrison, 2007). Thus, the convergent validity is identified by assessing the loadings, composite reliability as well as average variance extracted (Cohen et al. 2007). Furthermore, convergence validity is achieved when all the measures that are expected to reflect a construct are correlated.In other words, the legitimacy of a specific factual estimation scale is joined when values are > 0.7, > 0.7, and > 0.5 for loadings, composite reliability, and average variance extracted respectively. So, values above these stated points are considered to have enough convergence of validity (Cohen et al., 2007). In the current review, every one of the things have surpassed the suggested edge values for loadings, normal difference removed and composite reliability or unwavering quality (> 0.7, >0.7 and > 0.5) on their particular develops. Particularly, items A5, A6, A10, C1, C2, C4, D1, D2, D7, D8 and E12 which were dispensed with for having a low stacking. Table 4.5 presents the loadings, average variance removed, and composite dependability of this review. Table 4.5: Results of Measurement Model CONSTRUCTS ITEMS LOADINGS AVE CR COMPENSATION A1 0.751694 0.679988 0.936780 A2 0.856613 A3 0.875734 A4 0.772489 A7 0.856613 32 Global Journal of Management and Business Research Volume XXII Issue IX Version I Year 2022 ( ) A © 2022 Global Journals Mediating Role of Coaching on the Relationship between Compensation Job Involvement and Feedback on Employee Productivity