Do supportive organizations make for good corporate citizens?

In today's competitive business climate there is greater recognition of the importance that employee extrarole work behaviors play in organizational effectiveness (Borman and Motowidlo, 1993; Van Dyne et al., 1995; Van Scotter and Motowidlo, 1996). Extra-role behaviors, often labeled organizational

Given the importance of OCBs, it is critical to develop a more complete conceptualization of the construct and to identify the types of discretionary behaviors employees choose to exhibit under various conditions (Orr et al., 1989). Although there have been many conceptualizations of OCBs put forth in the literature, one of the most promising is to consider the beneficiaries of these behave iors (Williams and Anderson, 1991). Specifically, research supports a two-- factor classification of OCBs into behaviors that are interpersonally-focused and directly and intentionally aimed at assisting others (i.e., OCBI, which include behaviors such as orienting new employees and assisting a fellow employee with a heavy work load), and more organizationally-focused OCBs that function as a more impersonal form of citizenship directed at accomplishing organizational goals (i.e., OCBO; Latham and Skarlicki, 1995; Williams and Anderson, 1991). Although Williams and Anderson (1991) and others (Kemery et al., 1996; Moorman et al., 1998; Skarlicki and Latham, 1997) have provided preliminary evidence for the differential effects of the extrinsic and intrinsic components of job satisfaction on OCBOs and OCBIs, respectively, a logical next step is to extend these findings and consider additional antecedents as differentially predictive of these two OCB dimensions.

Therefore, the purpose of this study is to develop a more complete understanding of when and why employees decide to allocate discretionary effort toward different beneficiaries. Specifically, we focus on employees' perceptions of the extent to which they are valued by their organization (i.e., perceived organizational support; POS) (Eisenberger et al., 1990; Eisenberger et al., 1986; Wayne et al., 1997). POS develops through multiple exchanges between employees and their employers over time, and reflects the degree to which employees perceive that their work organizations genuinely care about their personal welfare (Eisenberger et aL, 1986). The purpose of the present study is to explore how these perceptions of organizational support are associated with employee engagement in two types of discretionary work behavior, OCBI and OCBO (e.g., Williams and Anderson, 1991).

POS was chosen as an antecedent based on both theoretical and empirical findings, suggesting the importance of this variable in influencing OCBs. That is, past research suggests a positive relationship between POS and extra-role behavior. Eisenberger et aL (1990) found a positive relationship between employees' perceptions of organizational support (POS) and the constructiveness of anonymous suggestions for improving organizational effectiveness (i.e., as judged in terms of the usefulness and concreteness of the suggestion). Although making suggestions for organizational improvement is one specific example of OCB (i.e., voice behavior) (Van Dyne and LePine, 1998), the narrow scope of the criterion used in the Eisenberger et al. (1990) study provides limited support for the relationship between POS and general OCBs. Specifically, the Eisenberger et aL (1990) study used employee suggestions for organizational improvement as the sole OCB measure. Despite Eisenberger et aL's (1990) results suggesting the presence of a general relationship, providing only self-report evidence on a single-item measure of OCB fails to allow exploration of the more complex relationship between POS and the different forms of OCB.

The goal of the present study is to first provide a stronger test of the relationship between POS and OCBs by retaining a two-dimensional OCB criterion and by using data collected from two independent samples at different field sites and within different types of organizations (i.e., service and manufacturing). Specifically, following the recommendations of Williams and Anderson (1991), who urge researchers to study the possible differential effects of variables on the two OCB dimensions, we ask whether POS is differentially related to OCBI and OCBO.

The following sections provide a literature review and hypothesis, the methods used, results found, and discussion of two independently conducted field studies examining the differential prediction of POS as an antecedent of both OCBOs and OG BIs. The methods, results and discussion for Study 1 are presented, followed by the methods, results and discussion for Study 2. Finally, specific managerial and theoretical contributions of these findings and discussion of the studies' limitations are presented in the general discussion section.

Literature Review and Hypothesis

Consistent with the literature on OCBs, we adopt social exchange theory (Blau, 1964) and the norm of reciprocity (Gouldner, 1960) as theoretical frameworks for predicting OCBs (Organ, 1988). Prominent in both of these perspectives is the concept of unspecified obligations. That is, when one party does a favor for another, there is an expectation of some future return, although exactly when that favor needs to be returned and what form it will take is unclear (Wayne et aL, 1997). One way in which these unspecified obligations develop for employees is through perceptions of organizational support (POS). POS is defined as 11. . . the global belief held by an employee that the organization values her/his contributions and cares about her/his well-being" (Eisenberger et aL, 1986: 501). It is primarily affected through the same social exchange factors people use in making attributions; namely, the frequency, extremity, and perceived sincerity of statements of praise and approval given to employees by organization representatives (Blau, 1964). Social exchange theory predicts that perceived support raises an employee's expectancy that the organization will reward greater effort toward meeting organizational goals (i.e., effort-outcome expectancy; Eisenberger et al., 1986). To the extent that perceptions of support also provide needed praise, the employee may incorporate organizational membership into self-identity and thereby develop an emotional bond to the organization (Eisenberger et aL, 1986).

Employees appear to seek a balance in their exchange relationships with organizations by demonstrating attitudes and behaviors commensurate with the amount of commitment they feel the employer has for them (Wayne et aL, 1997). Being an effective organizational citizen is one way that an employee may reciprocate the support he/she feels is being provided by the organization (Graham, 1991). In turn, making suggestions for improvement, helping co-workers, and other types of OCB incur obligations that the other party (i.e., the individual or the organization) will later reciprocate. In so doing, the repayment of these obligations reinforces giving and strengthens the mutually beneficial exchange between the employee and organization (Eisenberger et aL, 1986).

We propose that the two categories of organizational citizenship behavior (OCBO and OCBI) may be related differently to perceived organizational support. Perceived organizational support refers to the employee's perception about how an organization treats employees. Therefore, following social exchange theory (Blau, 1964) and the norm of reciprocity (Gouldner, 1960), POS should be more closely tied to employee behaviors that are specifically directed toward the organization (OCBO), than those directed toward assisting co-workers (OCBI). Employees who believe they are supported by the organization (high POS) may engage in discretionary work behaviors as a way to reciprocate for the feelings of importance and value received from the organization, but the target is the overall organization rather than a specific individual.

As noted, the empirical support for the relationship between POS and OCB reported by Eisenberger et al. (1990) provides indirect support for this prediction in that their OCB criterion measured whether employees made suggestions for organizational improvement (i.e., a measure of OCBO). In addition, Williams and Anderson (1991) urged researchers to retain a two-dimensional approach to the study of OCB because, as suggested above, different sets of predictors may better explain these two classes of OCB. Therefore, on the basis of these theoretical arguments and consideration of the OCB criterion used in Eisenberger et aL (1990), we predict that the relationship between POS and OCBO will be stronger than that between POS and OCBI.

Method

Two independently conducted studies are described. Although the goal of both studies was to provide a test of the POS-OCB relationship based on a two-dimensional OCB approach, the specific method employed by each study differed with respect to the measures, analysis techniques, rating sources, and samples used. To the extent the findings converge, these differences in methodology are complementary and can serve to strengthen the inferences drawn from results. Although both studies used cross-sectional designs and were conducted in field settings, the two studies differed in the industries that served as organizational settings.

Study 1

Participants and Procedure. The organization that participated in study 1 is a moderate-sized manufacturing organization based in the South. Participants were drawn from the 115 line-level employees who work in the Assembly Department of one division. Employees are responsible for building, assembling, and distributing plastic conveyer belts.

Data were collected from participants via a survey that was administered as part of a larger project on incentive pay. The response rate was 94% (108 of the 115 employees completed the survey). However, only 88 surveys were usable due to participant unfamiliarity with English, which was apparent both during survey administration and upon examination of the survey responses. The N=88 study sample was comprised of 50% females. The average respondent was between 31-40 years old, had held his/her current job for 3.5 years, and had at least a high school level of education. The ethnic composition of the sample was 21% White, 41% African American, 17% Vietnamese, 14% Hispanic, and 7% identified themselves as "other" or failed to respond.

One week prior to survey administration, the division manager explained to employees that as part of an examination of the department's incentive pay practices, employees were being asked to voluntarily complete a survey about the incentive pay plan and their work attitudes. Questionnaires were administered during company time in a group format one week later.

Measures. All scales used in the study were scored on a five-point Likert scale (1 = Strongly Disagree to 5 = Strongly Agree) with an additional sixth answer option to allow for "Don't Know" responses.

1. Organizational citizenship behaviors. OCBs are behaviors of a discretionary nature that are not part of employees' formal role requirements, but nevertheless contribute to the effective functioning of an organization (Organ, 1988). As other researchers have noted (e.g., Van Dyne and Le Pine, 1998), peers are often the most appropriate source for rating OCBs. Moreover, employees in this organization provide anonymous evaluations on their co-workers' performance as part of the organization's annual performance appraisal process. For these reasons, each participant was randomly assigned another member of their work group to rate on 12 OCB items. The OCB measure was adapted from Moorman and Blakely (1995) and Podsakoff et al., (1997) and measured three general dimensions of OCB (i.e., helping, civic virtue, and sportsmanship). Decisions on specific items to include in the OCB measure were made on the basis of previous reports of item factor loadings and reliability statistics.

Exploratory factor analysis using the principal axis method and oblique rotation showed the greatest support for a two-factor solution. The first factor is labeled OCBI. A sample OCBI item is, "[This employee] is always willing to help others if they need a hand." The second factor is labeled OCBO. One example of an OCBO item is, " [This employee] provides constructive suggestions about how our work group can improve." Although a three-factor model was considered, the three-factor solution revealed a third factor eigenvalue of .72, as well as several complex factor loadings. In the two-factor solution, however, both OCB factors revealed eigenvalues greater than 1 (see Table 1). The Cronbach's reliability estimates for the two OCB factors were .83 for the OCBO scale and .91 for OCBI scale. Although the OCBO and OCBI scales were positively correlated = .62, p<.01), considering the theoretical argument for retaining a two-factor OCB criterion, as well as adequate empirical differentiation to warrant the two-factor classification, we treated the two factors as separate OCB measures.

2. Perceived organizational support. POS is the extent to which an employee feels that the organization cares about his/her well-being. POS was assessed with a six-item scale. Items with high factor loadings from the original 36-item measure from the Survey of Perceived Organizes tional Support (SPOS) (Eisenberger el al., 1986) and from Kottke and Sharafinski (1988) were chosen to assure non-redundancy and adequate representation of the content domain. Items were written with upperlevel management as the referent, as management is typically seen as representing the beliefs, values and support systems of the organization (Eisenberger et al., 1986).

The reliability of the scale was found to be reasonable (alpha = 77). A sample POS item is: "Managers consider my best interests when they make decisions that affect me." A higher score reflects a stronger perception on the part of the employee that they are supported by the organization. An exploratory principle axis factor analysis supported a unidimensional structure for the six-item scale. The eigenvalue was 2.24, and all items loaded significantly on this factor (all loadings >.40; see Table 1).

Results. Pearson bivariate correlation analysis was used to test the study hypothesis. A one-tailed test was used to test for significance because of the proposed directional hypothesis (see Table 2). Marginal evidence for the differential prediction of OCBO and OCBI consistent with our hypothesis was shown for POS. Specifically, although the correlation between POS and OCBI was not significant = .10, ns), the relationship between POS and OCBO was significant = .23, p<.05). Using a one-tailed significance test, following procedures outlined in Cohen and Cohen (1983), the difference between these two correlations was found to be marginally non-significant (t(67)=1.20, p=.11).

Discussion. We hypothesized that the relationship between POS and OCBO would be stronger than that between POS and OCBI. Our results lend marginal support to this prediction, and to arguments for retaining a multidimensional OCB criterion. Marginal support for the differential hypothesis was likely a result of our modest sample size and corresponding lack of statistical power. Thus, additional research is warranted to determine whether the differential prediction between POS, OCBO, and OCBI can be replicated. Study 2 accomplishes this using a larger sample, different measures and analyses, and another organizational setting.

Study 2

Participants and Procedure. Participants were employees recruited from several restaurant companies located in the Midwest. General Managers were first contacted in person for authorization to approach employees. Once these managers agreed, employees were then asked during employee meetings to volunteer to complete questionnaires. The 257 participants for this study were drawn from approximately 350 eligible participants. Eligibility was determined by only including entry-level employees who had worked for the organization for at least one month. Data were collected from participants via two on-site surveys, one administered to employees and the other (the OCB measures) given to managers. The employee response rate was 74%. The sample was 75% female, 73% white with a mean age of 27 and an average tenure of 2.7 years. After completion, the OCB measures were hand-collected by the second author.

Measures. All scales used in the study were scored on a five-point Likert scale (1 = Strongly Disagree to 5 Strongly Agree).

1. Organizational citizenship behaviors. Managers (i.e., general and supervisory) who worked directly with the participants were asked to rate each respondent on their propensity to engage in two types of citizenship behavior. The two types of OCBs consisted of a six-item loyalty scale representing OCBOs (e.g., "Actively promotes organization's products and services"; Van Dyne et al, 1994), and a five-item helping scale representing OCBIs (e.g., "Helps others who have been absent"; Podsakoff et aL, 1990).

The two OCB subscales were chosen based on civic citizenship theory, which argues that the success of an organization may be influenced by individual behavior associated with relational ties between the citizen/employee and the company (Graham, 1991). This theory guided the classification of the citizenship behaviors into OCBOs and OCBIs. Based on specific scale items and the subscale definition, helping behavior was judged to primarily benefit others in a direct manner (OCBI) (Williams and Anderson, 1991), whereas loyalty was judged to primarily benefit the organization directly (OCBO). The Cronbach's reliability estimate for the helping scale was .91, and for the loyally scale was .82.

2. Perceived organizational support. The measure of POS used in the second study contained a total of nine items and was identical to that presented in Eisenberger et al. (1990). As in Study 1, responses were rated on a five-point Likert-type response format (1 = Strongly Disagree to 5 = Strongly Agree; ot=.86).

Results. Correlations among primary study variables for study two are shown in the top half of Table 2. The correlations between POS and helping/OCBI (r=.15) and loyalty/ OCBO (r=.27) were both statistically significant at p<.01 (see Table 2). The hypothesis was tested by comparing the difference in the POS-OCBI and POS-OCBO correlations using a one-tailed t-test following procedures described by Cohen and Cohen (1983). As predicted, the POS-OCBO relationship was significantly stronger than that between POS and OCBI (t(234) = 1.80, p<.05).

Additionally, hierarchical regression analyses were conducted using OCBO and OCBI as separate dependent variables, with age, race, gender, and tenure used as control variables. These control variables were included based on past research findings suggesting relationships between each of these variables with both POS and OCBs (e.g., Eisenberger et aL, 1986; Stamper and Van Dyne, 1999; Van Dyne et a.4 1994; Wayne et al., 1997). As argued by Cohen and Cohen (1983), hierarchical regression tends to be a more robust analysis technique than simple correlational analysis in detecting possible relationships among variables.

The results of both hierarchical regressions are illustrated in Table 3. After controlling for age, gender, race, and tenure, the addition of POS at step two yielded a significant F and a positive beta for both OCBI and OCBO (helping/OCBI: (Delta)F=6.40, p<.01, =.15, R^sup 2^=.13; loyalty/OCBO: (Delta)F=15.97, p<.001, 1=.24, R^sup 2^=.17). The hypothesis was tested by determining the significance of the difference between the beta coefficients using procedures outlined by Cohen and Cohen (1983). Despite the fact that the difference between the beta coefficients was smaller than that between the correlation coefficients described above (due to variance associated with the control variables), the relationship between POS and OCBO was still found to be significantly stronger than that between POS and OCBI (t(234)=1.76, p<.05). Thus, the hypothesis was supported.

In addition, we ran moderated regression.analyses to determine if tenure moderated the two POS-OCB relationships. In step two of both equations (when POS and Tenure were entered as predictors), tenure and POS both showed significant main effects in predicting OCBO and OCBI. However, when the interaction term (POS X Tenure) was entered in step three, there was not a significant change of F (OCBI: change-F = .09, p>.05; OCBO: change-F = 2.76, p>.05) indicating that the interaction term did not explain significant levels of additional variance in the dependent variables. Since Tenure did not moderate either of the POS-OCB relationships, we have not added this information to the paper. In addition, we performed moderated regressions for the other demographics in Study 2 (i.e., age X pos, gender X pos, and race X pos) and there is no evidence of moderation for any of the POS-- OCBO or POS-OCBI relationships. Thus, we feel that these results are not influenced by the demographic variables.

Discussion. Results of this study further support the differential hypothesis. Because POS reflects the attachment employees experience with respect to the broader organization, the types of OCB employees will most likely engage in are those that benefit the larger organization rather than specific individuals. These results are consistent with those from study 1 and demonstrate that retaining a twodimensional approach to OCB is important in order to uncover differences that might otherwise be masked by a unidimensional conceptualization.

General Discussion

Organizational citizenship behave iors (OCBs) are important behaviors that benefit effective organization functioning. In recent years, OCBs have received a great deal of research attention in an effort to obtain a better understanding of both the antecedents and consequences of these behaviors (Organ and Ryan, 1995). Based on the theory of social exchange (Blau, 1964) and the norm of reciprocity (Gouldner, 1960), this study contributes to our understanding of the POS-OCB relationship such that POS appears to encourage certain types of citizenship behaviors more than others. More specifically, POS encourages those more impersonal OCBs that contribute to the broader organization (e.g., actively promoting the organization's products and services, sharing ideas for new projects or improvements), rather than those citizenship behaviors that benefit specific individuals directly, such as co-workers and supervisors. This study also contributes to the broader conceptualization of the OCB construct by demonstrating the value in distinguishing the target recipient of citizenship behaviors.

The two-study design included several features that enhance the robustness of the present findings and delimit potential boundary conditions of the POS-OCB relationship. Specifically, generalizability is enhanced because the findings from the independently-conducted field studies were generated using samples from both manufacturing and service sectors. In addition, we examined both supervisory and peer ratings of OCB and used different measures of OCB and different analysis techniques, yet obtained similar results.

The main focus of our study was that of differential prediction between POS and OCBs. Accordingly, both studies measured OCBs multidimensionally in terms of the target beneficiary of the behavior. The two dimensions were defined either as behaviors directed toward the organization (OCBO) or as behaviors directed toward a specific individual (OCBI). Results are in the predicted direction, but the lack of power associated with Study 1 likely resulted in conservative estimates of results. That is, for Study 1, given the sample size of 69 using listwise deletion, alpha set at .05, and estimating a moderate effect size of .30 as suggested by Cohen (1988), the estimated power is only .71. However, in the second study, where we had a much larger sample and subsequently greater power, we clearly demonstrate that the relationship between POS and OCBO is stronger than that between POS and OCBI.

Implications and Future Research Directions

The current study does more than replicate prior research concerning the relationship between perceived organizational support and a general class of citizenship behaviors. Unlike previous work in the area, the present study provides evidence of differential prediction between POS and different facets of OCB. The difference in strength of these relationships is important because it suggests that employees seem to distinguish their effort levels based on the beneficiary of citizenship behaviors.

The findings of the present study not only speak to the POS-OCB relationship, but also suggest that a re-examination of past findings may be warranted in order to determine whether the relationships reported in the literature are accurate estimates of relationships between known antecedents and general OCB. Further, the results from this study have implications for the study of other forms of discretionary behaviors such as those that harm the organization or specific individuals. In this way, recent research on deviant behaviors (Robinson and Bennett, 1995) and organizational retaliatory behaviors (Skarlicki and Folger, 1997) may benefit by making similar distinctions between those behaviors directed toward the organization more broadly and those directed toward specific others.

Future research aimed at differentiating between OCB dimensions will offer stronger support to the findings reported in the present study. Research on differential prediction may look at other factors within the organization that may differentially affect the underlying employment exchange relationships and the prediction of OCB. Although this study only examined the effects of a single predictor (POS), other aspects of perceived support may offer convergent and discriminant evidence of differential prediction for OCB criteria. Specifically, support systems that pertain to the organization such as affective organizational commitment (Eisenberger et aL, 1986; Mowday et al, 1969) and situational variables attributed to the larger organization such as working conditions or other discretionary programs (e.g., child care) may be stronger predictors of OCBOs.

Conversely, variables that capture the quality of the relationship between employees and specific individuals at work such as supervisors may be better predictors of OCBI. For example, according to the theoretical perspective advocated here, variables such as perceived supervisory support (Kottke and Sharafinski, 1988), leader-member exchange (LMX) (Moorman et al., 1998) interactional justice (Skarlicki et al., 1999), and work group cohesion (Kidwell et al., 1997) should all be more strongly related to those OCB that benefit specific individuals such as one's coworkers or supervisor.

Although the academic literature on the contribution of OCB to overall organizational functioning has impacted managerial practice, there has been little recognition of the importance of viewing OCB criteria in terms of the target beneficiaries of the behavior. Explicit recognition of the different OCB beneficiaries (OCBO versus OCBI) may aid in focusing efforts to encourage OCBs. For instance, if the primary objective is to improve the extent to which employees engage in those behaviors that benefit the organization (e.g., contributing suggestions for organizational improvement), strategies that directly foster perceptions of organizational support (e.g., employee recognition programs, flextime) may take precedent. Conversely, if an organization is more concerned with promoting OCBIs such as inter-employee cooperation, greater attention should be devoted to developing stronger working relationships between employees and both their peers and direct supervisors (e.g., supervisory interactional justice training; Skarlicki and Latham, 1997). Specifically, future work is needed to examine whether employees differentiate OCBO sources such as team leaders, supervisors, or managers and between OCBI categorizations such as team member, co-worker, and friend.

The findings from our studies also have practical application for organizations. Specifically, our results suggest that employees are more willing to put forth extra effort in performing tasks that lie outside the boundaries of traditional job descriptions when they perceive that the organization cares about their well-being and when they perceive that their contributions are both acknowledged and valued. Promoting and fostering a supportive organizational climate may receive great benefits in terms of employees expanding their repertoire of behavior to include task as well as good citizenry actions. Thus, organizations may realize an increase in employee productivity without experiencing an increase in labor cost.

There are several ways that an organization may increase employee perceptions of support, based on the POS and organizational commitment literature (e.g., Eisenberger et aL, 1986; Mowday et aL, 1969). These include, but are not limited to: 1) giving employees timely and positive feedback about desired work behavior, 2) rewarding desired behavior through appointments to special projects or through employee-of-the-month programs, 3) job sharing or other nontraditional work programs for working parents, 4) on-site child care, 5) on-site physical fitness facilities, and 6) special benefits for long-term or chronic illnesses such as cancer.

Limitations and Future Directions

The present analyses are the result of combining data from two independently conducted field studies in different types of industries. In addition, the results were consistent even when using different OCB measures, analyses, and rating sources. However, it is important to note that these studies are cross-sectional in nature and, as such, care must be taken in drawing causal interpretations. Furthermore, although these findings support a broad two-dimensional approach to conceptualizing OCBs, it is possible that there are more fine-grained distinctions within the OCBO and OCBI categorizations. Finally, because these two studies used different measures, direct comparisons are somewhat difficult. For example, the scale reliabilities found in Study 1 were slightly lower than in study 2; therefore, it is possible that stronger results may have been found in Study 1 if more reliable measures had been used. Nevertheless, given that lack of reliability attenuates observed relationships, the results found in Study 1 would most likely be stronger had more reliable measures been used. Thus, we feel that despite these limitations, the combined findings for Study 1 and Study 2 are important evidence for the differential prediction of POS on OCBs.