Invariance of Organisational Citizenship Behaviour Measurement Model across Institutions in Malaysia

Organisational Citizenship Behaviour (OCB) is a discretionary behaviour that benefits organisations and their members. Even though OCB has been broadly researched over the years in the US, the measurement of OCB has received rather limited deliberation particularly in Asian contexts like Malaysia. This study tests the adequacy of the OCB measurement model and analyses the invariance of the model of 417 respondents in two Malaysian institutions. A three–dimensional OCB model was found to be adequate after having been subjected to Confirmatory Factor Analysis (CFA). As a result of discriminant validity, the two dimensions of OCB ‘helping’ and ‘civic virtue’ were found to measure the same thing. Consequently, the two dimensions were merged and named ‘concern’. Although the factors were found to have acceptable reliability and validity with very few modifications, there is a need to further test the model with different and larger samples. This study also shows that the CFA model experienced invariance across institutions. The results have practical implications for recruitment and training managers to utilise the instrument as well as to pay more attention to the importance of cultivating OCB among staff of the sampled institutions.


Introduction
Organisations require employees to transcend the formal requisites of the job role and be capable to act as 'good soldiers'.OCB has a reflective impact on the performance of individuals in the organisations.Performance becomes amongst the most important factors for success and demarcates the top performers from the underachievers (Rotundo & Sackett, 2002).The term OCB was coined by Bateman andOrgan in 1983 while in 1988 Organ defined it as, individual behaviour that is unrestricted, indirectly or ambiguously identified by the formal reward system and that, in the aggregate, supports the effective operation of the organisation.The three dimensions of OCB are helping, civic virtue and sportsmanship.Despite the popularity of OCB studies in industrial and organisational psychology, the measurement of OCB has received relatively limited consideration particularly in Malaysia (Lo & Ramayah, 2009).The purposes of this study are twofold: (1) to establish the construct validity of OCB and (2) to cross-validate its measurement invariance across institutions.

OCB Relationships with Other Constructs
In accordance with the relationships between OCB and other constructs, meta-analyses and reviews reveal that OCB is related with many indicators of organisational and group effectiveness (Podsakoff & MacKenzie, 1997;Podsakoff, Whiting, Podsakoff, & Blume, 2009).Superiors auspiciously assess their employees' OCB in relation to their overall promotion and job performance (Podsakoff, MacKenzie, & Hui, 1993;Podsakoff et al., 2009).OCB also facilitates affirmative working surroundings, which attract and retain the employees of those particular organisations (Organ, 1988).
In addition, OCB promotes the functionality and effectiveness of a work unit or organisation so long as employees are educated on its importance.In support to this, Hoffman, Blair, Meriac, John and Woehr (2007) examined the relationships between task performance and OCB and attitudinal variables with OCB.Their results reveal that a one-dimensional OCB model relates to both variables with a strong relationship between OCB and attitudes.

OCB Relationships with Mediating Effects
Efficacious leadership may contribute to the progress of more vigorous perceptions of organisational and individual instrumentalities which then leads to OCB.In relation to this, Jiao, Richards and Zhang (2011) assessed how perceived organisational instrumentality (the belief that OCB contributes to the functionality and effectiveness of the work unit or organisation) and perceived individual instrumentality (the belief that the employees' own interests is important to OCB) related to the employees involvement in OCB and the ways in which it mediates the leadership effect on OCB among mid-sized financial company of 161 supervisor-subordinate dyads in China.The results show that perceived organisational instrumentality variance explained in OCB was beyond perceived individual instrumentality.In addition, perceived organisational and perceived individual instrumentalities are partially mediated the relationship between leadership (transformational leadership and contingent reward) and OCB.

Validating the OCB Measure
Few studies have sought to validate the OCB measure.Becker and Randall (1994) assessed the validity of OCB against an objective behavioural criterion after examining the factor structure of a revised version of OCB.Factor analysis suggests that this revised version of OCB has only conscientiousness and altruism dimensions.The results show that the 16 restaurant managers in the fast food service industry assess and perceive subtle components of job performance with reasonable accuracy.In addition, Hoffman et al. (2007) used Confirmatory Factor Analysis (CFA) to determine OCB dimensionality and tested its construct validity.Their results support the idea that a single factor OCB model was strongly related to task performance although they are distinct to each other.Furthermore, the shared variance of OCB with 'attitudinal correlates' is beyond 'task performance'.Based on these validation studies, it is hypothesised that: Hypothesis 1: The OCB measurement model with three-factor structure is construct valid.
In accordance with the invariance analysis, Tayyab (2005) analysed the OCB invariance across two types of employees in Pakistan.The results revealed the equivalency of the items across permanent and contingent employees.It means that item content was identified in exactly the same way in both groups of employees whereby each group responded to the scale items in a similar conceptual frame of reference.Furthermore, Lievens and Anseel (2004) found clear support that the measurement structure of OCB had relative invariance across peer ratings of 215 employees and supervisor ratings of 259 subordinates in a Flemish part of Belgium within a Dutch-speaking context.Based on these researches, the second hypothesis was formed stating: Hypothesis 2: There is relative invariance of the OCB measurement structure across institutions.
Thus, the present study hypothesises that helping behaviour, civic virtue and sportsmanship dimensions of the OCB would be related positively to the quality and quantity of the organisational performance.Consequently, this study analyses the adequacy of three OCB factors CFA model together with its invariance analysis across institutions.

Respondents
The respondents were 560 staff members from two educational institutions in Malaysia whom were selected based on cluster sampling.Kerlinger and Lee (2000) stated that cluster sampling is the most effective sampling method for a large number of small clusters.A total of 560 questionnaires were disseminated to both institutions (280 questionnaires each institution) with at least eight questionnaires for each cluster.After cleaning up the data using SPSS, the remaining samples were 417 from a total of 467 completed questionnaires which were gathered at the end of the data collection process.Thus, only 74.5% questionnaires were usable for this study.The respondents consisted of 32.8% males and 67.2% females.The mean age was 36.54 (SD = 9.73) with 27 missing cases.In terms of educational level, 35.55% hold secondary schools certificates, 30.15% were diploma holders, 13.72% were bachelor degree holders and 20.58% attained post-graduate degrees (9 missing cases).

Research Instruments
The Organisational Citizenship Behaviour (OCB) questionnaire developed by Podsakoff et al. (1997) was used.It consists of 13 items measuring three dimensions which were: helping behaviour with seven items, civic virtue with three items and sportsmanship with three items.Responses were rated on a Likert Scale format, with answers ranging from 1 for 'strongly disagree' and 7 for 'strongly agree'.All of the sportsmanship items were reversed-coded.The internal consistency of the items was deemed reliable as the Cronbach's alpha was .804(Hair et al., 2010).Table 1 shows the descriptive statistics and list of items of the OCB measure used in this study.

Normality
The normality of the model is based on skewness and kurtosis.The value of skewness must be smaller than 3 (< |3|) (Chou & Bentler, 1995) and the value of kurtosis must be smaller than (< |10|) (Kline, 2011).Several outliers and the lowest loading items will be eliminated from the analysis until the normality is obtained.or normed chi-square.Marsh and Hocevar (1985) described that the minimum fit function for CMIN/DF of an acceptable fit is between 2 and 5 (2≤ χ2/df ≤ 5); (4) the root mean-square error of approximation (RMSEA) was also presented.Browne and Cudeck (1993) demonstrate that RMSEA of smaller than 0.08 (<0.08) as a reasonable error of approximation.In addition RMSEA of between 0.08 to 0.10 indicates a mediocre fit www.ccsen (MacCallu Cudeck, 19

Constr
The emplo proposed loading fo than 0.7 (≥ (>0.5) as a AVE of a l constructs 2011).

Invaria
The study was estim fulfilled th used was S to ensure t

Norma
The norma and 10 (<

Confir
In this s conceptual with its pa The result 3.756, the

1997) model
) was del. www.ccsen In this mo both 'help same thing reflected th 2 shows th

Constr
In this stu above 0.5 Furthermo discrimina therefore t validity of Table 3.

Invaria
In this stud  4, the invariance test across both groups resulted in a statistically insignificant change in the Chi-square value, Chi-square (df =9) = 12.673, p> .005.Simply said, the difference in the Chi-square values between the baseline model and the constrained model did not produce a poorer-fit model (Kline, 2011).The loadings did not vary significantly across institutions.It can thus justifiably conclude that the institution did not interact with the underlying traits to influence the staff members' responses to the indicators of OCB.This suggests that institution is not a moderating variable.

Discussion and Conclusion
This study investigates the dimensionality of OCB measures through confirmatory factor analysis across institutions in Malaysia.There were 417 respondents in this study.This study assessed the OCB measurement model with the three dimensions of helping, civic virtue and sportsmanship.In testing construct validity, there is a strong support for the discriminant validity issue for the two factors 'helping' and 'civic virtue'.In order to unravel this issue, these two constructs were combined into 'concern'.This scenario suggests that 'helping' and 'civic virtue' are in fact a similar attitude, experienced simultaneously by respondents in this research.It indicates that the two constructs are moulded together when staff members of both institutions were fulfilling their job responsibilities at their respective organisations.In other words, 'helping' comes together with 'civic virtue' and vice-versa.Hence, an organisation may deduce that staff members of both institutions with 'helping' behaviour are also those with 'civic virtue', which means staff members who possess the attitude of 'helping', also feel a strong sense of deep concern and active interest towards their organisations.Based on this finding, human resource managers of both institutions can use the findings to introduce new training programmes that use modules structured specifically to improve the new construct 'concern'.The 'concern' combines 'helping' and 'civic virtue', therefore, the module should consider matters related to 'helping' and 'civic virtue' as one construct.In addition to test the adequacy of the model, this study also sought to analyse the invariance of the OCB measurement model across two institutions.It was found that the measurement structure of the OCB was relatively invariant across these different sample groups.Therefore, the core idea is that the forms of citizenship behaviour observed signify well for the equivalence of OCB ratings across institutions.Future studies can include respondents from more than two institutions, so that the findings can be generalised to a wider group in a given population.Examining the invariance of OCB scales across different nature of institutions may be an interesting avenue for future research.In conclusion, this study provides support for both research hypotheses.
CFA of the Structural Equation Modeling version 18 AMOS model-fitting program was applied to test the validity of items and constructs as well as to test the research hypotheses.The maximum likelihood estimation (MLE) method was utilised for the estimation of parameters.The overall model fit was assessed by using four indices of the model goodness-of-fit: (1) the chi-square statistics; (2) the comparative fit index (CFI); McDonald andMarsh (1990) stated that CFI values close to 1 indicate a very good fit; (3) the minimum value of the discrepancy between the observed data and the hypothesised model divided by degrees of freedom (CMIN/DF) Figure 2. The nd Reliability r loading for t composite r readings range he squared corr eater than the enerally, the m

Table 1 .
The descriptive statistics and list of items of the OCB measure