Document View

In a management context, there is a strong belief that employees view outcomes more favorably when they result from fair procedures rather than unfair procedures. Yet academic and popular accounts indicate that some procedural enhancements have the potential to backfire - i.e. process improvements can unexpectedly bring about an increased sense of injustice, thereby doing more harm than good. This study attempts to provide an integrative framework for understanding this so-called "frustration effect": when does procedural justice enhance, and when does it diminish, distributive justice. To better understand the occurrence of frustration effects, the authors focus on Referent Cognitions Theory (RCT). Using structural equation modeling in a multi-sample framework, evidence suggests that the trigger for the so-called 'frustration effect" depends on a belief in one's entitlement to a preferable referent outcome. If outcomes are bad enough, and fail to meet individual expectations for a more desirable alternative, procedural fairness does little to enhance perceptions of distributive justice. The results are discussed in connection with practices used to allocate scarce goods and resources [PUBLICATION ABSTRACT]

INTRODUCTION

The research literature provides reasonably consistent information about two components that are crucial in deciding if the allocation process is just or fair: distributive justice (which focuses on the perceived fairness of outcomes, Adams, 1965) and procedural justice (which concerns the justice of the process used to determine those outcomes, Greenberg, 1990). Additional research efforts suggest not only that both process and outcome dimensions play a role in determining fairness perceptions, but that they play an interconnect role. In other words, individual reactions to outcomes may be enhanced by the perceived fairness of procedures used to distribute those outcomes (fair process effect), and conversely, individual reactions to procedures depend on the perceived fairness of outcomes that were obtained through those procedures (fair outcome effect).

There is evidence that procedural and outcome enhancements don't always have positive implications for justice judgments. For example, within the relative deprivation literature (e.g. Cropanzano & Randall, 1995; Folger & Martin, 1986) it has been argued that an improved outcome may provide a basis for increasing expectations. Under these circumstances, rising expectations create a new benchmark or reference point against which outcome allocations are measured. If improvements fail to meet these new benchmarks, the potential exists for individuals to feel "deprived" or dissatisfied with their improved outcomes.

Other researchers have found situations where a process improvement fails to have a positive impact on justice judgments. In particular, in a small but persistent number of studies researchers have found that increasing procedural justice can actually yield a decline in perceptions of distributive fairness, suggesting that the fair process effect doesn't always work. Folger and his colleagues labeled this finding the "frustration effect" (Folger, 1977; Folger, Rosenfield, grove & Corkran, 1979; Folger, Rosenfield & Robinson, 1983). The research literature offers few insights into the causes of this phenomenon. Certainly a disproportionate rise in employee expectations would help to explain such a finding. This explanation is consistent with the conclusions from other research that has observed increased expectation levels with the introduction of a fairer procedure (see Harlos, 2001; Robinson & Rousseau, 1994).

From a process improvement perspective, as organizations become increasingly reliant on systems that allow employees to express their voice (open door policies, grievance procedures, suggestion boxes, etc...), the question remains as to whether the introduction of a procedurally enhanced system will only intensify employee perceptions of unfairness. This article adds to this literature by offering insights as to how and why such procedural innovations could potentially fail to produce their intended effects. By analyzing conditions under which procedural fairness enhancements can actually decrease perceptions of distributive fairness, this study directs attention to a context that has been historically neglected in the research literature, a situation where process improvements can unexpectedly bring about an increased sense of injustice. Findings reported here suggest that the frustration effect is a natural and theoretically expected outcome of a generalized model of justice - an outcome that occurs predictably under specific conditions. In the current article we try to show that reactions to process enhancements depend in large part on an individual's frame of reference and conceptualize our understanding within the framework of Referent Cognitions Theory (RCT; Cropanzano & Folger, 1989; Folger, 1986, 1987, 1993; Folger & Cropanzano, 1998, 2001). We begin with a brief introduction to RCT, followed by an overview of relevant frustration effect research. After a test of our conceptual model, we discuss some practical implications along with avenues for future research.

LITERATURE REVIEW

Referent Cognitions Theory

In the language of RCT, alternatively imaginable outcomes are labeled referent outcomes. According to the theory, as an individual recognizes the existence of alternative end results, frames of reference are established to those alternative states of affairs. In a "high-referent" condition there is awareness to the possibility of an alternative results in which a more desirable outcome would have been obtained, this then becomes the referent outcome. Conversely, in a "low-referent" condition, individuals believe that there was never any possibility of obtaining the more desirable outcome. RCT argues that an unfavorable evaluation of an outcome is more likely in a high-referent condition, where an individual believes that a preferable referent outcome could have been obtained. In this situation, the more desirable outcome likely resembles a missed opportunity and serves to trigger discontent over disparities between existing and referent outcomes. This paper hypothesizes and tests a proposition that this discontent likely turns to resentment or frustration when an individual perceives unreasonable or unjustifiable allocation discrepancies, thereby minimizing the potential impact of any procedural enhancement.

Commonalities in Frustration Effect Studies

Three conditions appear to be necessary to trigger the frustration effect. First, in studies where a frustration effect was observed, outcomes tend to be relatively negative and inequitable. second, the studies all showed some form of improved procedures with inadequate improvements in outcomes. Third, there is reason to believe that individuals who exhibit the frustration effect have confirmation, perhaps enhancing certainty about the unjustifiable nature of an inequity. When these three conditions are met, it appears that the improved procedure has actually been associated with reductions in perceived fairness and lower levels of satisfaction.

Table 1 highlights those studies that have reported evidence of a frustration effect. It includes a brief description of the method by which each study gives consideration to the three common elements described above. The first condition, negative outcomes, has been represented in a variety of ways, including: a failure to redistribute excess tax payments (Thibaut, Friedland, and Walker, 1974); inequitable monetary distributions (Folger, 1977; and Folger, et.al., 1979, study 2); a biased allocation of rewards (Folger, et. al., 1979, study 1); deposition for a possible felony trial rather than conviction of a misdemeanor (Austin et.al, 1981); rejection for an employment opportunity (Kulik & Clark, 1993); and, an unusually low salary offer (Porter, Conlon, and Barber, 2004).

The second condition, procedural enhancement, has also been achieved with a variety of manipulations. Most frequently recipients have been given voice by either: allowing subjects to vote (Thibaut, et. al., 1974); or permitting subjects to share their ideas (Folger, 1977; Folger, et.al. 1979 study 1 & 2). In addition, one study used an adversarial rather than an inquisitorial legal procedure (Austin, et. al., 1981) and another study enhanced accuracy by making use of a second confirmatory test to gather information (Kulik & Clark, 1993). A final study used both voice - allowing job applicants an opportunity to express their concerns about the amount of a starting salary offer, and, information - providing explanations regarding how the starting salary amount was determined. (Porter, Conlon, and Barber, 2004).

The third condition common to these studies was inclusion of a manipulation which might be expected to enhance a subject's belief in their entitlement to a more desirable referent outcome. Of these three common characteristics, the referent manipulations have been the most diverse. We argue that each of these manipulations has the effect of increasing a subject's certainty as to the inequitable nature of the actual outcome by drawing attention to their right to a favorable referent outcome. As seen in the table, variations in the réfèrent were realized by: confirmations of injustice from other subjects/coworkers (Thibaut, et. al., 1974; Folger, et. al. 1979, study 1 & 2); favorable partial-adjustments of a reward allocation following voice (Folger, 1977); subject selection based upon prior beliefs about procedural inaccuracy (Kulik and Clark, 1993); and, introductory information defined a reasonable range for an outcome (Porter, Conlon, and Barber, 2004). Under each of these circumstances, providing the subject with corroborating evidence that supports the perception of unfairness highlights the unjustifiable nature of the allocation. We predict that when these conditions exist, an improved process is less likely to lead to improvements in perceptions of distributive fairness or, conversely, that an improved outcome is less likely to lead to improvements in perceptions of procedural fairness.

Enlarge 200% Enlarge 400%

A Catalyst for Frustration: Resource Scarcity

Deutsch (1985) contends that perceived fairness is especially important during times of scarce resources. The scarcity sensitizes individuals to their outcome share, causing individuals to scrutinize allocation practices and to react negatively when expectations are not met. When resources are scarce, acknowledgements (from the media, other recipients, or through first hand experience) increase the belief among recipients that a negative outcome/process must be unjust. Consequently, in the face of resource scarcity, we might expect to observe the traditional frustration effect; where enhancements to procedural fairness are likely to have little or no effect on perceived outcome favorability.

The commonalities we documented in frustration effect studies seem likely to arise where shortages exist. Specifically, when confronted with a resource of limited quantity, individuals are likely to make the following observations: First, given that resources are not sufficient to satisfy the wants of all recipients, many individuals will experience poor or negative outcome distributions. Second, allocators concerned over the scrutiny of their allocation practices are likely to adhere to relatively fair standards of procedure. The perception of a fair procedure builds an expectation for a desirable referent outcome, one that isn't subsequently realized. Finally, with scarcity, individuals often receive information from others that increases the certainty of their belief that an inferior outcome was either unjustifiable or avoidable. Such acknowledgements increase the likelihood of a frustration effect. Given the tendency of recipients to scrutinize process when there is a scarcity of resources, it seems likely that decisionmakers will use constant and standard procedures to avoid any perceptions of an impropriety. Prior evidence makes it clear that the pairing of a better process with the same bad outcome can produce a frustration effect. We anticipate that the pairing of a constant process with slightly improved outcomes may also yield similar frustrations, especially where the outcome improvement remains inferior to an imaginable, justified alternative. Under conditions of limited resources, we expect that outcomes frequently fail to improve at a fast enough rate. Given the inability to match rising expectations, scarcity provides a significant obstacle to what are believed to be reasonable referent outcomes. Here, an improved outcome is likely viewed as marginal at best and therefore remains inequitable. Under conditions of scarcity, if the improvements in the outcome are not sufficient to reestablish equity they are likely to have little or perhaps even negative effects on perceptions of procedural fairness.

METHOD

Sample.

2053 surveys were sent to a stratified random sample in a large northeastern university. Stratification occurred along three dimensions: status (i.e. student n=1317 vs. faculty/staff n=736); Campus (North, South, or commute between both); and experience with the parking process (i.e. appealed a parking ticket n=1053 vs. no appeal n=1000). A total of 786 surveys were returned with a distribution of 390 (30%) student and 396 (54%) staff.

Procedure.

Subjects were asked to supply their names in order to match responses with university parking records. The survey included forty five questions about process, outcomes, justice (procedural & distributive), and satisfaction (outcome & institution). Items on the survey were measured on 5-point Likert scales (strongly agree to strongly disagree). The measures used to operationalize each variable were developed based on literature in the area of organizational justice

In order to test the hypotheses, it is also necessary to provide some differentiation between resources allocated to different groups of recipients. Table 2 reveals three objective outcome measures used to differentiate between high, moderate and low scarcity groups: lot demand, wait for a parking place and parking tickets. Data from this table reveals an environmental context characterized by scarcity. A univariate analyses of variance revealed that the differences in perceived outcomes between bad, moderate, and good outcome groups were statistically significant (F=44.73, p=.00)

Enlarge 200% Enlarge 400%

RESULTS

Path Analysis - Group Models. A simple multiple-sample analysis (Schumacker & Lomax, 1996) was used to determine the extent to which the modified conceptual model simultaneously fit high, moderate, and low scarcity groups. In order to statistically determine whether the hypothesized parameters are equivalent across scarcity groups, three separate analyses were performed. Statistical parameters for each analysis are provided in Table 3.

Enlarge 200% Enlarge 400%

First, structural parameters were held fixed across groups. As seen in the table, parameter estimates represent a satisfactory fit between the modified conceptual model and the data with a χ^sup 2^ /d.f.=1.72, RHO=.900, TLI=.955, GFI=.964, AGFI=.942, AIC=197.99 (compared to 216.00 in the saturated model); and RMSEA=.031. In order to test the hypothesized relationship, two additional analyses were performed. First, the distributive justice procedural justice link was allowed to vary across groups. This created a 14.17 point gain in chi-square for 4 degrees of freedom (p < .007). Parameter estimates in the path model offered a small improvement in the fit statistic. second, the justice - outcome satisfaction links were then allowed to vary, resulting in a 20.27 point gain in chisquare for 4 degrees of freedom (p<.000). Once again, results reveal a small improved fit with a χ^sup 2^ /d.f.=1.36, RHO.921, TLI=.978, GFI=.975, AGFI=.953, AIC=179.55, and RMSEA=.022

Enlarge 200% Enlarge 400%

Group models and their structural parameters (path coefficients) are presented in Figure 1. As predicted the distributive - procedural justice link weakens as outcomes improve. Specifically, distributive justice has a smaller influence on procedural justice when outcomes are moderate (.43) as opposed to when outcomes are bad (.70). Under these circumstances, the pairing of a constant process with a slightly improved outcome, likely elicits frustrations which lessen the impact of what otherwise might be considered a fairer outcome. Interestingly, however this effect is only observed for moderate improvements in outcomes. Distributive justice regains its influence on procedural justice as outcomes move from moderate to good (.67). Interestingly, the relative influence of distributive justice on outcome satisfaction is greater as outcomes worsen. As seen in the figure, when outcomes are bad, distributive justice has a much larger influence on parking satisfaction (.56), than procedural justice (.27). Conversely, when outcomes are good, procedural justice has a higher level of influence on outcome satisfaction (.64) than distributive justice (.45). Clearly, as expected, when resources are scarce, individuals focus attention primarily on distributive norms when evaluating overall satisfaction with an outcome.

Enlarge 200% Enlarge 400%

DISCUSSION

Results from this study highlight the importance of considering how unjustifiable inequities impact perceptions of procedural justice. The present study found that when resources are scarce (a situation where perceived inequities are likely), moderate improvements in outcomes are less likely to lead to improvements in perceptions of procedural fairness. Specifically, the results show that during times of scarcity, as outcomes improve to moderate levels, distributive justice has less of an impact on procedural justice. However, as outcomes continue to improve, a more positive link is reestablished. There may be limits to the long held beliefs about the pervasive influence of procedural justice. Results from this study mirror those in frustration effect studies. Procedural improvements hold out much more promise under conditions of high or moderate outcomes than when outcomes are bad. For the frustration effect, bad outcomes that became somewhat better didn't lead to more favorable reactions to the process. We too confirm that no amount of process can remedy very bad outcomes. These results highlight the importance of considering the environmental context when applying referent cognitions theory.

Research to advance the notion developed here might take several directions. First, additional research is needed to confirm the existence of certain situational conditions associated with the frustration effect. Although this study attempts to lay out the specifics of this phenomenon, additional empirical support is needed to further support our assertions. Although this study used proxy measures of expectations, we believe that much of what is observed in frustration effects could be explained in the context of expectations. Marginal improvements to the outcome only serve to legitimize recipient claims over an outcome. This kind of acknowledgement sends a cue to the recipient that an allocator somehow agrees with their concerns. However, an agreement without addressing the substantive nature of a recipient's demands is likely to be perceived very negatively. Clearly, researchers must focus their attention on the identification of specific factors that influence expectations in an allocation situation.

Other limitations should prompt continued research in this area. Specifically, it should be noted that the observed effects rely on the existence of different groups, each with different outcomes. Consequently, results stem from aggregate group responses rather than experimental manipulations of specific outcome measures. In addition, given the aggregate nature of the data, this study could only demonstrate contextual effects rather than explain how they occur. This limitation is consistent with one raised by Mossholder, Bennett & Martin (1998), who advocated the use of longitudinal research to better understand the processes that underlie contextual justice effects.

Finally, our report of a frustration effect is limited to the extent that an improved outcome does not result in a complete reversal of the sign on the distributive - procedural justice link. Prior studies have reported a negative relationship between procedural and distributive justice with increments in perceived fairness (e.g. voice). In this study, increments in outcome favorability only serve to reduce the magnitude of this relationship. Nonetheless, given a significant degradation in the strength of the distributive - procedural justice link, we feel justified in labeling these findings as a frustration effect. From a practical standpoint, these findings suggest that organizations wishing to enhance fairness perceptions can only do so with substantive changes. We therefore caution against tokenism. Marginal improvements should not be used in an attempt to make up for an otherwise limited outcome distribution. From an employer's perspective, it is better to say the cupboard is bare than to make small, insignificant changes to an allocation. Furthermore, in the face of sever outcomes brought on by resource scarcity, managers might be cautioned against paying too much attention to process issues. Specifically, they should hesitate to recommend the inclusion of opportunities for process improvements, particularly where fairness represents a primary concern, i.e., when resources are scarce. Only as outcomes gradually improve do procedures themselves become important.