Voice in the Void: From Voice to Acquiescent Silence over Time as Learned Helplessness in Organizations

Introduction

Within organizations, individuals often seek control through the expression of “voice”—the act of expressing suggestions, concerns, and opinions to decision-making authorities in an attempt to indirectly influence a situation (Greenberg & Edwards, 2009; Morrison, 2014, 2023). Conversely, members may remain silent, refraining from voicing—a response of potential concern, particularly when they are faced with adverse workplace conditions. Numerous studies have explored the antecedents and consequences of voice and silence behaviors (for reviews, see Chamberlin et al., 2017; Hao et al., 2022; Knoll et al., 2016). However, as noted by Morrison (2023), this rich body of research is deficient in theory integration to understand what drives members of a structure to speak up or remain silent. This limitation becomes apparent when considering the recurrent parallels drawn between acquiescent silence (i.e., silence due to resignation motives; Pinder & Harlos, 2001) and the concept of “learned helplessness” (Maier & Seligman, 1976, 2016) that, to our knowledge, have not been subjected to convincing empirical validation (e.g., Knoll et al., 2021; Milliken & Morrison, 2003; O’Donovan et al., 2021). Learned helplessness refers to a state in which individuals disengage and resign themselves due to a perceived lack of control, stemming from learned experiences over multiple attempts where their actions failed to impact situations (Kofta & Sędek, 1989; Maier & Seligman, 1976, 2016; Peterson et al., 1993). Furthermore, the learned helplessness state is not merely passivity due to resignation in negative circumstances; it represents a deeper, depressive-like state that is genuinely concerning for individuals on broader aspects of their lives (Abramson et al., 1978; Maier & Seligman, 1976, 2016; Peterson et al., 1993; Seligman, 1975). Exploring the connection between acquiescent silence and learned helplessness could deepen our understanding of resignation-caused silence, uncovering organizational triggers and contribute to explain its consequences such as burnout or decreased job performance, possibly linked to this depressive-like state (Hao et al., 2022).

While prior theoretical linkages between learned helplessness and acquiescent silence have been posited, our research aim to explicitly test the hypothesis that acquiescent silence reflects a state of learned helplessness. We contend that this proposed framework enhances current understandings of voice and silence dynamics in organizational settings, particularly in navigating the interactions between approach and avoidance behaviors in organizations (Morrison et al., 2015; Sherf et al., 2021). Furthermore, we propose an experimental protocol to empirically test our hypotheses. As emphasized by Sherf et al. (2021), the current dearth of experimental research underscores the necessity of addressing concerns regarding the internal validity and robustness of theoretical frameworks concerning voice and silence behaviors.

Voice, Silence, and Control in Organizations

Voice is a discretionary, change-oriented behavior aimed at expressing ideas, opinions, suggestions, or concerns to influence organizational situations indirectly (Morrison, 2023; Van Dyne & LePine, 1998). Conversely, silence is not merely the absence of voice; it refers to the withholding of information that could be useful for constructive change (Knoll et al., 2016; Morrison, 2023). Voice and its content are a valuable source of information for organizations and decision-makers to drive desirable changes, such as innovation (Liang et al., 2019), improving members’ well-being (Colquitt et al., 2001), or preventing hazards (Noort et al., 2019). Voice can be manifested through various systems (e.g., direct communication with a manager, written grievance procedures; Harlos, 2001) and can take different forms based on individual objectives and situational parameters (Klaas et al., 2012; Liang et al., 2012; Maynes & Podsakoff, 2014). For instance, voice can be promotive, aiming to improve organizational functioning, or prohibitive, addressing present or past concerns.

Voice possesses two important features. First, it is a self-motivated action. To be considered as “voice,” there must be an expression of some information directed to someone else (Maynes & Podsakoff, 2014; Sherf et al., 2019). Second, this action is essentially focused on indirectly making a difference or changes in one’s environment. Therefore, voice behaviors can be qualified as attempts to exert some form of control (Brockner & Wiesenfeld, 2019; Thibaut & Walker, 1975). Following the expression of voice, its efficacy relies on the endorsement of the voiced content by authorities or decision-makers. The evaluation of the concrete impact of voice content concerns the degree of perceived instrumentality of voice in the situation (i.e., was voice instrumental in making a change in the situation?; Avery & Quiñones, 2002; King et al., 2019). In instances where voice does not yield expected results (i.e., low voice instrumentality), individuals may perceive a diminished sense of control, potentially leading to learning that exerting effort to voice is futile, especially if this kind of scenario is repeated. In other words, personal history of perceived action-outcome contingencies (e.g., voice instrumentation) in similar past situations is an important characteristic of perceived control for organizational members (Skinner, 1996). This perception of control, in turn, influences future motivation to act in similar situations (Leotti et al., 2010; Rauvola & Rudolph, 2022, 2023). Consequently, perceived control through voice in organizations should be influenced by the multiple experiences of its perceived degree of instrumentality, and therefore be a key antecedent for future intentions to voice or to stay silent.

Silence is multidimensional construct that encompasses various motives ( Knoll & Van Dick, 2013; Knoll et al., 2016; Pinder & Harlos, 2001; Van Dyne et al., 2003). On the one hand, organizational members can stay silent to protect themselves (defensive silence) or their peers (pro-social silence). These forms of silence are caused by fear of possible future repercussions for their voice (Knoll et al., 2016; Van Dyne et al., 2003). These forms of silence are proactive and members may then find other ways to alter the situation (e.g., counter-productive work behaviors; Jahanzeb & Fatima, 2018; Qi & Ramayah, 2022). On the other hand, workers can stay silent because they feel that voice is futile. This form of silence is identified as “acquiescent silence.” Acquiescent silence is a deep state of acceptance that one’s voice cannot impact the situation and that a more favorable situation is not attainable (Knoll et al., 2016; Knoll & Van Dick, 2013; Pinder & Harlos, 2001). An example is when an employee remains silent in a negative situation, thinking that it is useless to try to say something about it or that nothing would have changed anyway (Knoll, 2021; Knoll et al., 2016; Morrison, 2014; Pinder & Harlos, 2001; Van Dyne et al., 2003). For more than 20 years, scholars have often suggested that acquiescent silence could be an instance of learned helplessness, but this deduction has never been verified explicitly. Table 1 provides a non-exhaustive list of research articles affirming the connection between acquiescent silence and a state of learned helplessness in organizations.

Control-based Explanation for Acquiescent Silence

In a control-based perspective, acquiescent silence corresponds to a state where one could have acted to influence a situation with voice but remains silent and resigned. In the absence of explicit evidence, learned helplessness theory effectively serves as an adequate framework, positing that individuals can “learn” the inefficacy of their voice over time and lose personal control in the organizational environment. Learned helplessness stems from multiple instances of action-outcome non-contingency, progressively eroding personal control as individuals experience successive failures in attempts at controlling the situation (Kofta & Sędek, 1989; Maier & Seligman, 1976, 2016). In a helpless state, individuals are unable to see the potential impact of their actions and remain passive, even if their actions could have produced change. In other words, individuals must have tried to act in past similar situations or environments without results.

In the experimental studies presented here, we test the effects of repeated occurrences of low voice instrumentality (i.e., low voice impact) on a key manifestation of learned helplessness syndrome: diminished motivation to engage in actions referring to, in our case, the preference towards remaining silent. Our first hypothesis (H1) posits that individuals are less likely to utilize voice following exposure to repeated low voice-instrumentality situations, in contrast to situations characterized by repeated high voice-instrumentality. Our second hypothesis (H2) suggests that individuals exposed to multiple low voice-instrumentality situations are more prone to experiencing feelings of helplessness compared to those exposed to repeated high voice-instrumentality situations.

From Voice Behavior to Acquiescent Silence over Time

The dynamic, control-based proposition for the development of acquiescent silence in a way similar to the development of learned helplessness is in line with recent theorization about voice and silence. Sherf et al. (2021) relied on the approach/avoidance literature (Carver & White, 1994) to distinguish voice and silence as being provoked by two different biologically based self-regulation systems: the behavioral activation system (BAS) and the behavioral inhibition system (BIS). This theory posits that much human behavior is driven by these two systems, activated by environmental incentives. The BAS relates to the motivation to act towards rewards for the self, whereas the BIS motivates actions away from punishments or harm (Carver, 2006; Carver & White, 1994). Sherf et al. (2021) showed that voice was related to an activation of the BAS in case of high voice impact (i.e., high voice instrumentality) as it corresponds to a reward/satisfactory situation, resulting in more subsequent voice behaviors. Silence was broadly identified as corresponding to an activation of the BIS in the case of low psychological safety: repercussions for speaking up lead the individual to stay silent. One could argue that the activation of the BIS in this case is corresponding to silence motivated by fear (i.e., defensive or prosocial silence), but this explanation is not relevant for acquiescent silence, because its principal motive is resignation. Less attention in the literature is paid to cases where BAS/BIS systems are deactivated. Carver (2004) pointed out this issue and indicated that the systems may create both positive (when activated) and negative (when deactivated) outcomes for individuals. We propose that acquiescent silence can emerge and take place when the BAS is deactivated. As the BAS is triggered by voice impact incentives (i.e., instrumentality), the repeated absence of voice instrumentality could lead to decreased activation (and, ultimately, deactivation) of the BAS over time. As we pointed out, this mechanism leads to learned helplessness, in correspondence with Sherf et al.’s (2021) proposition that “when the environment is not conducive to change from employees (i.e., no opportunities to achieve gains or rewards, creating a sense of futility), the BAS is less likely to be triggered, leading to low levels of or less frequent voice” (Sherf et al., 2021, p. 117). Therefore, repeated low levels of voice instrumentality can lead people to develop acquiescent silence, through the deactivation of the BAS. The decrease in voice usage should be observable over time, as individuals exposed to lack of voice instrumentality will begin to engage lesser resources in their voice attempts, leading to a decrease in voice performance (i.e., quantity or quality of suggested ideas). We will test this hypothesis (H3), formulated as: Voice performance decreases over time for participants exposed to repeated low instrumentality of their voice compared to participants exposed to repeated high instrumentality of their voice.

Overview of the Research

In two experimental studies, we investigated the impact of varying degrees of voice instrumentality repetition on the generation of voiced propositions among individuals from different types of organizations. Traditionally, a triadic design has been utilized as the primary approach to assess learned helplessness in controlled laboratory settings, comprising three distinct experimental groups (Alloy et al., 1984; Maier & Seligman, 2016; Peterson et al., 1993; Seligman, 1975). For both studies, we introduced an adapted version of this traditional protocol to investigate learned helplessness, specifically tailored to situations involving the use of voice. This research program received ethical approval from our university’s ethics review board, and the data that support the findings of these studies are available in an Open Science Framework (OSF) repository1.

Study 1

Method

Participants and Design

University students were invited to take part in the study. An internet link directed participants to the experiment generated using Qualtrics (Qualtrics Labs, Inc., 2020). The study was presented as one led by a research group interested in compiling students’ propositions (i.e., recommendations they were willing to voice) regarding topics related to their quality of life at the university. The study was initiated by 1,233 students (clicked on the link), though only 454 (36.8%) participations were fully completed (i.e., went through the entire study and participated in the remediation session after the experiment). Of these 454 participations, we excluded participants who made the exact same propositions for all four topics (n = 4), who did not respect instructions (n = 12), and who did not make any propositions for the first topic in the study2 (n = 99). The final sample consisted of 339 participants (68.7% female, 25.6% male, 5.6% other/rather not say) from 18 to 63 years old (M = 22.2, SD = 5.48); 207 were undergraduates, 113 master students, 16 doctoral students, and 3 did not respond regarding their student status.

Procedure and Material

The study was portrayed as a research led by organizational psychologists and human resources professionals aiming to obtain propositions regarding various aspects of the quality of student life for future research purposes. Instructions indicated that participants would make suggestions (i.e., ideas, recommendations) on four topics (hereafter T1 to T4; Appendix A; topic order was randomized across participants). For each topic, instructions indicated that an experimenter would include only the most interesting propositions in the final pool to be compiled in the study. This cover story allowed us to manipulate voice instrumentality in a context relevant to students: Participants thought they had to make relevant suggestions for them to be considered (i.e., making their voice instrumental for inclusion in the pool). Every participant agreed to take part in the study after reading the instructions. Participants were then redirected to a “waiting room” page lasting seven seconds to simulate the time for the experimenter to join their session.

Voice Instrumentality Manipulation. Participants were randomly assigned to one of three experimental conditions: low voice instrumentality, high voice instrumentality, or control condition. They were then instructed to provide propositions on four consecutive topics. A pretest was conducted on 32 students from the same university and found no significant differences on ratings of the importance, relevance, and willingness of students to express opinions on each of the four topics3 (Appendix B for details). For each topic, participants could make up to 10 propositions in spaces provided; they were asked to express distinct ideas in each space. After providing propositions on a topic, a loading screen was displayed simulating the decision time of the experimenter. Then, a message presenting the experimenter’s feedback, dependent on the experimental condition, was displayed. For each topic, the feedback message for participants was: “After analysis of your proposals, we inform you that they [have/have not] been retained by the experimenters. Your proposals [will/will not] be added to the research list.” The message for participants in the control group (i.e., without specific feedback on their voice contingency) was: “For the moment, we need you to proceed on to the next topic.” Participants who made no propositions received the message: “You have chosen not to express yourself on this topic.” In the context of this protocol, expressing opinions aims to ensure the incorporation of proposals into the final study pool (similarly to Avery & Quiñones, 2002). Feedback regarding the inclusion of proposals in the final study pool serves as a noteworthy indicator that the act of voicing opinions played a crucial role in this scenario, facilitating the achievement of one’s goals in this context (having proposals retained by the experimenter). We computed the total number of propositions for each participant for each topic.

Intention to Voice. To test for the learning effect of instrumentality over time on voice behavior, we measured participants’ intention to use voice anew on one more unspecified optional topic after the four-topics phase. Participants could read: “You have the possibility to express yourself on an additional theme. Do you wish to express yourself on this last theme?” Responses were scored with a binomial rating (yes = 1, no = 0).

Measures. After the experimental manipulation, participants completed scales regarding how they felt in the situation. All measures used Likert 1 (strongly disagree) to 5 (strongly agree) point scales (see Appendix C). As a manipulation check, we measured participants’ perceptions of voice instrumentality with two items (α = .93). As there is no validated scale for feelings of helplessness in a particular situation, we used five items (α = .73) to measure participants’ feelings of helplessness. These items measured how much the participants believed their efforts were pointless in the experiment, as understanding the feeling of having no impact is crucial for studying learned helplessness (Peterson et al., 1993). One item was for example “No matter what I wrote, it appeared to have no impact on the selection of proposals.”

Results

The analyses were conducted using R Statistical Software (v4.2.2; R Core Team 2021).

Manipulation Checks

As data for perceived voice instrumentality did not meet the assumptions for one-way analysis of variance4, we analyzed data using the Kruskal-Wallis non-parametric test (MacFarland & Yates, 2016; Van Hecke, 2012). The analysis showed a significant group difference in perceived voice instrumentality, χ²(2, N = 339) = 161, p < .001, ϵ^2 = .48. Pairwise comparisons with the Dwass-Steele-Critchlow-Fligner method (DSCF; Critchlow & Fligner, 1991) indicated that participants in the low voice instrumentality condition perceived less voice instrumentality (M = 1.51, SD = 0.78) than participants in both the high voice instrumentality (M = 3.73, SD = 0.84, W = -16.11, p < .001) and control conditions (M = 3.40, SD = 0.80, W = -15.08, p < .001). In addition, participants in the high voice instrumentality condition perceived more voice instrumentality than participants in the control condition, W = -5.03, p < .01.

Silence and Helplessness Feelings

Hypothesis 1 proposed that participants in the low voice instrumentality condition would be less likely to seize a new opportunity to voice their suggestions (thus would be more likely to remain silent) in a new situation where voice was possible. To test this hypothesis, we performed a chi-square analysis to examine the relation between experimental conditions and participants’ choice to voice on a new topic. The test was significant, χ²(2, N = 339) = 17.9, p < .001, Cramer’s V = .23, indicating that voice behavior intentions differed across groups. Pearson’s residuals were computed for each cell of the contingency table to examine the nature of the relations. Positive standardized residual values specify an attraction (positive association) between the corresponding row and column variables while negative values indicate a repulsion (negative association). As depicted in Figure 1, there is a strong positive relation between preference for silence and repeated low voice instrumentality (ϵ² = 3.89), whereas this relation is negative for both the repeated high voice instrumentality and control groups (respectively ϵ² = -0.99, and ϵ² = -2.29). These results indicate that repeated low voice instrumentality is strongly related to silence whereas participants in the repeated high instrumentality and control groups were more inclined to voice, supporting Hypothesis 1.

Kruskal-Wallis non-parametric test was used5 to test Hypothesis 2 regarding helplessness feelings. The test was significant, χ²(2, N = 339) = 115, p < .001, ϵ^2 = 0.35. In line with Hypothesis 2, pairwise comparisons indicated that participants in the low voice instrumentality condition felt more helpless (M = 3.88, SD = 0.71) than participants in both the high voice instrumentality (M = 2.73, SD = 0.58, W = 13.8, p < .001) and control conditions (M = 2.79, SD = 0.56, W = 13, p < .001), the latter two conditions not differing in terms of felt helplessness (W = 1.69, p = .45).

Voice Performance Over Time

Hypothesis 3 stated that the number of participants’ propositions would decrease over trials in the low voice instrumentality condition compared to the high voice instrumentality condition, as repeated situations of low voice instrumentality should gradually deactivate participants’ BAS, thus decreasing the number of voiced propositions. As stated in the method section, we counted the number of voiced propositions for each participant and each topic. Dispersion statistics of these data revealed over-dispersed distributions in most cases (T2, T3, and T4, for every experimental condition, see Appendix D), indicative of negative binomial distributions of our count data (Coxe et al., 2009). Because negative binomial distributions produce biased analyses when least squared methods are used (Coxe et al., 2009; Gardner et al., 1995), we applied a generalized linear mixed effect model (GLMM) using the GAMlj package for R (Gallucci, 2019). This approach allows for the comparison of repeated dependent measures among multiple groups with non-parametric distributions, such as negative binomial distributions (Bono et al., 2021). Furthermore, GLMM are robust even in cases of unequal group sizes which is the case here (de Melo et al., 2022; Pinheiro & Bates, 2000). The model consisted of two fixed effects (“instrumentality”: low voice instrumentality, high voice instrumentality, and control group; and “time”: T1, T2, T3, and T4), their interaction, and participants as a random factor.

The analyses revealed significant main effects of instrumentality, χ²(2, N = 339) = 17.8, p < .001, and time, χ²(3, N = 339) = 32.2, p < .001, as well as a significant interaction between these factors, χ²(6, N = 339) = 12.9, p < .05. As shown in Figure 2, participants in the high voice instrumentality condition made significantly more propositions during the entire study than participants in the low voice instrumentality (M_{high inst.Â} – M_{low inst.Â} = 0.76, p < .001) and control conditions (M_{high inst.Â} – M_control = 0.44, p < .05). The difference between the low voice instrumentality and control conditions was in the expected direction but failed to reach significance (M_control – M_{low inst.} = 0.23, p = .08).

We performed post hoc analyses using Holm’s corrections to further examine the significant interaction. As expected, no group difference was found at T1 (all ps > .99). However, participants in the low voice instrumentality group made significantly fewer propositions than those in the high voice instrumentality one at T2 (M_{T2_low inst.} - M_{T2_high inst.} = -0.88, p < .05) and T4 (M_{T4_low inst.} - M_{T4_high inst.} = -0.92, p < .05). The difference between these groups at T3 was in the expected direction but failed to reach significance (M_{T3_low inst.} - M_{T3_high inst.} = -0.88, p = .08). No significant differences with the control group occurred at any time (all ps > .16). In addition, post hoc comparisons showed that the participants in the low voice instrumentality condition made fewer propositions at T2 compared to T1 (M_{T2_low inst.} - M_{T1_low inst} = -1.04, p < .001); however, the comparisons between T1 and T3 and between T1 and T4 were not significant (respectively, M_{T3_low inst.} - M_{T1_low inst} = -0.60, p = .56; M_{T4_low inst.} - M_{T1_low inst} = -0.75, p = .08). Overall, these results showed partial support for Hypothesis 3 as participants in the repeated low voice instrumentality tend to show lower voice behavior over the course of the trials.

Additional Analyses

As mentioned, the number of participants in the experimental conditions was not balanced (n = 86 for repeated low voice instrumentality, n = 133 for repeated high voice instrumentality, and n = 120 for the control group). This difference was particularly marked in the repeated low voice instrumentality condition with a large number of its participants not completing the study even though the number of participants beginning each experimental condition was roughly equivalent (n = 420 for repeated low voice instrumentality, n = 404 for repeated high voice instrumentality, and n = 409 for the control group). To further examine the relation between experimental condition and study completion, we conducted a chi-square analysis. The result was significant, χ²(2, N = 1,233) = 17, p < .001, V = .11. We again computed Pearson’s residuals for each cell of the contingency table. As shown in Figure 3, there was a strong negative relationship between study completion and the repeated low voice instrumentality condition (ϵ² = -2.74), whereas this relation was positive in the repeated high voice instrumentality and control groups (respectively ϵ² = 2.08 and ϵ² = 0.71). Therefore, the willingness to fully complete the study was influenced by voice instrumentality in the study.

Discussion

Study 1 tested the role of voice instrumentality over time on the intention to use voice in a new situation (H1), helplessness feelings (H2), and the number of voiced propositions over time (voice performance; H3). Results showed a significant impact of voice instrumentality over time: participants for whom voice was never instrumental were more inclined to stay silent when a new opportunity for voice was offered (H1) and felt more helpless compared to participants with high voice instrumentality or no specific feedback (H2). Furthermore, repeated lack of voice instrumentality influenced voice performance over time, as there was an overall decrease in the number of voiced propositions in the repeated low voice instrumentality condition compared to participants with high instrumentality (H3). These results bring two insights. First, the role of time and repetition is important in the emergence of silence due to resignation motives (acquiescent silence). Learning that an action (voice) is repeatedly inefficient in the environment (low voice instrumentality), as proposed by learned helplessness theory, seems to be the basis for the emergence of resignation motives. Second, these findings provide an interesting view on voice behaviors over time aligning with past research on voice and silence. As expected, participants did not differ on the number of voiced propositions before they were exposed to the first feedback (between T1 and T2), which produced a strong decrease in propositions for participants in the low voice instrumentality condition. That this effect was less present after the second feedback (at T3) may indicate a reactance effect with participants trying to regain control on the situation and using voice again (Tangirala & Ramanujam, 2008; Wortman & Brehm, 1975). But their voice is once again ineffective in this condition, and the number of propositions decreased again in the last trial (T4). Furthermore, we observed a notably higher level of withdrawal from the study (i.e., participant stopping the task before the end) in the low voice condition compared to the other two conditions. This withdrawal tendency may stem from factors such as frustration, reactance, or a perceived futility in continuing the assigned task. Despite the varied underlying reasons, this observation distinctly indicates that silence emerges as a prevalent response in situations characterized by low voice instrumentality, suggesting a proclivity towards disengagement from the given task. As university students’ main goal is achievement in their studies, they might be less motivated to engage in efforts to improve their organizational environment compared to employees working in organizations. To control for this potential bias, our second study replicated the experiment in a population of employees in organizations.

Study 2

The aim of Study 2 was to replicate the findings of Study 1 on a population of working organizational members. Thus, we restricted our population to a specific work sector to standardize our material and limit the influence of additional variance relating to potentially highly varied job characteristics. We conducted the study on retail workers in the United States, a population known to be subjected to less favorable working conditions (e.g., greater job insecurity, lower wages, part time positions, lower benefits) and potentially motivated to voice opinions regarding their working environment (Coulter, 2013; James et al., 2011). The hypotheses are identical to the ones in Study 1. Thus, we expected to observe lower tendencies to voice when an opportunity is available (H1) and higher feelings of helplessness (H2) after being exposed to multiple situations of low voice instrumentality. We also expected a decrease in the number of propositions over time (H3) with low voice instrumentality.

Method

Participants and Design

All participants were recruited through Prolific Academic and compensated £1.60 for their participation. The filters used on the platform to access the desired population were: English speakers (1) working in the retail sector (2) in the United States (3). As in Study 1, we portrayed the study as one piloted by a research group interested in compiling the propositions of retail workers on topics relating to their quality of life at work for future research purposes. The study was initiated 396 times, of which 319 participations (80.5%) were fully completed. We excluded participants who did not indicate being currently employed (n = 3) and who did not respect study instructions (n = 1). The final sample consisted of 315 retail workers from the United States (75.6% female, 20.3% male, 4.1% other/rather not say), aged 18 to 66 years old (M = 26.9, SD = 9.76), 145 indicating that they worked full-time (46%) and 170 part-time (56%). Most participants indicated working under supervisor (n = 297, 94.2%).

Procedure and Material

The procedure was identical to the one in Study 1.

Voice Instrumentality Manipulation. The four topics presented in this study related to general issues of quality of life at work and were presented in random order for each participant (Appendix A). A pretest, conducted on 36 US retail workers found no significant differences among the four topics on importance, relevance, and willingness of workers to express opinions (see Appendix D). The experimenter’s feedback for each experimental condition was the same as in Study 1, and the same method for counting participants’ propositions for each topic was applied.

Intention to Voice. We used the same item to measure the intention to use voice as in Study 1. Participants could read: “You have the possibility to express yourself on an additional theme. Do you wish to express yourself on this last theme?” (yes = 1, no = 0).

Measures. The same scales were used as in Study 1 for perceived voice instrumentality (as manipulation check) and helplessness feelings. Internal consistencies were good for both scales (respectively α = .96 and α = .84).

Results

Manipulation Checks

We conducted a non-parametric Kruskal-Wallis test as data for perceived voice instrumentality did not meet the assumptions for ordinary analysis of variance6. The analysis revealed a significant between-group difference in perceived voice instrumentality, χ²(2, N = 315) = 118, p < .001, ϵ²Â = 0.37, and DSCF pairwise comparisons indicated that participants in the low voice instrumentality condition (M = 2.31, SD = 1.19) perceived significantly less voice instrumentality than participants in the high voice instrumentality (M = 4.17, SD = 0.74; W = -13.84, p < .001) and control conditions (M = 3.90, SD = 0.76; W = 12.32, p < .001). Additionally, participants in the high voice instrumentality condition perceived more voice instrumentality than participants in the control condition (W = -3.86, p < .05).

Silence and Helplessness Feelings

To test Hypothesis 1, we conducted a chi-square analysis to examine the relation between the experimental group and willingness to voice on a new topic. The result was statistically significant, χ²(2, N = 315) = 8.82, p < .05, V = .16. Pearson’s residuals were computed for each cell of the contingency table. As shown in Figure 4, there was a positive relation between preference for silence and low voice instrumentality (ϵ²Â = 1.75), whereas the relation with silence was negative for high voice instrumentality and for the control group (respectively ϵ² = -0.23 and ϵ² = -1.42). These results indicate that repeated low voice instrumentality is related to silence, whereas participants in the high voice instrumentality and control groups where more inclined to voice, showing support for Hypothesis 1.

To test Hypothesis 2, we conducted a non-parametric Kruskal-Wallis7. The analysis revealed a significant difference between groups on helplessness feelings, χ²(2, N = 315) = 128, p < .001, ϵ²Â = 0.40, and DSCF pairwise comparisons revealed that participants in the repeated low voice instrumentality condition felt significantly more helpless (M = 3.57, SD = 0.72) than participants in the repeated high voice instrumentality (M = 2.29, SD = 0.65; W = 14.35, p < .001) and control conditions (M = 2.53, SD = 0.60; W = -12.99, p < .001), showing support for Hypothesis 2.

Voice Performance over Time

Data for the number of propositions on the different topics presented the same characteristics as in Study 1, thus the same analytic approach was used. The model revealed non-significant results for the main effects of instrumentality, χ²(2, N = 315) = 0.60, p = .73 and time, χ²(3, N = 315) = 2.28, p = .51, and for their interaction, χ²(6, N = 315) = 2.38, p = .88, indicating no differences in the number of voiced proposition among the different experimental groups and over the four times of the study, not supporting Hypothesis 3. The mean number of propositions over time across the three experimental conditions are presented in Figure 5.

Discussion

Study 2 was designed to replicate Study 1 to examine the generalizability of the Study 1 findings to a population of working professionals where intentions to voice and voice content might differ from those observed with university students. The procedure in Study 2 was identical to that of Study 1 but produced mixed results. In line with Study 1, participants under repeated low voice instrumentality were less prone to voicing on new topics (H1) and reported greater feelings of helplessness (H2). However, unlike Study 1, we found no significant intergroup differences in voiced propositions over time (H3). In opposition to H3, the mean number of propositions varied little over time and across experimental groups (Appendix D). A crucial aspect of our procedure that could elucidate this outcome is the recruitment method: participants were sourced through a crowdsourcing platform and compensated for their involvement. This method is known for yielding quality data in psychological experiments (Peer et al., 2022). On these platforms, participants are incentivized to perform well to ensure payment post-experiment, knowing that subpar performance can lead to rejection and loss of compensation. Moreover, frequent rejections lower participants’ “acceptance scores,” diminishing their chances of future study participation (Palan & Schitter, 2018). Given these factors and our procedure, participants might have perceived providing propositions as essential for validating their participation. Consequently, irrespective of feedback, the incentive of payment likely motivated continued participation in voicing across the four trials. Still, despite this possible concern of losing payment, participants in the low voice instrumentality condition showed a greater tendency toward silence when a new optional opportunity for voice was offered and still felt more helpless after the multiple experiences of low voice instrumentality.

General Discussion

We proposed a control-based dynamic approach to the emergence of resignation motives to stay silent in organizations and explicitly tested, for the first time, the association between acquiescent silence and learned helplessness. This work sought to provide a framework for understanding how acquiescent silence emerges, through progressively learning that voice is inefficient. We drew on learned helplessness theory stating that the low impact of one’s actions in a situation would cause a decrease in perceptions of control, and that the repetition of this scenario over time could lead to an absence of personal control and learning that future potential actions in the situation are futile (Kofta & Sędek, 1989; Maier & Seligman, 1976, 2016; Peterson et al., 1993). We suggested that repetitive situations of low voice impact (low instrumentality) are similar to learning that voice (i.e., an action aimed to control the environment) is inefficient, resulting in helplessness, reluctance to use voice again, and the emergence of acquiescent silence. Our work expands on recent theorizing on voice and silence relating to approach/avoidance systems (Sherf et al., 2021). Voice is related to an approach behavior, triggered by the perception that it can have an impact in one’s environment. Thus, repetitive low voice instrumentality could cause a progressive deactivation of this approach system, reducing the motivation to voice with the same intensity in similar situations (Sherf et al., 2021).

The results of the studies presented support the learned helplessness framework to understand the emergence of acquiescent silence. Across both studies, participants exposed to repeated situations of low voice instrumentality were significantly less likely to use voice anew when an opportunity was available. They also felt more helpless, indicative of resignation motives to stay silent in the situation relating to acquiescent silence. With regards to the quantity of voiced propositions (voice performance), the results obtained were mixed, but nonetheless instructive. Overall, the number of voiced propositions decreased over time for participants exposed to low voice instrumentality in Study 1, but did not for participants in Study 2, probably because participants were compensated to complete the latter study. All in all, the results obtained in those studies could facilitate an explicit connection between acquiescent silence and a state of learned helplessness regarding motivational deficits.

Implications for Theory

Our research provides an initial response to Knoll’s (2021) query regarding the incorporation of an individual’s historical interaction with a specific organization in the study of voice and silence. The incorporation of learned helplessness theory provides a framework on how acquiescent silence emerges as resignation caused by the individual’s progressive learning that voicing is inefficient. To our knowledge, the present studies are the first to examine experimentally the mechanisms leading to acquiescent silence over time. These studies therefore highlight the importance of considering individuals’ past voice attempts and their perceived instrumentality when treating acquiescent silence, an often-overlooked characteristic of voice/silence studies which typically focus on only one situation (King et al., 2019; Knoll, 2021), not taking into account repeated events. Our results suggest that the emergence of acquiescent silence lies in the repetition of failures to affect a situation with voice, as predicted by learned helplessness theory (Kofta & Sedek, 1989; Maier & Seligman, 2016; Peterson et al., 1993). These conclusions are also in line with indications that merely offering opportunities for voice is not enough to provide positive consequences for organization members (Avery & Quiñones, 2002; Bashshur & Oc, 2015; Detert et al., 2013; King et al., 2019), especially when voice falls on deaf ears (Harlos, 2001). Our findings that the learning of voice inefficacy also occurs with the presence of extrinsic incentives for voice such as monetary compensation is an additional observation of interest for theoretical considerations on the development of acquiescent silence.

This research expands on recent distinctions on the antecedents of voice and silence. It has been established that acquiescent silence theoretically relates to silence caused by resignation, and not fear, which defined defensive or prosocial silence (Hao et al., 2022; Knoll & Van Dick, 2013; Maynes & Podsakoff, 2014). Silence was found to relate to the activation of the BIS (behavioral inhibition system) in cases of low psychological safety (i.e., fear of repercussions for voice; Sherf et al. 2021). We argue that this might not be the case for acquiescent silence. As resignation is caused by the repeated lack of perceived impact of one’s actions, it is expected that acquiescent silence might be caused by an abnormal deactivation of the approach system (i.e., BAS; regulating voice) rather than the activation of the inhibition system (regulating silence; Sherf et al., 2021). Indeed, as Sherf et al. (2021) pointed out, an under-activated BAS should lead to a decrease in voice behaviors. This is supported by the decrease in voiced propositions over time in Study 1.

The protocol used in our studies allowed for the observation of the effects of different levels of voice instrumentality on individuals’ voice behavior over time (i.e., the number of generated propositions as voice performance). In the repeated low voice instrumentality condition of Study 1, participants showed behavioral patterns that could be explained by learned helplessness: a decrease in voice performance after the first feedback, followed by a slight increase, similar to a reactance effect (Wortman & Brehm, 1975). Finally, performance dropped once again on the last voice opportunity. These results are in line with previous control-based explanations for voice in organizations (Tangirala & Ramanujam, 2008), where low levels of perceived control indeed produced an increase in voice behaviors, but the effect did not last over time, resulting in a lack of motivation to use voice. It is important to highlight an unexpected finding in Study 2, where participants did not exhibit variations in voice performance over time despite the manipulation of voice instrumentality. The recruitment of participants through crowdsourcing platforms could potentially influence experimental procedures assessing behavioral deficits, particularly in cases where extrinsic motivations such as compensation are present.

Implications for Practice

The present findings suggest several practical implications for organizations. To start, it is crucial for organizations and managers to recognize the potential harm of situations where employees share their thoughts repeatedly but see little impact, such as routine suggestion meetings or surveys without follow-up. Continuously suggesting opportunities for input without acting on ideas can negatively affect employees’ willingness to share new ideas and even lead to resignation. Merely offering opportunities for voice is not enough and is even potentially threatening for organizational members and the organization itself. Also, organizations should be cautious, for example, when incorporating new members, who may initially seek to impact their new environment proactively and progressively learn that they cannot (Ashforth & Saks, 2000). Observing the development and progressive installation of acquiescent silence among newcomers should alert managers that individuals are losing a sense of personal control and that actions are needed to restore it (e.g., more challenging goals, training and resources, removal of obstacles to performance, information about job related events; Spreitzer, 1996). Keeping voice instrumental is especially important, as it can be lifesaving in situations where potential work-related dangers or worker mistreatment must be brought to the attention of authorities. We showed that external motivators, like monetary rewards, may not effectively encourage employees to continue sharing ideas if those ideas are consistently rejected. Study 2 findings reveal that even when employees persist in proposing ideas, there is a sense of exhaustion when those suggestions never make an impact (Chung et al., 2017). This has significant implications, especially for innovation-focused teams like research and development. The continuous dismissal of proposed innovations may lead to a reluctance to suggest and champion new ideas, significantly affecting organizational performance.

Limitations

The limitations in our studies should allow for a better understanding of our results and open future perspectives. As the experimental procedure was designed to focus on one psychosocial mechanism (i.e., control perceptions with voice behavior over time), there are four main limitations to consider. First, in our studies, acceptance of voiced propositions was the sole indicator of voice instrumentality, whereas in real organizational contexts, voice must be actively listened to, and the proposals must genuinely impact one’s environment to be considered as having a high level of instrumentality. Our protocol was deficient in this aspect, mainly due to the impracticality of acting upon every proposal from each participant in a real-life setting. Furthermore, online participation with a cover story might have influenced the motivation to genuinely participate in the study. Second, due to ethical considerations, the experiments were brief, lasting approximately 15 minutes. It is important to consider two important aspects of this point: (1) voice non-endorsement in real life settings life does not typically occur in such short bursts but rather over extended periods, in various forms, and from diverse sources. Nonetheless, within this short span, we were able to set up conditions conducive to learning voice non-contingency; (2) during this brief experimental duration, we observed the immediate repercussions of repeated low voice instrumentality. However, the lack of long-term measurement clouds the persistence of these immediate consequences. Under different conditions, a longer duration might permit more repetitions of low voice instrumentality, potentially leading to more severe consequences over extended periods. Third, the generalization of our findings to real organizational settings can be constrained by the low ecological validity commonly associated with experimental studies. Factors such as the temporal aspect mentioned earlier, potential participant disinterest, the constraints inherent in online experiments or the elimination of potential moderator for voice (e.g., type of leadership or individual dispositions; Hu et al., 2018; Yi et al., 2022) contribute to this limitation. Despite these challenges, our efforts were directed at creating an experimental scenario that closely simulated instances where individuals express their opinions on significant organizational topics. Even in those cases where the implementation time was brief or participant interest could have been low, the experimental conditions remained credible, as evidenced by the results obtained. Fourth, in our procedure, only one form of voice was used: requesting suggestions from organizational members. Results may be different when examining other voice systems (e.g., suggestion boxes, open door policies).

Notwithstanding their limitations, these studies carry significant ramifications for the advancement of theories related to voice and silence in organizations, as well as implications for practitioners.

Conclusion

People may stay silent in organizations because they perceive that their voice is futile. Repeated failures to achieve a desired goal lead people to develop learned helplessness. Our results indicate that the choice to voice ideas, propositions, or concerns in organizations or to choose to stay silent is influenced by learning the lack of impact of voice in those settings. The present research advances the literature on voice and silence by providing an experimental and theoretical framework for understanding how resignation to voice can emerge in organizations as a learned helplessness phenomenon. This approach emphasizes individual learning through past experiences in organizational contexts.