Job involvement and performance among middle managers in Sri Lanka

Abstract

This paper examines performance evaluation outcomes for middle managers in the garment sector in Sri Lanka and seeks to explain variations in levels of job involvement and perceptions of organizational support. Some 155 middle managers across three firms were segregated into high- and low-performing groups. High-performing managers tended to have higher perceptions of organizational support, resulting in a social exchange relationship with their employer, and expressed higher job satisfaction and lower turnover intentions. Managers assessed as low performers experienced more negative perceptions of organizational support, lower job satisfaction and an economic exchange relationship with their employer. Significantly, for both high and low performers these outcomes were moderated by job involvement. These findings highlight the need for organizations to pay careful attention to the factors influencing job involvement and perceived organizational support. The paper concludes with a discussion on the practical implications of the findings for human resource managers.

Keywords::

• economic exchange
• job involvement
• job satisfaction
• perceived organizational support
• performance evaluation
• social exchange
• turn over intentions

Acknowledgements

The authors would like to thank Ms Anagi Karunasena for her valuable contribution in data collection and an anonymous reviewer for very helpful comments on an earlier draft of this paper.

Notes

^1. The discriminant validity was assessed by following Fornell and Larcker (1981), where the square root of the AVE of each construct should be greater than the raw entries and column entries of the correlation matrix, when the square root of AVE of the constructs is placed on the relevant point in the diagonal of the correlation matrix.

^2. PLS provides two indicators for the assessment of hypothesized relationships among the constructs. They are path coefficient and R². Path coefficient is similar to standardized regression coefficients (β values) and can be used to assess the strength of relationships, while R² can be taken as the measure of the overall effect. According to Hulland (as cited in Schepers, de Jong, Wetzels and de Ruyter 2008), R² of endogenous constructs indicate whether a given PLS model accomplishes the purpose of maximizing the variance explained.

^3. In arriving at the results in Table 6, the indicators of both the predictor variable and the moderator variable were used to create the product indicators by the multiplication of indicators of both the predictor and the moderator. In creating the product term indicators, following Chin et al. (1996), the indicators are standardized before their multiplication. They recommend standardizing indicators, when the ordinal to interval-level items such as Likert scale is used, as in this study. First, these product indicators are used to create the latent interaction construct, that is, the predictor*moderator. Second, the R² of the effect of predictor and moderator after creating the interaction effect is obtained. Third, to assess the statistical significant of the main effect and the interaction effect, a bootstrapping procedure was used.