Human Service Integrative Course

Journal of Applied Psychology 1998, Vol. 83. No. 5, 788-797

Copyright 1998 by the American Psychological Association, Inc. 0021 -9010/98/$3.00

The Value of Voice in Participative Decision Making

James E. Hunton University of South Florida

Thomas W. Hall and Kenneth H. Price University of Texas at Arlington

Relying on concepts found in prospect theory (D. Kahneman & A. Tversky, 1979), the value function of voice-based participation (i.e., the relationship between the amount of voice received and the value attached to that quantity) was examined. In keeping with tenets of prospect theory, the value function of voice exhibited a nonlinear pattern. Points were identified in which voice displayed significant improvements and diminishing marginal returns on response measures of process fairness, decision control, and outcome satisfaction. Task meaningfulness, a moderator of voice-based participation, did not change the general shape of the value function but did influence the intensity of participant reactions at low and high levels of voice. Voice influence, a second moderator of voice- based participation, had minimal impact on participant responses.

Participation by stakeholders in decision processes has

received considerable research attention, particularly in

the fields of psychology and management (e.g., Barley &

Lind, 1987; Locke & Schweiger, 1979; Vroom & Jago,

1988). From a procedural justice perspective, participa-

tion by affected parties through voice (i.e., the opportu-

nity to express one's opinions, concerns, and preferences)

increases the perception of process fairness (e.g., Lind &

Tyler, 1988).

The instrumental model of procedural justice posits that

voice increases perceptions of fairness because it provides

for the possibility of producing a favorable .outcome

(Barley & Lind, 1987; Lind, Kanfer, & Earley, 1990). In

the relational model, the solicitation of voice increases

perceptions of fairness because participants feel they are

being treated with the dignity, politeness, and respect ap-

propriate for full group members (Lind, 1995; Tyler &

Lind, 1992). Although factors mediating voice effects dif-

fer for instrumental and relational models, both frame-

works agree that procedural fairness is a critical element

associated with decision acceptance and commitment

across a number of different situations (Greenberg, 1990;

Shapiro, 1993).

Given the key role of voice in procedural justice judg-

ments, the primary focus of this study is to provide a first

James E. Hunton, School of Accountancy, University of South Florida; Thomas W. Hall, Department of Accounting, University of Texas at Arlington; Kenneth H. Price, Department of Manage- ment, University of Texas at Arlington.

All authors contributed equally to this article. Correspondence concerning this article should be addressed

to Kenneth H. Price, Department of Management, University of Texas, UTA Box 19467, Arlington, Texas 76019. Electronic mail may be sent to [email protected].

approximation of the value function of voice based on

concepts found in prospect theory (Kahneman & Tversky,

1979). As voice is examined along a continuum from zero

to 100% participation, findings can help to identify points

where voice displays significant improvements and dimin-

ishing marginal returns on consequence variables (i.e.,

perceived fairness of the procedures, perceived decision

control, and outcome satisfaction). Current theory and

research on the magnitude of voice are reviewed in the

next section.

Magnitude of Voice

Research has examined the magnitude of participant

voice as a facet of the meaningfulness of the voice solici-

tation effort. Meaningfulness of voice can be viewed along

three dimensions: (a) when voice occurs in the decision-

making process, (b) the degree of congruence between

desired (expected) and actual levels of voice, and (c) the

amount of voice that is permitted across different stages

of decision making (Hunton & Price, 1994). Earley and

Kanfer (1985) have reported that voice-based participa-

tion in earlier rather than later stages of decision making

has a stronger, more positive impact on satisfaction and

performance. Doll and Torkzadeh (1988) have reported

that participation efforts are dysfunctional when desired

participation exceeds actual participation. Cooper and

Wood (1974) have found that voice in all stages of deci-

sion making (planning through postimplementation) re-

sults in more positive perceptions of involvement than

voice occurring in only one stage.

Consequently, the literature is generally supportive of

a positive relationship between the meaningfulness of the

voice solicitation and the perception of fairness. In some situations, however, more voice may not be better. Van

788

RESEARCH REPORTS 789

den Bos, Vermut, and Wilke (1996) have reported that participants who expect no voice yet receive voice per- ceive the process as less fair and perform at lower levels than participants who neither expect nor receive voice. These results argue that the positive impact of higher levels of voice is reduced when the solicitation of voice is inconsistent with expected procedures (Van den Bos et al., 1996). In spite of moderating variables of this type, a value function approach may be of considerable use in understanding the consequences of different levels of voice-based participation when higher levels of voice is expected to have more positive outcomes. The value func- tion approach is discussed next.

Value +

Reference Point

.Gain

The Value Function of Voice

Decision theory models of human judgment typically incorporate a function depicting the relationship between

the quantity of a commodity received and its value (i.e., a value function). A commodity can be anything valued by an individual, including money, leisure time, or even voice-based participation. In modeling the value of a com- modity, prospect theory (Kahneman & Tversky, 1979) asserts that individuals code outcomes as deviations (e.g., gains and losses) from a reference point. This reference point may be the amount of a commodity that a person expects to receive in a given situation. Although people vary in their expectations, the commodity value function

for each person is thought to be nonlinear, with a convex shape below the reference point and a concave shape above the reference point. The rationale for a convex shape in the loss domain (below the reference point) and a concave shape in the gain domain (above the reference point) is based on the principle of diminishing marginal significance. In particular, people are thought to be most sensitive to variations in the general region of the refer- ence point (Hogarth, 1989). As the amount of a commod- ity realized moves further away from the reference point, the marginal significance of the incremental loss or gain declines; thus, smaller and smaller changes in perceived value of each additional increment of commodity received are produced. This functional relationship is illustrated in Figure 1.

To examine the value function for voice, a scenario

methodology that describes the need to make 20 different decisions regarding the attributes of a new computer sys- tem is used. Depending on the treatment group, partici- pants receive different quantities of a voice commodity by voicing their preferences to their supervisor for either 0, 5, 10, 15, or 20 attributes of the new computer system. Our expectations are that higher levels of voice will result in more positive perceptions of process fairness, decision control, and outcome satisfaction, as reported in other voice studies (e.g., Barley & Lind, 1987; Greenberg,

Figure I. The proposed value function from prospect theory

(Kahneman & Tversky, 1979)

1990). However, within these parameters, the actual value function for voice is expected to be nonlinear based on prospect theory (Kahneman & Tversky, 1979). Conse- quently, this study examines whether the voice value func- tion exhibits nonlinear tendencies and, if so, whether the approximate shape of the curve is convex, concave, or both.

Experiment 1

Method

Participants and procedures. Participants (N = 105) were

recruited from introductory courses in management at a large

southwestern university. The modal age of the participants was

21 years, 58% of the respondents were female and 42% were

male, and 26% worked full-time, 63% worked part-time, and

11% did not report employment.

The experiment was conducted over seven sessions each with

10 to 20 participants. Within each session, participants randomly

received one of five versions of a case scenario. The case in-

structed participants to assume the role of a department manager

whose company was in the process of updating its personal

computers. Each participant was informed that his or her super-

visor had developed a list of 20 attribute choices (e.g., size of

hard drive and speed of the processor) that the supervisor would

make before the purchase decision and that the participant had

seen this list and judged each choice to be equally important.

Participants were told to assume they spent approximately 85%

of their workday using personal computers. The case manipu-

lated the independent variable, the magnitude of voice, across

five levels by telling participants to assume that their supervisor

either had not asked them to express their opinions and prefer-

ences for the 20 choices (mute condition) or that they had been

asked to express their opinions and preferences for 5, 10, 15,

or 20 of the 20 choices.

Dependent measures and manipulation check questions. All

dependent measures were assessed using 9-point bipolar scales.

790 RESEARCH REPORTS

Items adopted from other justice studies (e.g., Earley & Lind,

1987) included three questions that measured perceptions of

procedural justice (a = .87) and decision control (a = .69).

Participants were asked to rate the procedures used to determine

the attributes of the new computer system on scales ranging

from unfair (1) to fair (9), unjust (1) to just (9), and strongly

disagree (1) to strongly agree (9) for the fairness item measur-

ing satisfaction with the procedures. Perceptions of decision

control were measured with scales ranging from no control

(1) to complete control (9), and no influence (1) to complete

influence (9). The third control item asked the extent to which

participants felt that their opinions, concerns, suggestions, and

preferences were considered in determining attributes of the

new computer system on a scale ranging from no extent (1)

to a great extent (9). We adopted three questions measuring

anticipated outcome satisfaction, a = .93 from Doll and Torkza-

deh (1988). Participants were asked to anticipate their degree

of satisfaction with and their feelings toward their new computer

system on scales ranging from not satisfied (1) to completely

satisfied (9). The third satisfaction item asked whether they

would be satisfied with their new computer on a scale ranging

from strongly disagree (1) to strongly agree (9).

A manipulation check question for voice solicitation indicated

that 70% of the respondents were able to correctly identify the

number of computer attributes on which voice was solicited.

Although the accuracy rate was less than desired, supplemental

analysis indicated that research findings did not differ signifi-

cantly when participants who incorrectly answered the manipu-

lation check question were excluded from the analyses.

Results

Relationship among dependent measures. Correla-

tions among the dependent measures ranged from .74 for

the measures of fairness and decision control to .76 for

the measures of fairness and satisfaction. Given the inter-

correlation and the similarity of the results across the

dependent measures, only the results for perceptions of procedural fairness are reported.

Shape of the value function. Tb identify whether the

shape of the voice value function was nonlinear across

the five levels of voice, a trend analysis was performed for the fairness response measure. Using orthogonal

polynomial coefficients provided by Beyer (1988), spe-

cialized contrasts testing for linear, quadratic, and cubic

effects were performed. Results documented a nonlinear

value function for voice on the response measure of

procedural fairness that included significant linear (p

- .022), quadratic (p = .007), and cubic (p = .007)

components. The shape of this value function is plotted

in Figure 2.

Additional tests. A one-way analysis of variance

(ANO\%) followed by Duncan's multiple-range tests

(a = .05) was used to examine whether the approximate

shape of the value function was convex, concave, or

both. For the measure of procedural fairness, there was

a significant main effect for the magnitude of voice,

F(4, 99) = 9.08, p < .001. Cell contrasts documented

a significant stepwise increase in perceptions of fairness

from the mute condition (M = 3.4) to the solicitation

of voice on 5 attributes (M - 5.8). Increases in the

magnitude of voice from 5 to 10, from 10 to 15, and

from 15 to 20 decisions did not result in significant

stepwise increases in perceptions of fairness (Ms = 5.9,

5.7, and 6.0, respectively).

9

8

7 (ft

I 5

•I 4

2

1

0

10 15 20

Magnitude of Voice

Figure 2. Shape of value function for perceptions of procedural fairness for Experiment 1.

RESEARCH REPORTS 791

Discussion

Voice as a value function. Findings from this first experiment are supportive of a nonlinear relationship be- tween the magnitude of voice and perceptions of proce- dural fairness. Inspection of Figure 2 indicates a concave function with little change after participants voiced on 5 of 20 attributes. In keeping with prospect theory, solicita- tion of voice above a specific point, in this case 5 deci- sions, did not result in significant increases in perceptions of fairness. The relatively large stepwise increases that occurred when voice magnitude moved from no voice (mute condition) to 5 decisions demonstrates that a mini- mum amount of voice was necessary to ignite increases in perceptions of procedural fairness. In this regard, pro-

viding participants with some voice as opposed to no voice seemed to separate relatively immaterial from more substantive levels of voice-based participation. This argu-

ment is strengthened if one assumes that the midpoint on the scale represents a neutral point for most participants.

Experiment 2

The focus of the second experiment was to replicate the nonlinear value function for voice with a sample of midlevel executives. In addition, this study examined two

important moderators of voice-based participation and their potential impact on the perceived value of voice. Both moderators, task meaningfulness and the influence of voice (Folger, 1987; Hunton & Price, 1997), are dis- cussed in the following sections.

Task Meaningfulness

Task meaningfulness arises from the decision context, refers to the personal salience of the task on which voice occurs, and is independent of the activity of expressing one's voice. Reactions to voice-based participation are expected to be more extreme under conditions of high task meaningfulness and less extreme under conditions of low task meaningfulness (e.g., Hunton & Price, 1997). Consequently, more positive perceptions of procedural fairness, decision control, and outcome satisfaction are expected in the high- compared with the low-task-mean- ingfulness condition when high levels of voice are solic- ited. At low levels of voice, more negative perceptions of fairness, decision control, and outcome satisfaction are expected in the high- compared with the low-task-mean-

ingfulness condition. Although the anticipated Voice Mag- nitude X Task Meaningfulness interaction is expected to intensify reactions to different levels of voice, it is un- known if or how the nonlinear shape of the value function for voice found in the first experiment will change.

Differential reactions to voice, as a function of task meaningfulness, are expected because (a) salient deci-

sions have a more powerful impact on the participants

(Brickner, Harkins, & Ostrom, 1986), (b) participants

typically desire more input on personally salient tasks (Vroom & Jago, 1988), and (c) in keeping with the instru- mental model of procedural justice, the expression of voice is valued because it is perceived to increase the possibility of securing favorable outcomes (Thibaut & Walker, 1975). Consequently, more personally salient de- cisions are expected to result in higher outcome satisfac- tion when voice is high because there is more to gain and

lower levels of satisfaction at low levels of voice because there is more to lose.

When the response measure of decision control is con- sidered, the objective amount of control over a decision

does not change as a function of the meaningfulness of the decision. However, increases in the perception of con-

trol have been reported when the effects of noninstrumen- tal voice (i.e., voice having no influence on the decision outcome) were examined (Lind et al., 1990). In keeping with the relational model of procedural justice, research- ers theorize that heightened feelings of inclusion resulting from noninstrumental voice lead to increases in perceived decision control (Lind et al., 1990). In this regard, varia- tions in the level of voice may result in higher feelings

of inclusion when the decision is higher in meaning- fulness. In turn, these perceptions may enhance percep- tions of fairness, directly or indirectly, through heightened perceptions of control.

Influence of Voice on the Decision Maker

A second moderator of current voice solicitation efforts is the influence of outcomes of past voice solicitations. Whereas higher levels of voice are expected to lead to more favorable outcomes, previous participant experience with decision makers may reinforce or negate such expec- tations. Consequently, at each level of voice (5, 10, 15, or 20 decisions), perceptions of process fairness, decision control, and outcome satisfaction are expected to be higher under high, as compared with low, past participant influence.

When there is a history of following rather than ignor- ing the voice of affected parties, expectations of outcome satisfaction are generally higher (Miller & Crush, 1988). In keeping with the instrumental model of procedural jus- tice, researchers have reported corresponding decreases in decision control and fairness perceptions, as well as increases in frustration, when expectations of participative

input are not met (e.g., Hunton, 1995; Lind & Tyler, 1988). Expected differences in perceptions of fairness, decision control, and satisfaction as a function of past outcome expectations are also consistent with the rela- tional model of procedural justice (e.g., Lind & Tyler, 1988). Soliciting voice should have fewer symbolic bene- fits when voice has not been followed in the past.

792 RESEARCH REPORTS

Expectations for participants in the mute condition are

less clear. Based on referent cognition theory (e.g., Folger,

1986, 1987), no voice may be especially frustrating and

cause resentment when past recommendations have been

followed. Because this assumes that participant input had

been allowed in past decisions, it may be easy for these

participants to imagine more positive outcomes that could

have been obtained if only different procedures were used

(i.e., higher levels of voice were permitted). In contrast,

when participant voice has not been followed in the past,

frustration and resentment may be lower when voice is

not allowed. It may be more difficult for these participants

to imagine more positive outcomes resulting from higher

levels of voice because no one is listening. As levels of

frustration and resentment increase, one would expect de-

creases in perceptions of fairness, decision control, and

outcome satisfaction.

Method

Participants and experimental design. Participants were

members of a professional association of accounting and finan-

cial managers from the publishing, paper, and chemical indus-

tries. Of the 1,000 survey instruments distributed, 667 usable

surveys were returned. The average age of respondents was 37

years, 58% of the population was male and 42% were female,

and supervisors, middle managers, and upper managers repre-

sented 43%, 36%, and 21% of the sample, respectively.

We used a 5 (magnitude of voice) X 2 (task meaningfulness)

X 2 (voice influence) fully crossed factorial design. Participants

were randomized to treatment conditions, and the number of

respondents per cell ranged from 29 to 37. Preliminary tests

indicated no significant difference in mean respondent age, in-

dustry, gender, or managerial position by treatment condition.

Procedures and experimental manipulations. The experi-

mental materials were distributed and returned through profes-

sional association Internet E-mail systems available to all poten-

tial respondents. The scenario that participants read manipulated

the magnitude of voice (i.e., mute, 5, 10, 15, and 20 decisions)

in the same manner as reported in Experiment 1. We manipulated

task meaningfulness by telling participants that their personal

computers were used for either 85% or 15% of the workday.

We manipulated voice influence by telling participants that on

the basis of their past experience, they were either certain or

uncertain that their supervisor would incorporate their requests

into the final decision. Following the scenario, participants in

each condition scrolled through the accompanying question-

naire, which contained sets of randomly ordered dependent mea-

sure questions followed by manipulation check questions. Com-

pleted sections could not be retrieved, so responses could not

be changed. Dependent measures and manipulation check questions. As

in Experiment 1, dependent measures included perceived proce-

dural fairness (a = .98), perceived decision control (a = .96),

and outcome satisfaction (a = .93). Manipulation check ques-

tions assessed the level of voice, perception of task meaning-

fulness, and influence of previous voice. All manipulation ques- tions were answered correctly.

Results

Intercorrelations among dependent measures. Depen-

dent measures were significantly and positively associated

with one another, ranging from .61 for the fairness and

decision control measures to .49 for the decision control

and satisfaction measures. Response measures were not

collapsed because the interactive effects of the indepen-

dent variables differed across the response measures.

Shape of the value function. A separate trend analysis

(Beyer, 1988) for each of the three response measures

was conducted within each level of the two independent

variables (task meaningfulness and voice influence). For

each of the 12 analyses, findings indicated a nonlinear

voice value function, with significant linear, quadratic,

and cubic components (p < .0001).

Additional tests. Given confirmation of the nonlinear

components, an ANOV\ model and associated Duncan's

multiple-range tests were used to check for the expected

interaction effects, as well as to identify the approximate

shape of the value function for voice. For the measures

of procedural fairness, decision control, and outcome sat-

isfaction, significant Voice Magnitude X Task Meaning-

fulness interactions were found, F(4, 647) = 26, p <

.001; F(4, 647) = 9.4, p < .001; F(4, 647) = 30.3,

p < .01, respectively. Consequently, value functions for

fairness, decision control, and satisfaction are plotted sep-

arately within each level of task meaningfulness to facili-

tate visualization of the shape and nature of each value

function (Figure 3), and significant main effects for voice

magnitude and task magnitude are not discussed further.

Table 1 contains cell means, difference scores, signifi-

cance levels, and effect sizes (Cohen, 1988) for the three

response measures within each level of task meaning-

fulness. Discussion of the approximate shape of each of

these value functions is presented below.

Voice value function under high task meaningfulness.

The top half of Table 1 documents no significant step wise

gains in perceptions of fairness or satisfaction when parti-

cipants with no voice (mute condition) were compared

with participants expressing preferences for 5 computer

attributes. Only the perception of control significantly in-

creased from mute to 5 preferences. An increase in voice

from 5 to 10 decisions did not result in significant step-

wise gains on response measures of fairness or control;

only the perception of satisfaction increased. Increasing

the level of voice from 10 to 15 decisions resulted in

significant, as well as the largest, stepwise increases on

all three response measures. The increase in voice from

15 to 20 decisions resulted in significant, but smaller,

stepwise gains in perceptions of fairness and satisfaction,

but there was no significant increase in the perception of

control.

Voice value function under low task meaningfulness.

RESEARCH REPORTS 793

8 3 -

S

= 9 .2 8

Z 5

° 4

1 3

g - 2

g 1-

* 0

10

Magnitude of Voice

15 20

I —•— High Task Meaningfiilness - •• - Low Task Meaningfulness |

Figure 3. Shape of value functions for perceptions of procedural fairness, control, and satisfac- tion for Experiment 2.

A similar pattern was uncovered when task meaning-

fulness was low (see lower half of Tahle 1). The increase

in voice magnitude from mute to 5 decisions and from 5

to 10 decisions did not result in significant stepwise gains

in perceptions of fairness or satisfaction. However, step-

wise increases in control were significant from mute to 5

decisions and from 5 to 10 decisions. The largest stepwise

gain for each of the response measures occurred when

voice increased from 10 to 15. Changes in the level of

voice from 15 to 20 decisions did not result in significant

increases for any of the three dependent variables. In addi-

tion, the region where the greatest changes took place

(i.e., between 10 and 15 decisions) did not change

whether task meaningfulness of the decision was high or

low; instead, only the elevations (slopes) of the curves

were altered. Interactive effects of task meaningfulness. Table 2 con-

tains cell means, difference scores, significance levels, and

794 RESEARCH REPORTS

Table 1

Means, Mean Difference Scores (MD), and Effect Sizes (ES) by Task Meaningfulness

Difference in means for row-column comparisons

Mute 10 Perception and magnitude

of voice M MD ES MD ES MD ES ES

High task meaningfulness

Fairness Mute 5 Decisions 10 Decisions 15 Decisions 20 Decisions

Control Mute 5 Decisions 10 Decisions 15 Decisions 20 Decisions

Satisfaction Mute 5 Decisions 10 Decisions 15 Decisions 20 Decisions

1.6 2.1

2.6 6.7 7.9

2.7

3.6 4.2 8.2 8.5

2.9 2.7 4.0 7.0 7.9

0.5 1.0*

5.1* 6.3*

0.9 1.5*

5.5* 5.8*

-0.2 1.1* 4.1* 5.0*

0.6 3.2 3.9

0.5 0.9 3.2 3.4

—

0.7 2.5 3.0

0.5 4.6* 5.8*

0.6 4.6* 4.9*

1.3* 4.3* 5.2*

—

2.9 3.6

—

2.7 2.8

0.8 2.6 3.2

4.1*

5.3*

4.0*

4.3*

3.0*

3.9*

2.6 3.4

2.3 2.5

1.8 2.4

1.2* .8

0.3 —

0.9* 0.6

Low task meaningfulness

Fairness Mute 5 Decisions 10 Decisions 15 Decisions 20 Decisions

Control Mute 5 Decisions 10 Decisions 15 Decisions 20 Decisions

Satisfaction Mute 5 Decisions 10 Decisions 15 Decisions 20 Decisions

2.1

2.6 3.1 5.0 5.6

2.5 3.5 4.2 6.4 6.8

4.6 4.8 4.6 6.3 6.3

0.5 1.0*

2.9* 3.5*

1.0* 1.7*

3.9* 4.3*

0.2 0.0 1.7*

1.7*

0.6 1.8 2.1

.6 1.0 2.3* 2.5*

— —

1.0 1.0

0.5 2.4*

3.0*

0.7* 2.9* 3.3*

-0.2 1.5*

1.5*

—

1.5 1.8

.4 1.7 1.9

—

.9

.9

1.9* 2.5*

2.2*

2.6*

1.7*

1.7*

1.2 1.5 0.6 —

1.3 1.5 0.4 —

1.0 1.0 0.0 —

Note. Effect sizes (Cohen 1988, p. 49) were calculated as difference in means/root-mean-square error. A dash is reported in the effect size column when means are not significantly different. *p < .05.

effect sizes of each response variable by task meaningfulness

within levels of voice. At lower levels of voice (i.e., mute,

5, and 10 decisions), only the perceptions of satisfaction

were significantly lower in the high- compared with the low-

task-meaningfulness condition. At high levels of voice (15

and 20 decisions), perceptions of fairness, control, and satis-

faction were all significantly more positive in the high- com-

pared with the low-task-meaningfulness condition.

Effects of voice influence. Results indicated only a

significant main effect of voice influence on satisfaction,

F(l,647) = 12.2,p < .01, with higher levels of satisfac-

tion when decision makers were consistent (M = 5.3)

as compared with inconsistent (M = 4.9) in following

participant recommendations. Within each level of voice

magnitude (mute through 20), differences between cell

means in high- and low-voice-influence conditions were

consistent with the main effect for the response measure

of satisfaction, but individual cell contrasts were not sig-

nificantly different. There were no significant effects of

voice influence on perceptions of fairness or control.

Discussion

Voice value function. Results of Experiment 2 were

supportive of a nonlinear value function for voice with

RESEARCH REPORTS 795

Table 2

Means, Mean Difference Scores, and Effect Sizes by Task Meaningfulness

Within Levels of Voice

Task meaningfulness

High Low High Low High Low High Low High Low

High Low High Low High Low High Low High Low

Magnitude of voice

Mute Mute 5 Decisions 5 Decisions 10 Decisions 10 Decisions 15 Decisions 15 Decisions

20 Decisions 20 Decisions

Mute Mute 5 Decisions 5 Decisions 10 Decisions 10 Decisions 15 Decisions 15 Decisions 20 Decisions 20 Decisions

M

Perceptions of fairness

1.6 2.1 2.1 2.6 2.6 3.1 6.7 5.0 7.9 5.6

Perceptions of control

2.7 2.5 3.6 3.5 4.2 4.2 8.2 6.4 8.5 6.8

Difference score

-0.5

-0.5

-0.5

1.7*

2.3*

0.2

0.1

0.0

1.8*

1.7*

Effect size

—

—

—

1.1

1.3

—

—

—

1.1

1.0

Perceptions of satisfaction

High Low High Low High Low High Low

High Low

Mute Mute 5 Decisions 5 Decisions 10 Decisions 10 Decisions 15 Decisions 15 Decisions 20 Decisions 20 Decisions

2.9 4.6 2.7 4.8 4.0 4.6 7.0 6.3 7.9 6.3

-1.7*

-2.1*

-0.6

0.7*

1.6*

1.0

1.3

—

0.4

1.0

Note. Effect sizes (Cohen 1988, p. 49J were calculated as difference score/root-mean-square error. A dash is reported in the effect size column when means are not significantly different. *p < .05.

both convex and concave components. Perceptions of fair-

ness, control, and satisfaction were more positive as the

magnitude of voice increased. The largest stepwise in-

crease in the response measures occurred when the solici-

tation of voice rose from 10 of 20 to 15 of 20 attributes.

Diminishing marginal returns were found on all three re-

sponse measures when voice further increased from 15

to 20 attributes. Allowing participants to express prefer-

ences for 15 attributes seemed to separate material from

immaterial levels of participation. Perceptions moved from at or below the midpoint of the scale to above the midpoint of the scale on nearly all response measures.

The only exception was in the low-task-meaningfulness

condition, where perceptions of fairness increased to near

the midpoint of the scale. Thus, for these participants

voice must occur at sufficiently high levels (i.e., between

50% and 75%) to have a substantial impact.

Influence of task meaningfulness. The more extreme

reactions on the response measures when the task mean-

ingfulness was high, compared with low, documents how

the consequences of voice can be dependent on the deci-

sion context. These results suggest that it may require

more voice to make people feel satisfied with the outcome

when voice is solicited on a personally salient task. How-

ever, when high-task-meaningfulness participants received

more voice, they not only were more satisfied with the outcome but also perceived the process as fairer and felt

more in control than participants who voiced preferences

on a less meaningful task.

These results extend findings reported by Brockner and

796 RESEARCH REPORTS

Wiesenfeld (1996). In this regard, procedures have a

greater impact on participants, not only when they receive

unfavorable outcomes but also when the task is more

meaningful. This research is also consistent with the work

of Hunton and Price (1997), who reported stronger perfor-

mance effects when voice was solicited on a more mean-

ingful task.

Voice influence. The weak effects of voice influence

could be related to the manipulation itself. The manipula-

tion may not have created strong enough differential cues

of what to expect in the future to impact perceptions of

fairness and control or to strongly influence perceptions

of satisfaction. Although this may represent a failure of

the strength of the manipulation, it may also be that the

very act of voice solicitation creates a contrary set of

expectations. Unless there is strong evidence to the con-

trary, there may be a tendency to treat voice solicitation

attempts as genuine. That is, participants might feel that

a decision maker would not take the time and expend the

effort required to solicit voice and then discard or ignore

such input. If this is the case, the solicitation of voice

could have weakened the impact of the voice influence

manipulation, which could partially account for the non-

significant effects reported in this study. These results do

not support prior findings of Hunton (1995). However,

the manipulation of voice influence in the current study

was considerably weaker than the manipulation used by

Hunton (1995), where voice was promised but actually

denied.

General Discussion

Results of Experiment 1 and 2 are supportive of a non-

linear value function for voice, with diminishing marginal

returns as voice nears its maximum possible level. When

the participants were midlevel executives, the shape of

the function resembled more of an S, particularly at the

convex portions of the curve. Although smaller increments

in the manipulation of levels of voice could more clearly

define the nature of this curve, our results do document

the need for higher levels of voice for the midlevel execu-

tives, as compared with students, to perceive the process

as fair (i.e., above the midpoint of the scale). Perhaps,

for the executive sample, soliciting voice on half or less of

the decisions may be inconsistent with their expectations

based on their work experience and position. For the col-

lege students, uncertainty regarding typical operating pro-

cedures or lower perceived status may have influenced

their responses. Our results also document beneficial ef-

fects for both samples when voice-based participation is

less than fully comprehensive (i.e., < 100%).

The instrumental (Thibaut & Walker, 1975) and rela-

tional (Lind & Tyler, 1988) models of procedural justice,

as well as referent cognition theory (Folger, 1987), can

explain increases in perceptions of fairness across the

levels of voice examined in this study. However, these

theoretical perspectives do not account for or specify a

region such as the one found in this study where voice

had diminishing returns. Future research could further

define and refine this value function by focusing on the

reasons for different value functions among different pop-

ulations. Measurement of participant expectation levels

may help clarify such differences. Alternatively, partici-

pant expectations could be manipulated (see Ambrose,

Harland, & Kulik, 1991), and the shape of the value func-

tion for voice could be plotted to confirm the S shape

function that was approximated in this study.

The precise impact of increasingly higher levels of

voice has not been closely examined in the procedural

justice literature. We have attempted to further this line

of inquiry by applying the concept of perceived value to

voice and by providing a first approximation of the voice

value function. Despite acknowledged limitations of the

methodology (e.g., scenario approach and voice solicita-

tion on a single issue), the results are generally consistent

with theory and research using alternative methodologies,

which have documented the beneficial effects of voice in

a variety of different situations (Greenberg, 1990). In

addition, our research suggests the theoretical and practi-

cal usefulness of treating voice as a valued commodity

to answer a variety of questions, such as who should

participate in decisions, under what circumstances is

voice-based participation likely to be most effective, and

at what levels of voice solicitation do the marginal benefits

of additional voice become negligible?

References

Ambrose, M. L., Harland, L. K., & Kulik, C. T. (1991). Influ-

ence of social comparisons on perceptions of organizational

fairness. Journal of Applied Psychology, 76, 239-246.

Beyer, W. (1988). CRC handbook of tables for probability and

statistics (2nd ed.). Boca Raton, PL: CRC Press.

Brickner, M., Harkins, S. G., & Ostrom, T. M. (1986). The

effects of personal involvement: Through provoking implica-

tions for social loafing. Journal of Personality and Social

Psychology, 51, 763-769.

Brockner, J., & Wiesenfeld, B. M. (1996). An integrative frame-

work for explaining reactions to decisions: Interactive effects

of outcomes and procedures. Psychological Bulletin, 120,

189-208.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum.

Cooper, M. R., & Wood, M. T. (1974). Effects of member par-

ticipation and commitment in group decision making on in-

fluence, satisfaction and decision riskiness. Journal of Ap-

plied Psychology, 59, 127-134.

Doll, W. J., & Torkzadeh, G. (1988, February). The measure-

ment of end-user computing satisfaction. MIS Quarterly, 12,

258-274.

RESEARCH REPORTS 797

Barley, P. C., & Kanfer, R. (1985). The influence of component

participation and role models on goal acceptance, goal satis-

faction and performance. Organizational Behavior and Hu-

man Decision Processes, 36, 378—390.

Barley, P. C., & Lind, E. A. (1987). Procedural justice and par-

ticipation in task selection: The role of control in mediating

justice judgments. Journal of Personality and Social Psychol-

ogy, 52, 1148-1160.

Folger, R. (1986). Rethinking equity theory: A referent cogni-

tion model. In H. W. Bierhoff, R. L. Cohen, & J. Greenberg

(Eds.), Justice in social relations (pp. 145-162). New York:

Plenum.

Folger, R. (1987). Reformulating the preconditions of resent-

ment: A referent cognitions model. In J. C. Masters & W. P.

Smith (Eds.), Social comparison, justice, and relative depri-

vation: Theoretical, empirical, and policy perspectives (pp.

183-215). Hillsdale, NJ: Erlbaum.

Greenberg, J. (1990). Organizational justice: Yesterday, today

and tomorrow. Journal of Management, Ifi, 399-432.

Hogarth, R. (1989). Judgement and choice. New \brk: Wiley.

Hunton, J. E. (1995). Procedural justice and user involvement

in developing information systems. Journal of Information

Systems, JO, 20-35.

Hunton, J. E., & Price, K. H. (1994). A framework for investi-

gating involvement strategies in accounting information sys-

tems development. Behavioral Research in Accounting, 6,

128-158.

Hunton, J. E., & Price, K. H. (1997). Effects of the user partici-

pation process and task meaningfulness on key information

system outcomes. Management Science, 43, 797-812.

Kahneman, D., & Tversky, A. (1979). Prospect theory: An anal-

ysis of decisions under risk. Econometrica, 47, 263-291.

Lind, E. A. (1995). Justice and authority relations in organiza-

tions. In R. S. Cropanzano & K. M. Kacmar (Eds.), Organi-

zational politics, justice and support: Managing the social

climate of the workplace (pp. 83-97). Westport, CT: Quorum

Books.

Lind, E., Kanfer, R., & Barley, P. C. (1990). Voice, control, and

procedural justice: Instrumental and noninstrumental con-

cerns in fairness judgments. Journal of Personality and Social

Psychology, 59, 952-959.

Lind, E. A., & Tyler, T. R. (1988). The social psychology of

procedural justice. New York: Plenum.

Locke, E. A., & Schweiger, D. M. (1979). Participation in deci-

sion-making: One more look. In B. Straw (Ed.), Research in

organizational behavior (Vol. 1, pp. 265-339). Greenwich,

CT: JAI Press.

Miller, J., & Grush, J. E. (1988). Improving predictions in ex-

pectancy theory research: Effects of personality, expectancies,

and norms. Academy of Management Journal, 31, 107-122.

Shapiro, D. L. (1993). Reconciling theoretical differences

among procedural justice researchers by re-evaluating what

it means to have one's views "considered": Implications for

third party managers. In R. Cropanzano (Ed.), Justice in the

workplace (pp. 51-78). Hillsdale, NJ: Erlbaum.

Thibaut, J., & Walker, L. (1975). Procedural justice: A psycho-

logical analysis. Hillsdale, NJ: Erlbaum.

Tyler, T. R., & Lind, E. A. (1992). A relational model of author-

ity in groups. Advances in Experimental Social Psychology,

25, 115-191.

Van den Bos, K., Vermut, R., & Wilke, H. A. (1996). The

consistency rule and the voice effect: The influence of expec-

tations on procedural fairness judgements and performance.

European Journal of Social Psychology, 26, 411-428.

Vroom, V. H., & Jago, A. R. (1988). The new leadership. Engle-

wood Cliffs, NJ: Prentice Hall.

Received August 11, 1997

Revision received March 27, 1998

Accepted March 30, 1998 •