Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Background

An important and unique feature of leader-member exchange (LMX) theory is its emphasis on dyadic relationships. Yet, research on supervisor–subordinate relationships has shown convincingly that leaders do not behave consistently and similarly toward all subordinates. Instead, leaders form different quality relationships with their subordinates. High-quality LMX dyads exhibit a high degree of exchange in superior–subordinate relationships. Subordinates in these dyads are often given more information by the superior and reported greater job latitude. Lower-quality LMX relationships are characterized by more traditional “supervisor” relationships based on hierarchical differentiation and the formal rules of the employment contract.

Problem Statement

The leader-member exchange (LMX) theory of leadership focuses on the quality of relationships built between leaders and subordinates; LMX measures are designed to assess the quality of these relationships. Since the leader and subordinate are jointly embedded in the relationship, it is reasonable to assume that their ratings of the relationship will converge to some reasonable extent. However, supervisor–subordinate relationships has shown convincingly that leaders do not behave consistently and similarly toward all subordinates. Instead, leaders form different quality relationships with their subordinates.

A good and valid instrument helps to determine how clearly leaders behave toward their subordinates. Therefore, the objective of this paper is to verify this instrument, the supervisor-rated LMX of their subordinates.

Literature Review

This section reviews the measurement of leader-member exchange (LMX) relationship. The purpose is to provide research background of the scale.

Leader-Member Exchange Relationship

Rooted in social exchange theory (Blau 1964) and the norm of reciprocity (Gouldner 1960), LMX focuses on the quality of the dyadic, interpersonal relationship between the supervisor and subordinate (e.g., Gerstner and Day 1997; Graen and Uhl-Bien 1995; Liden et al. 1997). Supervisors have been shown to give favourable treatment upon subordinates with whom they have high-quality LMX relationships. In return, subordinates have been shown to reciprocate favourable treatment upon their supervisors by engaging in extra role and extra task effort (e.g., Organ and Ryan 1995; Settoon et al. 1996).

Definitions of Leader-Member Exchange Relationship

Interpersonal interactions that employees secure in organizations have important implications on individual well-being, morale, effectiveness and competence, as well as organizational success and productivity (e.g., Shao et al. 2011; Podsakoff et al. 2009). The relationship allows employees to have more opportunity to perform citizenship behaviour for those people who are close to them, regardless of race, gender, or age (Bowler and Brass 2006). Interpersonal exchange relationships with supervisors are important that it ultimately determines how employees define and play their roles within the organizational context (Dienesch and Liden 1986). At the same time, subordinates are not passive, but rather proactive participants who would try their best to change their work environment. As high-quality exchange relationships develop, mutual internal goals and attitude similarities between managers and employees are linked positively to job-related outcomes (Lo et al. 2006).

An important and unique feature of LMX theory is its emphasis on dyadic relationships. Yet, research on supervisor–subordinate relationships has shown convincingly that leaders do not behave consistently towards all subordinates (Dansereau et al. 1975; Graen 1976). Instead, leaders often form different quality relationships with their subordinates (Liden and Graen 1980).

High and Low Leader-Member Exchange Relationship

High-quality LMX dyads exhibit a high degree of exchange in superior–subordinate relationships and are characterized by mutual liking, trust, respect, and reciprocal influence (Dienesch and Liden 1986). Subordinates in these dyads are often given more information by the superior and reported greater job latitude. Lower-quality LMX relationships are characterized by a more traditional “supervisor” relationship based on hierarchical differentiation and the formal rules of the employment contract (Graen and Scandura 1987; Scandura and Graen 1984).

Subordinates in the in-group claimed to have more power as they receive more information, are more influential and confident, and obtain personal attention from their leaders as compared to the out-group subordinates (Liden et al. 2000). In-group members are willing to do extra tasks to which their leaders will reciprocate (Graen and Scandura 1987), but out-group members receive lesser attention and support from their leaders and thus, might see their supervisors as treating them unfairly. Literature (Dansereau et al. 1975; Brower et al. 2000) in the past has revealed that supervisors do differentiate between their subordinates in terms of the exchange. It is an advantage to be in a high-LMX as it is associated with higher trust, greater warmth, and support, and there is more frequent interaction between the members of the dyad (Dansereau et al. 1975; Brower et al. 2000). However, it should be noted that members having initially low LMXs are not necessarily poor performers in the work units. The initially low-LMX group clearly has the potential to consistently produce at higher levels, but it appears that they do not perceive higher performance as being worth the effort. After the one-on-one leadership intervention, the initially low-LMX group responds more positively to the new opportunities than do their colleagues (Scandura and Graen 1984).

It is important for the employees to know where they stand with their supervisors in an organization. To do this, employees need to compare themselves with other colleagues and assess how supervisors and colleagues react to them (Van Breukelen et al. 2006; Lamertz 2002). Employees are aware that with high-LMX, leaders are sensitive to subordinates’ professional credentials and potential work contribution (Henderson et al. 2009), thus would lead to a greater increase in organizational commitment (Leow and Khong 2009).

Relational facets based on power are likely to be more important than instrumental facets. When supervisors demonstrate power capacities, such as treating subordinates with respect, subordinates develop feelings of positive self-worth. Subordinates feel that the supervisors are treating them fairly while using the power, hence yielding positive effects on subordinates’ organizational outcomes (Asgari et al. 2008).

Existing LMX research departs from previous ones in that, LMX has recently been related to behaviours such as organizational citizenship behaviours (Deluga 1998; Hui et al. 1999), task performance (Howell and Hall-Merenda 1999), turnover intention (Ansari et al. 2000), organizational outcomes (Omar 2001), and influence tactics (Liew 2003). LMX literature has found that subordinates in high-quality exchange relationships received more desirable assignments, more rewards, and had greater support from their supervisors. This is congruent with social exchange theory, where individuals who are engaged in high-quality relationships will behave in such a way that their exchange partner will also receive benefits (Lo et al. 2006).

When employees have high exchange relationships with their supervisors, they feel that this would lead to positive treatment by their supervisors. This would induce an obligation on the part of the followers to reciprocate positive treatment from leaders with extra role behaviours. Hence, these employees are motivated to help their leaders and, equally, the organizations achieve their goals (Chan and Mak 2012).

Reciprocity

The basic premise of LMX theory is that managerial actions that demonstrate positive regard for an employee create a desire on the part of the employee to reciprocate through behaviours that are highly valued by the manager (Settoon et al. 1996). Dienesch and Liden (1986) argued that both the theoretical model of LMX and the measure of this concept should be multi-dimensional and proposed a three factor model that included contribution, loyalty and affect.

As mentioned by Murry et al. (2001), the positive exchanges (Bernerth et al. 2007) are typically reciprocated with positive outcomes from the subordinates. Each member of the dyad has the other’s best interest at heart and this is reflected in more supportive behaviour (Lo et al. 2006). Expectations and intention exist in helping behaviours. When a person does a favour by completing a colleague’s task, there is indirectly an expectation of a return (Kandan and Ali 2010) and also could affect the rate of the return as well (Banki 2010). According to Blau (1964; cited in Bernerth et al. 2007), employees will try to balance between inputs and outputs of any of social transaction to stay out of debt and yield towards reciprocity. Even a leader expresses altruism for returning back the effort of his followers’ effort.

Out of reciprocity of LMX, employees tend to repay what they feel they owe. A reason why a person stays committed to the organization might come from the feeling of reciprocity (Felfe et al. 2008). To balance the interpersonal interaction to achieve equity, employees who want to repay a previous kindness may choose to do acts of citizenship behaviour (Poile 2010).

To sum up, LMX is a relationship-based approach to leadership that focuses on the two-way (dyadic) relationship between leaders and followers. It suggests that leaders develop an exchange with each of their subordinates, and that the quality of these LMX relationships influences subordinates’ responsibility, decisions, and access to resources and performance.

Methodology

This research is based on a descriptive and exploratory study that employs the quantitative approach to obtain respondents’ perspectives on LMX.

The methodology employed in this study is a cross-sectional correlational research, a quantitative and deductive research that describes the linear relationships between two or more variables (Sekaran and Bougie 2013). This study involved the gathering of data over a period of three months. The extent of researcher interference was minimal as the study was conducted in the natural environment of the organizations with the normal flow of work and non-contrived study setting. The data was analyzed using the statistical package in the Social Sciences Software (SPSS) version 20 and Rasch Model Measurement (WINSTEPS 3.72.3).

The population refers to the entire group of people, events, or things of interest that is to be investigated (Sekaran and Bougie 2013). Population is defined as a group of potential participants on whom the researcher wants to generalize the results of the study (Salkind 2014). The population of reference for this study included the total number of administrative officers, as supervisors, obtained from 20 public higher education institutions (referred to as public universities) working in various departments and units.

The sampling technique used was by stratified random sampling on 210 supervisors at public universities. A deliberate effort was made to obtain a representative sample by including administrative officers and their supervisors from 20 public universities in Malaysia. According to Sekaran and Bougie (2013), a sample of 100–500 is large enough to generalize the population. Furthermore, with reference to Krejcie and Morgan (1970), for a population size of 1919 (the round up number in the list is 1900), the sampling size should be 320. Thus, consistent with the above suggestions, the minimum sample size targeted in this study was set at 320.

On average, the response rate of survey questionnaires collected is approximately 30–50 % (Wallace and Mellor 1988). Using a response rate of 50 %, the number of questionnaires that need to be distributed should be a least 960 to achieve an effective sample of 320. Out of 979 pairs of questionnaires distributed, only 577 (58.94 %) administrative officer respondent questionnaires’ were returned. Meanwhile, for supervisor respondents, only 343 (35.04 %) were collected. However, there were only 210 (21.45 %) pairs that matched.

Stratified random sampling is the most efficient technique among all probability designs where all groups are adequately sampled and comparisons among groups are possible (Sekaran and Bougie 2013). Proportionate stratified random sampling technique was used whereby the population was first divided into meaningful segments (e.g., organizational categories); thereafter the number of subjects from each stratum would be altered, while keeping the sample size unchanged. The purpose of employing this type of sampling technique is to infer the data results that represent all local universities in Malaysia. Hence, proportionate sampling decision was made because it was suspected that more variability within a particular segment as well as administrative officers’ designations (strata).

In this study, two levels of sampling techniques were conducted. First, the population was divided into organizational (i.e., university) categories. Stratified random sampling involves stratifying the elements along meaningful levels and taking proportionate samples from the strata. This technique is well-represented and more valuable and differentiated information would be obtained with respect to each group (Sekaran and Bougie 2013).

Second, for each important segment (i.e., subordinates and immediate supervisors) of the population, simple random sampling was used because every element in the population has a known and equal chance of being selected as a subject (Sekaran and Bougie 2013).

Findings of Study

An appraisal of data fit to supervisor-rated LMX relationship scale was conducted as a mean of verifying the measure. A total data point of 1770 evolved from 197 respondents on the 9 items analyzed. It produces a Chi-square value of 3309.74 with 1562 degree of freedom (p = 0.000, Table 1). This means that the overall fit to the measurement is good. For this set of data, it can be seen that the mean infit and outfit for item mean square are 1.01 and 1.00 respectively, very much as the expected value of 1.00 (Linacre 2011). Similarly, the mean infit and outfit person mean square are both at 1.02 and 1.00 respectively.

Table 1 Data fit to supervisor-rated LMX relationship scale
Full size table

The mean Z standardized infit and outfit values are expected to be 0.0. As displayed in Table 1, the mean infit and outfit values for item’s Z-standard are both 0.0, while the mean infit and outfit for the person’s Z-standard are both −0.2. Since the values for the mean square and the Z-standard are very much close to the expected values, therefore, it can be said that the data for the actual research does fit the Rasch model reasonably well and appropriate analysis conducted can reveal the outcome of this research.

Reliability of the Instrument

The process of reliability and validity is very much needed in evaluating the quality and reconstructing of a diagnostic tool. According to Jackson et al. (2002), reliability is the consistency of an instrument in measuring what is supposed to be measured. As for this study, it is desirable to verify the supervisor-rated LMX relationship measure to obtain a good and valid instrument that can help to determine how clearly leaders behave toward their subordinates. In terms of the item reliability, the final version of the measure produced ‘excellent’ item reliability (Fisher 2007) of 0.98 logit. It indicates that the probability of the difficulty levels of every item remaining exactly the same if the instrument were given to a different group of supervisors is high. Hence, the instrument holds an “excellent” position of not being dependent on the respondents.

The person reliability is identified as ‘good’ at 0.78 logit by Fisher (2007). In addition, the Cronbach Alpha (KR-20) Person Raw score test reliability is slightly higher at 0.81 logit. With the reliability at 0.81, if a similar set of instrument measuring the to obtain a good and valid instrument, then the likelihood of obtaining a similar pattern of ability in the person measure order table and the location of these supervisors on the person–item distribution map would be fairly similar (Aziz 2010).

Table 2 shows the comparison between cleaned and uncleaned values of supervisor-rated LMX relationship scale.

Table 2 Summary of comparison between cleaned and uncleaned values (9 measured item)
Full size table

In summary, the results showed a good reliability for both item and person measured at 0.98 (SE 0.11) and 0.78 (SE 0.54), respectively. The PCA of explained variance improved from 42.7 to 48.3 %, determining strong measurement dimension.

Discussion and Conclusion

The LMX measures were adopted from a 7-item (LMX7) construct of Scandura and Graen (1984) and additional 12 items were adopted from Bernerth et al. (2007). Several iterations were done by deleting the items identified as misfits. The better instrument was constructed, showing marked improvement across various fit statistics.

Quality control procedures have resulted in an item reduction from 19 to only 9 items, thereby producing a better instrument in measuring the supervisor rating of LMX of their subordinates. Referring to Fig. 1, after assessing the 9 items, the measurement can be divided into two categories; work-related and nonwork-related items. For work-related items, a supervisor expects a loyal employee to reciprocate through work contributions (Maden 2015). Whereas, for nonwork-related items, a supervisor expects his/her employees to hold the feelings of respect and affect towards the supervisor (Long et al. 2015). This finding also could perhaps be attributed to the fact that Malaysian society is a hierarchical and relationship-oriented society (Ansari et al. 2004; Hwa et al. 2008).

Fig. 1
figure 1

9 measured items of supervisor-rated LMX relationship measurement

Full size image

Future Directions

This study developed a psychometrically sound LMX scale that captures the exchange process between leaders and members. A significant drop in the number of supervisor-rated LMX scale items which was originally from 19 to 9 items was an alarm. Even though, necessary steps were taken in ensuring that the items were of “good quality”, the large number of item deletion needs to be investigated. Future investigation on this issue should be done to validate further on the 10 misfit items which could be related to high power distance, high collectivist, and high performance orientation of Malaysian culture.

However, the results using the revised instrument may yield awareness and understanding among subordinates how their supervisors perceive LMX of them; hence, necessary action can be executed by the employees, supervisors, and the organization, to improve LMX relationship among employees.