1. Introduction
Constituents use local public meetings to exert influence over the local policy-making process. However, some studies have concluded that public meetings in general either have minimal effects on policy outcomes or are used in self-serving ways by officials in order to claim that they have solicited “citizen input” on policy proposals (Checkoway, 1981; Cole & Caputo, 1984; Ratliff, 1997) . On the other hand, others have disputed what McComas calls the “minimalist” view of public participation, the use of public meetings by agencies “to satisfy minimum legal requirements for public participation without ever giving much weight to the public’s input” (McComas, 2001: p. 38) . For example, Rosener (1982) found that public meeting participation does affect outcomes—citizen participation at public hearings had an impact on the denial rate of permits under consideration by the California Coastal Commission. A more recent study (not conducted in the U.S.) identifies the conditions under which constituent input can shape the policy decisions of public officials (Migchelbrink & Van de Walle, 2020) .
On the other hand, other studies have considered the utility of public meetings that are not confined to their policy impacts (McComas, 2001; Halvorsen, 2003; Adams, 2004) . Evidence suggests that officials conducting public meetings perceive the criteria of success as not only encompassing outcomes success—the adoption of constituent recommendations into policies, but also process success—the exchange of genuine dialogue between public officials and constituent participants (McComas, 2001) . Moreover, high quality dialogue in public meetings can contribute to constituent perceptions of institutional responsiveness (Halvorsen, 2003) . Additionally, Adams (2004) argues that while citizen participation in public meetings, specifically city council meetings, does not always have immediate impacts on policy proposals, they may serve other important functions, such as conveying information about public opinion to political officials, allowing citizens to occasionally set the agenda, and causing delay in the voting of an agenda item, giving citizens time to consider and present alternative proposals.
The common theme underlying the concerns in these studies is whether officials are actually listening to the public. For some, listening might require nothing short of implementing constituent-recommended proposals. For others, listening need not encompass such policy adoptions but rather involves an activity that orients officials to an understanding of and receptivity to constituent concerns (McComas, 2001; Baker et al., 2005) . While research has focused on the former type of listening, very little is known about the extent to which public officials actually listen to constituent comments in public meetings. Moreover, constituents commonly observe that officials conducting these meetings appear distracted by other activities, such as using their cell phones and talking to one another. Perhaps such “non-listening” is inconsequential if officials ultimately adopt constituent recommendations into public policies. However, this view of listening is narrow for various reasons. First, listening, when defined as “the active and dynamic process of attending, perceiving, interpreting, remembering, and responding to the expressed (verbal and nonverbal) needs, concerns and information offered by other human beings” (Purdy, 1997: p. 11) , does not require constituent success in persuading public officials. Secondly, there is rarely an uniform public opinion on any given issue in public meetings and thus, officials often fail to satisfy the preferences of all interested groups on a contentious policy issue. This, however, does not mean that officials fail to listen to groups whose preferences were not implemented into policy.
Thirdly, in the absence of government compliance with constituent preferences, listening is especially important. Attending public meetings can be a costly activity, as demonstrated by those who do not attend because of limited resources that fail to compensate for the inconvenience of attendance (Halvorsen, 2003) , and therefore, those who do participate expect at the very least they will be heard, particularly when their recommendations are not implemented into policy. Listening can help promote understanding between constituents and officials when public officials not only consider constituent recommendations but also enable them to recognize the rationale for certain policy outcomes (Baker et al., 2005) . In doing this, officials help constituents realize that meeting participation is not a wasted effort.
While scholars evaluating public meeting activities should take auditory listening seriously, they should also look into inequities that potentially influence the behavior of officials conducting public meetings. Past studies have certainly paid attention to problems of social equity in public participation and have called for designing better processes to attract more participants, especially those who have been and continue to be marginalized from the political system (Bryson et al., 2013; Clark, 2018) . The assumption is that without the participation of these groups, a range of perspectives will not be considered. However, when such groups participate in public meetings, are their voices acknowledged and understood? Unfortunately, there is a paucity of empirical research on how public officials treat non-White participants relative to their White counterparts in public meetings. To what extent does racial group membership affect the inclination of officials to listen to constituent messages in public meetings?
On a general level, this article encourages scholars to consider the listening behavior of officials as one of several important metrics with which to assess the outcomes of public meetings. More specifically, it investigates the potential group disparities that might affect this listening behavior. Looking specifically at public hearings and open commentary periods in city council meetings, the author investigates the degree to which the race of the council member and constituent affects specific dimensions of the listening process—the recall and understanding of constituent messages. The author finds limited support for the expectation that shared racial identity affects the extent to which city council members recall and comprehend constituent messages. However, given the limitations of the study, the author urges caution in interpreting the results and suggests future directions for research and ways to refine the study in order to draw more generalizable conclusions.
2. Studying Listening in Local Public Meetings
While listening is a multi-step process, it is difficult to assess in its entirety the listening that occurs in public meetings. For example, if rules governing a council meeting forbid members from responding to public statements, then scholars will be unable to assess the response dimension of listening. This article focuses on the recall and comprehension part of listening. Admittedly, recalling and understanding information does not capture the comprehensive process of listening. For example, the failure to recall a message some time after a meeting does not mean that public officials were not listening to a constituent during the meeting. However, in other instances, failing to recall a message could actually reveal that the official consciously tuned the speaker out during the meeting. Furthermore, even if officials listened to every constituent who spoke, they are unlikely to remember every single message. Thus, the retention or discarding of messages may reveal a pattern premised on the social identity of the constituent.
Message comprehension is also an important component in the listening process because it affects how public officials respond to the message. If public officials misunderstand what constituents say in these meetings, they will not be able to offer a proper remedy or response. The ability to comprehend messages, especially in a way the speaker intended the message to be understood, depends not only on the attentiveness of the listener but also on the interpretation or meaning the listener assigns to the message. While the lack of or limited comprehension can stem from poor language enunciation or differences in native languages and jargons between individuals, it can also be tied to the listener’s eagerness to tune out or dismiss the speaker’s message because the listener and speaker possess different group affinities, perspectives, and frame of references. On the other end of the spectrum, some might depict a good outcome of comprehension as something that captures a high level of “listening fidelity”, reflecting a high degree of congruence between the cognitions of the listener and that of speaker (Mulanax & Powers, 2001; Powers & Bodie, 2003) . In ordinary parlance, the listener more or less understands the statement in the manner the speaker intended. Processes that facilitate this kind of comprehension include, but are not limited to, the reasonable inferences and evaluations the listener makes regarding the message the speaker intends to communicate. Moreover, good comprehension does not require that the listener be convinced by the message if the message is intended as persuasive communication.
However, outcomes of comprehension cannot simply be dichotomized as “good” or “bad” understanding but oftentimes falls somewhere between the two extremes. At times, listeners may engage in “distorted” listening, interpreting the meanings of the speaker’s words through biases and filters that prevent them from understanding the speaker’s message (Bickford, 1996; Dobson, 2012) . In distorted listening, the individual may understand the message’s basic point, but the individual goes beyond the original speech to inappropriately contextualize the information in that speech. While Gaines et al. (2007) are not explicitly concerned with distortion in auditory listening, their framework helps illustrate how social identities can affect the interpretative processes to render a meaning for the listener that differs from the meaning given by the speaker. For example, they discuss how interpretation of facts occurs through one’s evaluation of and explanations for the information contained in those facts. In their study of the effects of partisanship on message interpretation, Gaines and his colleagues found that partisanship influenced whether individuals perceived a certain number of casualties from the Iraq War as high, moderate, or low (Gaines et al., 2007: p. 959) . In other words, Democrats and Republicans received the same information about the number of deaths but drew different meanings from it. Gaines and his colleagues also found that Republicans and Democrats attributed different meanings to the same fact—“the U.S. did not find weapons of mass destruction in Iraq”—because they offered different explanations for the lack of weapons (“they were never there” vs. “Iraq either hid them or destroyed them”). They add that drawing inferences (such as a future state of affairs) from the information given can also affect the meaning assigned to a message. Inferences made about the speaker’s motives or character can also impact how messages are interpreted. In general, one can distort a message by inappropriately contextualizing the information in the message, such as drawing certain inferences from the information or giving certain evaluations of or explanations of it.
3. Theoretical Expectations: The Influence of Race in Public Meetings?
While social identities like partisanship affect how one perceives a message, does race exert a similar effect on officials tasked with facilitating local public meetings? Does race affect an official’s inclination to ignore, misunderstand, or distort constituent messages in public meetings? Empirical evidence suggests that public officials are less attentive to the messages of racial minorities than those of Whites, specifically in non-public communicative interactions. Results from field experiments reveal that even when bureaucrats received the same email from alleged constituents with either White or Hispanic aliases, they were less likely to respond to emails from the latter than the former (White et al., 2015) or provide less friendly responses to the latter (Einstein & Glick, 2017) . Additionally Butler and Broockman (2011) found that White state legislators, even among those who identified as Democrats, were more likely to respond to emails with White aliases than those with Black aliases. On the other hand, minority legislators were more likely to respond to constituents they perceived as Black than to those they perceived as White. While such studies provide evidence of racial biases in “figurative” listening, such studies provide inconclusive information about racial biases that may occur in public interactions between public officials and constituents. On the other hand, Hoang’s observational study (Hoang, 2019) suggests that racial differences in the verbal responses that officials issue to constituents in public meetings are conditional upon the statements constituents make. However, given the constraints of some public meetings, the lack of a verbal response does not constitute sufficient evidence of non-listening. Additionally, cursory verbal responses do not fully capture racial differences in other aspects of the listening process, such as how well officials comprehend messages.
While prior studies show that racial biases influence the attention given to a message in the first place, studies in linguistics suggest that the racial identity of the speaker also affects how the message is interpreted and subsequently understood. In reverse language stereotyping (RLS), “attributions of a speaker’s group membership cue distorted perceptions of that speaker’s language style and proficiency” (Kang & Rubin, 2009: p. 442) . In one investigation of RLS, Kang and Rubin (2009) found that ascribing the same spoken lecture either to a White or East Asian instructor affected the students’ ability to detect the two lectures as being voiced by the same person. Additionally, a person’s propensity to infer negative traits about a speaker based partly on the speaker’s social identity predicted lower levels of comprehending the lecture ascribed to the East Asian instructor, who was also depicted as being a non-native speaker even though the lecture was delivered with no distinct accent. Such findings are consistent with the conclusions of previous studies (Rubin, 2002; Lindenmann, 2002, 2003) , especially one similarly concluding that ascribing a racial and language proficiency identity to a speaker affects how that speech will be assessed and interpreted (Rubin, 2002) .
The impact of race on message comprehension is not confined to the instructional setting. In an experimental study with a cautionary note on the optimisms of racial cue-taking, Kuklinski and Hurley found that Blacks were more likely to agree with a certain statement if the message was attributed to an African American leader than to a White leader even in circumstances where “ideological reputation…presumably could serve as an even more telling contextual information than race” (Kuklinski & Hurley, 1994: p. 748) . Furthermore, African Americans varied their interpretations of the same statement depending on the racial identity of the alleged source of the message.
Given that social identities, particularly race, affect message comprehension and a person’s inclination to disregard messages, we might expect these relationships to play out in public meetings. While the aforementioned studies on race and message interpretation focus on non-elites and students, public officials and other elites are certainly not immune to racial biases in their thinking, as previously discussed. Furthermore, bias toward one’s co-ethnics exists among Democrats (Butler & Broockman, 2011) and need not be confined to the behavior of Whites. In the case of Black officials, however, preferential treatment toward in-group members might occur for reasons different from those that motivate White racial bias. Perhaps shared experiences with racial marginalization in the United States propel Black members toward identification, sympathy, and/or empathy when listening to the views and experiences of their co-ethnics speaking at public forums, even if the constituent’s statements have no bearing on race-related matters. Moreover, Black council members may feel that Black constituents do not receive adequate representation from the political system or from White council members, so they may be especially inclined to listen more attentively than usual when their co-ethnics voice their concerns at public meetings.
Overall, the author investigates the expectation that local public officials are more likely to remember the messages of co-ethnics than members of out-groups. If true, the evidence will manifest as higher levels of recalling messages spoken by co-ethnic members. Additionally, the author investigates the extent to which these officials comprehend the messages of constituents who do not racially identify with them.
4. Methodology
While many types of public meetings exist, the study focuses on local government meetings because they occur regularly and frequently. Additionally, compared to single-issue public meetings, city council meetings are more likely to attract participants who run the gamut of policy interests since councils takes up many different local issues. The data for this study was derived from observations of council meetings that took place in 2014 and 2015 and interviews with city council members hailing from the southern region of Michigan, specifically four small to moderately-sized cities in the region. The councils in the study varied according to size and gender and racial composition. Councils ranged from a nearly homogeneous White council to one that was composed exclusively of racial minorities. However, almost all members on these councils, including the non-partisan councils, identified as Democrats. The cities represented by these councils ranged from being majority-White to majority-minority, and the median household income for most of the cities was below the state median. While among the cities chosen, considerable variation existed along certain dimensions, no large cities were represented, and partisan heterogeneity was lacking among the council members. Therefore, much caution should be taken in extrapolating the results of this study, as the author later discusses. Table 1 below summarizes the characteristics of the cities and their respective councils in the sample.
![]()
Table 1. Demographic information for councils and municipalities.
4.1. Collecting the Data
After each meeting in question, the author transcribed nearly verbatim the verbal statements of every constituent who participated in the meeting. The author then conducted one-on-one, semi-structured interviews with 15 council members who agreed to speak with the author. A few council members agreed to participate in only one interview. The majority of members consented to do two interviews; and a handful agreed to be interviewed a third time. Overall, the author conducted a total of 31 interviews. In the interviews, the author asked each council member if that member was able to remember what each constituent had spoken about during the meeting and if so, what the member understood the constituent to be conveying to the council. Each member commented on the messages of anywhere from 6 to 17 constituents in a given interview.1 In order to minimize suspicion that the author was investigating the members’ listening behavior, the author asked several questions soliciting members’ perceptions of the central purpose of public hearings and their views about alternative resources for information about public opinion. After the interviews, each member’s responses were transcribed and then paired with the relevant constituent statement.
Thereafter, three students, who were unaware of the study’s goals and the social identities of the subjects, were instructed to read the paired statements and evaluate on a Likert scale how well they thought the member remembered and understood the constituent’s message.2 They were asked to rate the member’s listening behavior on a scale from 1 to 5, with 1 representing a complete lack of recall or comprehension of the message; 2 representing marginal recollection or comprehension; 3 representing moderately good recall or comprehension; 4 representing very good recollection or comprehension, and finally, 5 representing extremely good recall or comprehension.
Not surprisingly, ratings of comprehension were mostly based on and tied to ratings of recall. For example, interview responses revealing a conscious decision to ignore the speaker or the inability to recall a message resulted in a rating of 1 for the recall and comprehension of that particular message. Likewise, extremely good comprehension was tied to high levels of recall. However, in their assessments of comprehension, evaluators were instructed to not only consider the extent to which council members recalled or dismissed the constituent’s message but also the degree to which they thoughtfully reflected on the information in the message. Thus, responses demonstrating high levels of comprehension may show the member’s efforts to think critically about the information in the message even if the member only moderately recalled the message. Evaluators were also instructed to consider in their ratings of comprehension the degree to which members distorted the constituent’s message (e.g. offered unwarranted or unreasonable explanations of or inferences about the message or the messenger’s motives). Therefore, in instances of very high recall, a member’s comprehension may receive a lower rating if evaluators determined that the member listened to the message in a distorted way.
4.2. Coding the Responses
Below are examples of how the interview responses would be coded.
Example 1: Dismissive Listening
Constituent A attends city council meetings regularly and frequently expresses a variation of the same message, even during public hearings where his message may not apply. In one particular meeting, he made following comments:
“I am in favor of requiring that the amendment be attached to all annexation issues, zoning and site plan issues brought before the council that would require open access to the property involved by members of all levels, including certainly the most vulnerable residents of the city and that the attendant opportunity for transportation to the properties through para-transit, handicapped transportation, and senior ride transportation… (the constituent doesn’t complete sentence). This should be a blanket tape amendment that would assure that the historic prejudice involved and bigotry involved in red lining within the city and the county be overcome and be eliminated. This is too proud of a city, with two prestigious educational institutions together—with the University’s Law School inside of the city and the impressive educational institutions, including the University—that we have today the status quo, where properties are annexed, but they are not annexed under conditions requiring opportunity for access by people of all income levels and transportation to these sites by people of all income levels. This is something that is a historic source of bigotry and discrimination and ongoing black eye to the city. Needs to be redressed. Thank you.”
The responses from Council Members A and B below show a disconnected orientation toward the constituent’s statement. Therefore, they were also unable to comprehend the message. Evaluators were instructed to give such responses a rating of 1 for recall and comprehension.
“I don’t know precisely what this constituent said. I’ve taken to not listening to this constituent. He is someone who comes and speaks at every single meeting and at every single opportunity, at the exact same topic, which is generally speaking, social equity and affordability—affordable housing—social and economic equity…”
- Council Member A
“I don’t remember what he said. I kind of tuned him out.”
- Council Member B
Example 2: Very Good Comprehension
While the responses from Council Members A and B show that they consciously tuned out the message of Constituent A, Council Member C’s response demonstrate not only a willingness to listen to the same constituent but also an effort to reflect on the statement.
“He has important things to say. But he isn’t judicious in how he says them. He spoke later at two different public hearings. And I found myself being distracted by one of the things he was saying because he was saying that we shouldn’t accept a property into the city before guaranteeing that that property had access to the bus, which is okay, fine except that property HAS access to the bus line. And so telling us this was not helpful.”
“I actually spent time looking up the property itself and looking at its adjacency to the #8 Bus. Because I felt like, ‘Here it is. Here is the bus.’ But the constituent wants that included in everything. He wants it to be part of the public record that every parcel is within a certain distance, every parcel is handicap accessible, every parcel is earmarked for affordable housing. And it’s hard to say if he’s being reasonable about those things because he seems to simply be automatic. He’s not looking at the property. He’s not doing any research. He’s not determining the best way to make an impact. If instead he said, ‘I looked at a map, I checked the bus routes, and the closest bus comes here only once an hour and you have to walk half a mile to get it. That’s not acceptable. So if we’re bringing this into the city, we should be working with city public transportation authority to ensure that the use of this property is enhanced by the access to the bus. THAT would be compelling.”
While the response reveals the member’s disagreement with the constituent’s views, it also shows that the member not only remembered, but also understood, the point the constituent was attempting to convey. Additionally, the response shows that the member had critically thought about the reasons for the message’s lack of persuasiveness and the necessary considerations that would make the message more compelling. While such a response may warrant a rating of moderate recall (3), students were instructed to consider giving responses like this an assessment of high comprehension (4).
Example 3: Distorted Listening?
In other situations, however, council members neither completely detach themselves from the constituents speaking nor fully understand the constituent’s message as the constituent intended. Consider the following view expressed by the constituent, who attended a council meeting to encourage council members to vote against a resolution that would cut funding to his organization. Constituent B said:
“I just want to speak and to encourage you NOT to approve Amendments #1 and #15 and to encourage the Council to continue to be a part of the successful public-private partnership that has created great results in this city over the last nine years. Those other partners include the county, the University, 35 companies, all the other municipalities in the County that are part of that partnership. I did provide through the city manager this particular handout. I think we’ve all received it. It clearly spells out our results, specifically in the city. Last year, your investments resulted in $21 million of projects and 752 jobs. I’d like to point out that the 752 jobs are half of all the jobs that S Company (constituent’s company) was able to develop through partnerships throughout the County.” (The constituent then continued talking about the projects and benefits that his company had brought into the city).
When Council Member D is asked to provide an account of what he/she perceived to be the constituent’s message, the member expressed:
“Yeah, this speaker is one of those—he’s the head of S Company. He just basically said: ‘give us all the money.’ Well, it’s just corporate welfare. So, I don’t have much sympathy for corporate welfare because I think rich people have all the sources in the world to help them. Poor people and ordinary people don’t, so I feel that my duty is to make sure that I take care of people who don’t have other resources. I’m sorry—he just gave the same old kind of feedback—‘oh we’re just doing so much good, just give us all the money; we need all your money.’ I don’t know which amendment he came to speak about but basically Amendment #1, which is mine (that was) basically going to cap the income (that) we give to S Company. The other one was taking out $75,000 we give to S Company for marketing and giving it to (the) homeless shelter. So he opposed both of (those amendments).”
The response above distorts what the constituent had said. Although the member recalls some details of the constituent’s argument and understands the basic point conveyed—he doesn’t want the Council to reduce funds to his company– the member, without justification, attributes the motives of the speaker to being greedy and offers an exaggerated account of his request as desiring to “take all the money.” While good listening does not preclude the member from disagreeing with the speaker on the value his company contributes to the city, the distortion of the message prevents the listener from fully hearing the arguments of the speaker. Thus, such listening induces the member to downplay the accomplishments of the company and is likely to prevent the member from considering that perhaps the funds being curtailed are necessary for the company to thrive. The author instructed the student assessors to refrain from giving this type of response a high comprehension score (4 or 5) because a significant portion of the response does not accurately characterize the constituent’s statement. In other words, if the author had shown the constituent this response and asked whether such a view reflected his request to the council, the constituent would have rejected such a characterization of his message.
4.3. Compiling the Dataset
After every student completed the assessments, the author took several steps to determine the overall rating of each member’s response. First, the author checked to ensure that each student appropriately assigned the ratings at the extreme ends—that the ratings of 1 or 5 did indeed reflect, respectively, complete lack of recall/comprehension or extremely good recall/comprehension.3 While the three evaluators were given the same detailed instructions on how to rate the responses, differences inevitably arose in the ratings. For both recall and comprehension assessments, there was agreement among the three raters in approximately 60% of responses. However, for most of the remaining responses, two of the three raters offered the same or very close ratings. In the instances of diverging ratings (even if slight), the author and a fourth evaluator extensively discussed the ratings and members’ responses and reconciled the differences. In most cases, the author and fourth student coder assigned a final score that mirrored the one agreed upon by two of the three original raters.
A dataset was then derived from the ratings given to each member’s response and the information collected about the members interviewed and the constituents who spoke at the meetings. The unit of analysis is the dyadic relationship between each constituent who spoke and each council member who commented on the constituent’s message. Thus, a given interview yields, for example, 100 observations if a member was asked to comment on the messages of 10 constituents who spoke at the last meeting. Overall, interview responses yielded a total of 389 observations. From these observations, 296 were from 120 White constituents while 93 were from the 47 non-White constituents in the sample (39 observations were associated with 24 Black constituents, 28 with 13 Latino constituents, 8 with 3 Asian constituents, and 18 with 7 minority constituents who are neither Black, Asian, nor Latino). Also, of the 389 observations, 305 were from 11 White council members, 48 were from 3 Black council members, and 36 were from 1 non-Black minority council member.
4.4. Dependent Variables
The main dependent variables capture how well the member recalled and understood the constituent’s message, according to the assessments of independent evaluators. Because each exercise asked for a rating based on a Likert scale, the dependent variable is an ordered variable. The original scale asked evaluators to assess recall and comprehension on a scale from 1 to 5. However, significantly fewer ratings of 5’s and 1’s were assigned than those in between the extreme ends of the scale. With few observations at the extreme ratings, an ordered logit model with a five-level dependent variable is unlikely to generate results when control variables are included. Therefore, I collapsed the original 5-level “recall” and “comprehension” variables into 3-level variables. The first level of the variables captures recall or comprehension assessed as either completely lacking (1) or negligible (2). The second level captures moderately good recall or understanding (3). The third level captures recall or comprehension assessed as very good (4) or extremely good (5). Overall, in the modified recall variable, there are 123 observations (31.62%) associated with the first level, 113 observations (29.05%) associated with the second level, and 153 observations (39.33%) associated with the third level. In the modified comprehension variable, there are 74 observations (19.02%) associated with the first level, 120 observations (30.85%) associated with the second level, and 195 observations (50.13%) associated with the third level.
4.5. Independent Variables4
Race of the constituent and member constitutes the primary independent variables. The constituent’s race is a nominal variable with three categories: White, Black, or non-Black minority.5 Likewise, the member’s race is a nominal variable of three categories: White, Black, or non-Black minority. The variables representing the member and constituent’s race are interacted to test the expectation that a member better understands or recollects a constituent’s message when the constituent speaking shares the member’s race.
Additionally, the author included in the analysis variables that may impact the relationship between race and a member’s recall or comprehension of constituent messages. The author controlled for the constituent and member’s gender. The author also included a dummy variable that reflects whether or not the constituent is at least 50 years old (Age 50+). Given that significantly older individuals attend these meetings, members may be more inclined to listen to these constituents than to their younger counterparts. Furthermore, the author controlled for the independent effect of constituent statements about legislation (Speak on Legislation). Since a significant portion of public comments in these council meetings is directed at legislative matters being debated by the council, local legislators may feel compelled to be more attentive to these messages than to messages unrelated to prospective local ordinances.6 Another variable accounted for was the expression of an opinion that was similar to a view asserted by another constituent during the same meeting (Same Opinion as Another Constituent). This dummy variable also captures whether or not the constituent comes to speak as a member of a group. All else equal, council members may be less likely to remember the details of a particular individual’s message if the message is similar to that of another and especially if it is similar to those of several other constituents. However, they may be more likely to understand a message if they hear it from others more than once or if they hear it repeatedly. To consider the effects of a particular council’s culture or norms on interactions with constituents, the author included fixed effects or dummy variables for three of the four cities (City B, City C, and City D).
In addition to considering individual and institutional-level characteristics that potentially impact the relationship between race and listening, the author controlled for aspects of the interview process that can affect such a relationship. Members may experience more difficulty recalling and hence understanding constituent messages with each additional day that elapses between the meeting and the interview. Thus, the author included a continuous variable to capture the number of days that had passed from the night of the council meeting to the day the council member was interviewed (Days Elapsed). Finally, the quality of a member’s recall and comprehension of constituent messages may also depend on the length of the interview. Some officials had limited time to converse with the author, as they agreed to be interviewed during a short lunch break or in the morning before work. Thus, it is conceivable that a council member who interviews for approximately one hour may provide more thorough and vivid responses concerning each constituent’s message than someone who only has 20 minutes to converse with the author. Hence, the author included a continuous variable to capture the total number of minutes the council member spoke (Length of Interview).
5. Results
Table 2 displays the descriptive statistics associated with the members’ levels of recall and comprehension of messages, disaggregated by the race of the members and constituents. It shows that the levels of recall and comprehension exhibited by White members appear to vary with the constituent’s race. Perhaps surprisingly, the descriptive statistics suggest that White constituents are not particularly advantaged when it comes to having their messages heard or understood by
![]()
Table 2. White council members only—recall and comprehension.
White council members. In fact, White members seem to recall and comprehend better the messages of non-Black minorities than those of Whites.
On the other hand, while Black constituent messages seem less likely than those of their non-Black counterparts to elicit moderate recall and comprehension levels from Black members, their messages seem to exceed those of their non-Black counterparts in eliciting the highest level of recall and comprehension. Additionally, Black members appear less likely to dismiss or only minimally recall and comprehend the messages of Black constituents than those of their non-Black counterparts. In general, descriptive statistics seems to show that shared racial identity does affect how Black members recall and comprehend the messages of their constituents (Table 3).
![]()
Table 3. Black members only—recall and comprehension.
While for the most part, the descriptive statistics show that shared racial identity does not exert similar effects on the recall and comprehension levels of Black and White council members, a regression model can determine whether racial differences in comprehension and recall levels are statistically significant. Again, the unit of analysis in the data is the dyadic relationship between each constituent who spoke and each councilmember interviewed. Therefore, the observations are not independent because each council member and each constituent compose several dyads and are crossed with one another. Furthermore, because the dependent variables are composed of ranked categories, I use a mixed effects ordered logit regression to analyze the data—more specifically, the model is a crossed random effect model, in which both constituents and members are treated as random effects.
5.1. Results from the Multilevel Models: Ordered Logit
The models in Table 4 show the impact of race on council members’ recall of constituent messages. With the inclusion of relevant control variables, the first model provides a test of the independent effect of White racial advantage. The second model, which includes the interaction of race variables, provides a test of the conjecture that council members recall better the messages of their co-ethnic constituents. The results support some of the expectations discussed in this study. Not surprisingly, members recall messages less when they are confronted with
![]()
Table 4. The effect of race on message recall.
Notes: Entries are coefficients and their standard errors from a crossed random effects ordered logit model. The dependent variable is level of message recall (no/little recall, moderate recall, high/very high recall). White constituent and White member are the reference categories. In Model 2, observations associated with the one non-Black minority member are excluded from the analysis. Cutpoint 1 is the estimated cutpoint on the latent variable used to differentiate no/very little recall from moderate and very/extremely high recall when the values of the independent variables are set at zero. Cutpoint 2 is the estimated cutpoint on the latent variable used to differentiate very/extremely high recall from moderate and no/very little recall when values of the independent variables are zero. Recall with a value at cutpoint 2 or greater would be classified as very/extremely high recall when the independent variables are evaluated at 0. Recall with a value between cutpoint 1 and cutpoint 2 on the underlying latent variable would be classified as moderate recall. For a discussion of interpreting cutpoints associated with the results of ordered logit models, please refer to the following website: UCLA Advanced Research Computing: Statistical Methods and Data Analytics from https://stats.oarc.ucla.edu/stata/output/ordered-logistic-regression/. +p < 0.10. *p < 0.05. **p < 0.01.
constituents who express a viewpoint that is also conveyed by others in the same meeting. Also, as expected, there appears to be an inverse relationship between the level of message recall and the number of days that have passed between the meeting and the interview. Other controls, such as the gender and city level variables, also affect recall, but its effects are only marginally significant at the 0.10 level. Specifically, male council members are assessed as having lower levels of recall than their female counterparts. Furthermore, council members of City A recall constituent messages less than their counterparts in City C. The coefficients of these variables, with the exception of gender, retain similar significance levels in the interaction model (Model 2).
As for the primary variable of interest, I find no evidence to support the conjecture that White constituents are generally advantaged over their non-White counterparts in having their messages remembered. In fact, members demonstrate higher levels of recollecting messages articulated by non-Black minorities than those expressed by White constituents, although the effect of being a non-Black minority is not statistically significant at the 0.05 threshold. Additionally, there is insufficient evidence to indicate that White members exhibit higher levels of recalling messages expressed by White constituents than those conveyed by their non-White counterparts. On the other hand, the positive interaction term in Model 2 suggests that Black members remember better the messages of Black constituents than those of White constituents (although the interaction coefficient falls short of the 0.05 significance threshold).6
Table 5 reports the results from the models predicting the relationship between race and message comprehension while controlling for other relevant factors. Similar to the results associated with message recall, the author finds that interview features affect members’ comprehension levels, with the increasing number of days the interview is delayed having a negative impact on a member’s comprehension of the message. However, the coefficient for this effect is marginally significant at the 0.10 threshold. City-level factors again exert their marginally significant effects, as evaluators rated the comprehension levels of members from City C better than the comprehension levels of members from City A.
![]()
Table 5. The effect of race on message comprehension.
Notes: Entries are the coefficients and their standard errors from a crossed random effects ordered logit model. The dependent variable is level of message comprehension (no/little understanding, moderate understanding, high/very high understanding). White constituent and White member are the reference categories. In Model 2, observations associated with the one non-Black minority member are excluded from the analysis. +p < 0.10. *p < 0.05. **p < 0.01.
Interestingly, I do not find evidence that council members, in general, or White members, in particular, better comprehend the messages of White constituents than those of their non-White counterparts. However, as shown in Model 1, messages expressed by non-Black minority constituents elicit higher levels of comprehension than the messages expressed by White constituents, although this difference falls short of the 0.05 significant level. Moreover, this result is driven by the behavior of White members, as shown by the coefficient associated with the “Other Minority” variable in Model 2.7 While White constituents do not appear to derive an advantage over minority constituents, especially in the listening outcomes of White members, the converse seems to be the case for Black constituents. Specifically, as shown in Model 2, Black members express higher levels of comprehension when Black constituents convey the message than when their White counterparts do so.
5.2. Predicted Probabilities of Comprehension
Because coefficients of ordered logit models are difficult to interpret given that they are in log-odds, I report how the predicted probabilities of low, moderate, or high recall and comprehension vary according to the constituent and member’s race. The predicted probabilities are calculated with the control variables set at theoretically meaningful values. The results are shown in Table 6 and Table 7. The predicted probabilities reported and discussed below are those associated with race coefficients that exert statistically significant effects on the members’ behavior at the 0.05 threshold.8
![]()
Table 6. Predicted probabilities of each outcome for comprehension (When Member is Black).
Variables held at the following theoretically meaningful values: Black member = 1; male member = 1 male constituent = 1 M, days elapsed = 2; minutes = 30; same opinion = 0; City B = 1; Age 50 and above = 1; +p < 0.10. *p < 0.05. **p < 0.01.
![]()
Table 7. Predicted probabilities of each outcome for comprehension (When Constituent is Black).
Variables held at the following theoretically meaningful values: Black constituent = 1; male member = 1 male constituent = 1 M, days elapsed = 2; minutes = 30; same opinion = 0; City B = 1; Age 50 and above = 1; +p < 0.10. *p < 0.05. **p < 0.01.
As Table 6 shows, the racial difference in the predicted probabilities of poor (or a lack of) comprehension among Black members is not statistically significant. In other words, there is insufficient evidence to suggest that Black members are more likely to misunderstand or only marginally understand the messages of Whites than those of their co-ethnics. On the other hand, while Black members are less likely to moderately understand the messages of Black constituents than those of White constituents, they are more likely to understand very or extremely well the messages of Black constituents than those of White constituents. Furthermore, this difference of 32.3 percentage points is not only statistically significant but substantively significant as well.
Black members not only comprehend better the messages of Black constituents than those of White constituents but they are also more likely than White members to understand better the messages of Black constituents. Specifically, as shown in Table 7, Black members are less likely than White members to fail to comprehend or only marginally comprehend the messages of Black constituents. Furthermore, while they are less likely than White members to moderately understand the messages of Black constituents, they are significantly more likely to understand their messages very or extremely well. In listening to messages from Black constituents, the predicted probability of high message comprehension among White members is approximately 0.58, which is fairly substantial, but increases to approximately 0.94 when the member is Black, which means that Black members almost always understand very well the messages of Black constituents.
6. Discussion
In general, I find limited support for the expectation that shared racial identity positively affects the member’s understanding or recall of constituent messages. According to the results, Black members in this study understand better the messages of their co-ethnics than those of out-group members. They are also likely to exceed White members in their ability to understand the messages of Black constituents. Given that race significantly shapes lived experiences, inevitably giving rise to different frame of references and perspectives to understand social realities, these results may not be surprising. However, as will be discussed shortly, caution should be taken in generalizing from the findings.
On the other hand, the results also motivate scholars to consider circumstances in which White members may not racially discriminate in the way we might speculate them to. Specifically, the study yielded no evidence to show that White council members privilege their co-ethnics. Moreover, the results suggest that White members might even be more receptive to messages of certain minority groups, although again, this difference falls short of the standard significance level. While the differences in lived experiences between Whites and non-Whites might result in different policy priorities and diverging perceptions of events, it is possible that the members in this study, most of who identify as liberal, may be conscious of how different life experiences shape the perceptions of various groups, and this awareness might motivate them to make a concerted effort to listen to and understand racial minorities who speak at public meetings.
Furthermore, like their Black counterparts, these members may be cognizant of the political disadvantages minority constituents encounter. For example, in the cities under consideration, racial minorities, especially non-Black minorities, are underrepresented in their active participation in council meetings, even in cities with a plurality or bare majority of racial minorities. Thus, in the infrequent occasion where racial minorities do participate, members may make an effort to be especially attentive. For some members, the recall and comprehension of messages expressed by minority constituents may also be aided by the ease in remembering what is different and perhaps out of the ordinary relative to what is routine and familiar. Therefore, even if a non-White constituent expresses a message alongside several White constituents, this non-White constituent may stand out to White officials, especially if the message is different from the others. The salience of this difference potentially explains why White members may be particularly attentive to non-White individuals who irregularly participate in public meetings. Obviously, the author makes no definitive claims about the motives of the council members in the study but offers potential explanations that future investigations should look into.
6.1. Limitations
This investigation into specific aspects of the listening process comes with limitations, some of which were already stated. While the aforementioned speculations about my findings may be plausible explanations, it is too premature to conclude that Black members exhibit race-related differences in their listening given the very small number of Black council members who were interviewed. These members may not constitute a representative sample of the myriad of Black public officials out there. Furthermore, while the author interviewed more White members, these members sit on councils representing smaller to moderately sized cities, and almost all identify as Democrats, with ideologies ranging from moderate to liberal. Thus, it is unknown whether or not the findings here will hold among conservative council members and those who represent cities in major metropolitan areas. The limited number of interviews and hence the limited number of observations preclude the author from drawing more general conclusions. Thus, future investigation would certainly benefit from additional interviews, especially with officials who are non-White.
However, interviews constitute an imperfect method to evaluate the listening behavior of public officials. For example, the inability to recall constituent messages, even during an interview given one day after the occurrence of the meeting, does not demonstrate that public officials failed to listen during the meeting, although it might reveal a discriminatory pattern regarding the retention of messages. As a result, the recall variable does not make distinctions between those who consciously choose to ignore or dismiss a constituent’s message and those who simply forget the message over time. To be clear, several members did admit to consciously tuning out constituents but others made no such admission even though their responses clearly demonstrated an absence of recollection. Thus, the author cannot, with certainty, differentiate the feature of the recall variable that captures deliberate dismissiveness from the feature that captures the eventual non-retention of information.
Finally, interviews and observational data may be inappropriate for investigating a comprehensive range of listening distortions. To be clear, interview responses can reveal racial differences in the inclination of members to infer questionable or objectionable motives to constituent speakers. During the interviews, a few council members responded in ways that exaggerated the messages of constituents or made inferences unsupported by the information in the message, but the number of these instances was insufficient to detect any race-related patterns. While interviews may help scholars investigate distorted listening as it occurs through certain mechanisms, interviews may be unhelpful in examining other forms of distorted listening. For example, when public officials evaluate an appeal for assistance from a director of a community organization, members may find the appeal unreasonable due to their perceptions that the director is making an excessive number of requests. Furthermore, officials are not necessarily listening in a distorted manner if they happen to disagree with the assessment offered by a constituent who is not of the same race. However, the listening becomes distorted if their evaluation of the same information changes merely because the person presenting the information racially identifies with the officials. Obviously, it would be difficult to determine through observational data and elite interviews the extent to which officials are inclined to engage in this type of behavior because two or more constituents almost never deliver the exact same information with the same degree of emotional intensity at a public meeting. In order to evaluate racial bias in this type of distorted listening, experimental methods are more appropriate.
6.2. Other Considerations
In addition to the implications for future research as noted above, research on listening comprehension can benefit from other considerations. Future studies can look at how the structure and institutional norms of a city’s council meeting might influence public officials’ understanding of constituent messages. For example, are officials more likely to understand what consitutent say when the council reserves time within the meeting to respond to public comments than when the council is forbidden to do so by law? Additionally, future studies can investigate how the widespread use of social media and cell phone recordings might induce public officials to pay more attention to constituent concerns expressed during these meetings. Last, but not least, scholars can also examine how the cultural and educational background of government officials might shape their behavior in public meetings and whether these factors can alleviate racial bias in listening.
7. Conclusion
In summary, this article elucidates the importance of listening in public meetings, a space that provides constituents with the opportunity to shape legislative outcomes. In order for public officials to consider and even respond to constituent opinion, they should be able to properly interpret, contextualize, and hence understand the constituent’s message. Failure to do so not only results in a lack of action but also constituent frustration with officials and the public meeting process. Also, racial differences in comprehension may compound these problems. The results of this study suggest that race does not necessarily impede the ability of council members to comprehend and recall constituent messages. While the evidence shows that shared racial identity positively affects the listening behavior of Black council members, it comes from a very limited number of Black members. Thus, additional scholarship should follow up on this investigation, which to date, has not been undertaken in prior studies on public meetings.
Acknowledgements
I would like to thank the council members for taking time out of their busy day to do interviews with me. Their input and thoughts on public participation are very much appreciated. I would also like to express my gratitude to my research assistants for their excellent work on coding the constituent statements and member responses. I take responsibility for all errors in this paper.
Supplemental File
Appendix A
![]()
Table A1. Demographic Information of Council Members Interviewed
Appendix B: Interview Questionnaire for Council Members
Script: I would like to get your opinion on citizen participation in public meetings.
I. Recall (Go down the list of constituent’s who spoke)
Do you remember what Constituent X said during the public comment period?
II. Content (For the statements that are remembered) prompt each member if they forget
a) What was the main point or problem the constituent spoke about?
b) What was Constituent X asking the Council to do?
III. Reasonable Messages and Unreasonable Messages
a) Do you find Constituent X’s message reasonable (even if you don’t agree with it) or unreasonable?
b) Why do you find Constituent X’s message reasonable or unreasonable?
(Some Potential Responses for Unreasonable Messages)
i) The information in the message appears to be misrepresented.
ii) The constituent or information in the message appears biased.
iii) The constituent appears uninformed about the matter.
iv) The constituent relies too heavily on emotions.
c) Do you happen to know if Constituent X was from your ward?
IV. Perceptions of Public Participation and Public Opinion
a) What do you see as the purpose of the public comment period during the public meetings, such as those of city council, local commissions, school boards, and so on?
b) What sources of information do you mostly rely on to get your constituents’ opinions on issues?
c) Have there been times where public comments affected what you did on council?
V. Additional Information
Please use this opportunity to add any information that you think I would find helpful or to clarify any of your responses.
Appendix C: Marginal Effects for Recall and Comprehension
![]()
Table C1. Predicted probabilities of each outcome for recall (When Member is Black).
Variables held at the following theoretically meaningful values: Black member = 1; male member = 1 male constituent = 1 M, days elapsed = 2; minutes = 30; same opinion = 0; City B = 1; Age 50 and above = 1; +p < 0.10. *p < 0.05. **p < 0.01
![]()
Table C2. Predicted probabilities of each outcome for comprehension (When Member is White).
Variables held at the following theoretically meaningful values: White member = 1; male member = 1 male constituent = 1, days elapsed = 2; minutes = 30; same opinion = 0; City B = 1; Age 50 and above = 1; +p < 0.10. *p < 0.05. **p < 0.01.
Appendix D: Variable Information
![]()
Table D1. Mean/median and distribution of the independent variables.
NOTES
1Please see Appendix A for demographic information on the council members interviewed and Appendix B for a list of questions that the author asked.
2The author did not ask constituents themselves to rate the listening behavior of members because the author would have failed to secure the cooperation of every constituent who spoke at these meetings. Therefore, independent evaluators would have been necessary in such circumstances. Additionally, the constituents may have been able to deduce the identities of the members during the interviews even without the author’s revelation of the response source.
3At times, the author essentially recoded one rater’s score to a value that was less extreme after a careful consideration on the member’s response and the ratings of other evaluators. For example, in the instance that a comment received scores of 4, 4, and 5, the score attributed to that comment would be a 4.
4Please see Appendix D for the descriptive statistics associated with the independent variables.
5Non-Black minorities consists of Latinos, Asians and those of Middle-Eastern descent. Compared to Blacks, there were significantly fewer constituents in each of the non-Black minority categories. Thus, I grouped these constituents into one category.
6Messages that have no bearing on pending legislative matters may include, but are not limited to, complaints about city services, request for assistance with a community or personal issue, and criticisms of or compliments to local officials.
6More specifically, the total effect of shared race on Black members’ recall of messages is obtained by adding the interaction term coefficient to the Black Constituent coefficient. The overall value is positive.
7Since Model 2 captures an interaction effect, the coefficient associated with Other Minority actually captures the effect of Other Minority when Black Member is equal to 0. Hence it captures the conditional effect of being a non-Black minority when interacting with a White member.
8For additional information on the predicted probabilities associated with other race coefficients that attained significance at the 0.10 level, please refer the Appendix C.