Abstract
Research in public opinion and political communication indicates that those who hold viewpoints that are in the minority may be more hesitant to express their opinion when compared to those in the majority. Gauging hesitation through response latencies has been put forth as a measure of the internalization of majority pressure. In a laboratory setting, participants are asked to offer simple judgments (e.g., “like” or “dislike”) for various digitized images of both political and nonpolitical persons, things, and ideas. Responses are recorded and categorized according to majority or minority status and then analyzed at both the subject and the object level in an attempt to better understand the link between the climate of opinion and response hesitancy. Overall, those in the minority take longer to offer “like” or “dislike” responses when compared to those in the majority. This relationship is positively correlated with the size of the majority. Furthermore, individual differences are found to moderate the minority slowness effect (MSE). The effect is also found to be stronger for political objects when compared to nonpolitical objects.
In attempting to obtain a more comprehensive understanding of how public opinion is shaped and formed, the individual internalizations and reactions regarding this much-studied social force must be investigated. In other words, what goes into the mix when an individual offers a viewpoint to others? And how is this process affected by holding the majority—versus the minority—viewpoint?
This empirical tack seems particularly relevant, given the information-saturated age in which we live. We are certainly aware of our own preferences, but we are also cognizant of the views of the masses—both overtly through a litany of public opinion polls and more subtly in the form of mediated messages received via news outlets, such as print and electronic journalism, as well as social media websites and television shows. And let us not forget the interpersonal discussion of these messages. Noelle-Neumann put forth the idea that people possess a “quasi-statistical organ” (Noelle-Neumann, 1974) or a “quasi-statistical sense” (Noelle-Neumann, 1993), which she envisioned as a kind of social barometer in gauging public sentiment. This interpersonal awareness provides a basic perceptual reference point from which we can navigate our way through social interactions. Additionally, this method of assessing public opinion allows individuals to discover which issues are more controversial and may, therefore, cause conflict in future interactions.
The current study reports one method of measuring the degree to which the internalization of public opinion influences individuals’ responses when asked to offer simple evaluations of everyday objects. By using response latencies as a gauge of the hesitation involved in expressing an opinion, we highlight the pervasive effect that the quasi-statistical sense can have, even in a private laboratory setting. By replicating and extending Bassili’s (2003) “minority slowness effect” (MSE) studies, we offer a perspective on the basic processes involved in expressing one’s opinion.
Personal Opinion Development, Expression, and Self-Actualization
The origin of personal opinions has fascinated social scientists for centuries. Mead (1934) recognized the importance of the self as a socially constructed image. He described the individual as an aggregation of his or her perceptions of how others perceive him or her (i.e., the “looking glass self”). Simply put, who we are is in large part shaped by the assertions of those around us—and it is an iterative relationship. Our opinions, then, can be viewed as an important portion of this social self, which can vary according to the size and heterogeneity of our various social networks (Eveland & Hively, 2009; Mutz, 2002). When considering any type of expression, we mingle our own thoughts and desires with the normative considerations of our environment (e.g., Glynn & Huge, 2007). The merging of the two often generates questions as to how our own expressions will be received by others: Will the group respect or revile me if I share my true views on this issue?
Through expressing viewpoints—in Walter Lippmann’s words (1922, p. 3), merging the “world outside and the pictures in our heads”—some individuals feel more efficacious in governing their lives. On the other hand, some are more content to withhold their opinions, feeling that their world is mostly ruled by forces beyond their control (Ajzen, 2002) or choosing to avoid the potential interpersonal conflicts that can arise from political discussions (Mutz, 2002; Ulbig & Funk, 1999).
Conformity
At the individual level, the possibility of being seen as an outcast often has the power to limit—or halt altogether—the expression of an unpopular viewpoint (e.g., Neuwirth, Frederick, & Mayo, 2007). As Moscovici (1985) suggested, “When disagreement does arise, it is experienced as undesirable, as a tension and a threat” (p. 352). Indeed, the mere notion of disagreement is enough to make some people uncomfortable. If we are able to envision a future interpersonal exchange involving our own opinion colliding with the opinion of another, it may cause tension. In order to avoid such disagreements and maintain social bonds, we may seek to alter or conceal opinions that may differ from the majority view of a certain group.
Over the course of five studies, Bassili (2003) unearthed a pattern of expression inhibition whereby respondents were slower to offer minority opinions as compared to the response latencies of those who offered majority opinions. Put simply, those in the minority hesitated when expressing their views across various communicative contexts (e.g., phone interviews, responding privately to computer images as stimuli, or responding with the anticipation of having those responses revealed to others). Bassili also found a strong correlation associated with the discrepancy between the size of the majority and the minority, on the one hand, and the difference between the mean response latencies for minority and majority responses, on the other hand.
In other words, the larger the size of the majority, the longer it took respondents in the minority to express their viewpoint. He dubbed this the “minority slowness effect” (MSE). In one particular study, this effect was discovered even for judgments of everyday objects presented as images on a computer screen, such as apples and spiders. That is, those who disliked apples or liked spiders—both minority viewpoints—were measurably slower to offer their opinions when compared to those who liked apples and disliked spiders (i.e., the majority positions). As a means of explanation for this somewhat mysterious phenomenon, Bassili reasoned that “expression hesitancy is a byproduct of the conflict engendered by pressures to conform when one’s opinion does not match what one assumes to be the dominant position on an issue” (2003, p. 261).
The MSE persisted even when the most social elements involved in opinion expression were stripped away. Respondents offering judgments at individual computer stations still hesitated when offering unpopular viewpoints. As Bassili (2003) found, the lack of a social environment did not seem to completely erase the pressure that such a context would bring to bear. He reasoned that such hesitation resulted from internal conflict. Furthermore, this inner turmoil might be so deeply ingrained—and therefore below the level of consciousness—that it arises automatically, even in the absence of others. Our “quasi-statistical sense” (Noelle-Neumann, 1993) reminds us of the majority and minority status of opinions. In other words, it is possible that we cannot think of a minority viewpoint without also considering the conflict or tension that such a view might cause when expressed to others. The two cognitions seem powerfully intertwined.
When faced with the prospect of offering a minority judgment, we may feel torn in opposite directions. On the one hand, we would like to express our minority view, whether in an attempt to exert influence on our own environment or for the sheer sake of honesty. On the other hand, we may also weigh the consequences of expressing an unpopular viewpoint. Will we be looked down upon by others or perceived as strange?
When Personal and Public Opinion Intersect
It is interesting to note that while collective opinion is certainly learned and internalized by the members of the group, the context in which the debate occurred is also remembered. We are quick to associate past experiences and things we have learned in order to make sense of this new information. Zajonc (1980) reminds us that we are prone to internalize more than just the stimulus at hand—we often internalize the context in which that stimulus was observed. That context invariably contains social cues, emotion, and associated affective reactions.
Furthermore, individuals will not always have thoroughly contemplated opinions at the ready for expression. In this case, an individual might rely more heavily on social norms or cues in an attempt to appear reasonable and intelligent. Mutz and Soss (1997) refer to this recognition of the importance of an issue to others as “collective issue salience” (p. 434). As mentioned earlier, the desire to be correct and the desire for affiliation with others may pull us toward the majority view, even if we have contradictory thoughts of our own. Thus, the opinions put forward by the public are prone to much outside influence.
The main theoretical support for the MSE stretches back to Asch’s (1956) work on conformity, in which it was discovered that respondents were prone to go along with an obviously incorrect majority rather than expressing their own minority response when judging the distance of lines. By inducing respondents to offer these incorrect responses as an outcome of being surrounded by confederates who expressed a unanimous view, Asch demonstrated the power that conformity can have on an individual in a social setting.
In the realm of public opinion, Glynn suggested a “normative” approach to the opinion process. The key relationship here is between the strength of the perceptions others’ opinions and the strength of the individual’s actual opinion. In some cases, the perceptions of others’ opinions will dominate, and an individual will perceive the reference group’s opinion as being more important than his or her own opinion. In the case of the reference group’s opinion being relatively less important, an individual’s actual opinion is key. An individual’s “expressed opinion” becomes the result of “an interaction between what an individual actually thinks and what he or she believes is an appropriate and normative response for relevant social groups” (Glynn, 1997, p. 176).
Noelle-Neumann’s (1974, 1993) spiral of silence theory proffers the idea that the social environment, or “climate of opinion,” can have a marked effect on an individual’s willingness to express an opinion, or, on the other hand, to remain silent. She proposes that those in the minority are less willing to express a dissenting viewpoint or a viewpoint that is perceived to be losing ground. Social pressure can be especially powerful when there is a known majority position within a given group—whether the group is composed of family members, coworkers, or citizens of an entire nation. This pressure often causes those in the minority to feel a kind of segregation. Noelle-Neumann (1974, 1993) credits “fear of isolation” as a social force capable of silencing those in the minority. This loosely refers to an individual’s trepidation related to social removal from a group or other people in general. Fear of isolation is said to drive the spiral with greater force as the size of the majority increases.
However, as Bassili (2003) notes, Noelle-Neumann’s (1974, 1993) theory offers only a dichotomy of response options—speaking out or remaining silent. It can be assumed that there is much more to the process of opinion expression than a simple yes–no or agree–disagree relationship. There is most likely a continuum underlying this dichotomy (e.g., Hayes, 2007; Neuwirth et al., 2007). At one end are those who are eager to express minority opinions. At the other end are those who would do almost anything to avoid offering a less-than-popular viewpoint. The MSE may offer a more precise measure of response hesitancy in that it proposes a continuum rather than a dichotomy. In other words, the more social pressure an individual feels with regard to expressing an opinion, the more a measurable delay is likely to occur between the presentation of a stimulus object and the response. More specifically, Bassili posits a linear relationship between the majority size and response hesitancy for those in the minority.
In Bassili’s studies, as the size of the experimental majority increased (i.e., more respondents liked or disliked a given object), so did the discrepancy between the mean minority response time and the mean majority response time, with the minority responses taking longer. That is, if 90% of respondents liked a given object, those in the minority (i.e., the other 10%) were likely to hesitate more in their responses as compared to a case in which the majority–minority split was relatively smaller (i.e., a 60/40 split). Bassili labeled this relationship between the experimental size of the majority and the discrepancy between the mean majority and minority response times the “tell-tale correlation.” This correlation was typically of a magnitude between .50 and .60 (p. 274).
Individual Considerations
In the context of Bassili’s (2003) work, outside influences seem to remain part of the equation even in private contexts. This might be evidence of a kind of postactivation process, in which respondents became conflicted between a desire to express their actual opinion and a desire to remain with the majority, and therefore, hesitated as they mulled over this conflict.
Although it is apparent that these pseudosocial forces seem to have a strong impact on respondents in a private context, it must also be acknowledged that individuals bring very different personality traits to a given social environment. In arguing for a more individualized focus on members of the minority group in the spiral-of-silence context, Csikszentmihalyi (1991) stated that individual differences may mediate or moderate the effects of social pressure on various individuals and should be investigated as predictors of expression in order to obtain a better understanding of the process as a whole (Hayes, Glynn, & Shanahan, 2005a; Matthes, Morrison, & Schemer, 2010; Willnat, Lee, & Detenber, 2002).
Fortunately, there are personality measures available that tap into these individual differences in one’s awareness of the opinions of others and the amount of importance those opinions are given. Even if an individual has an opinion at the ready, he or she may decide to withhold that opinion. The Willingness to Self-Censor Scale (WTSC; Hayes et al., 2005b) measures respondents’ “active withholding of one’s true opinion from an audience perceived to disagree with that opinion” (p. 1). Self-censorers are said to be more anxious about being in a social setting and especially about communicating in such a context.
Similarly, certain people become apprehensive when faced with the possibility of being evaluated negatively by others. Those who are high in the Fear of Negative Evaluation (FNE) tend to adjust their behavior in an effort to avoid being looked down upon (Leary, 1983; Watson & Friend, 1969). In a minority context, it makes sense that high FNEs would be more likely to hesitate at the prospect of offering a minority response due to a fear that those in the majority (i.e., the larger number of people) would not approve. Public Self-Consciousness (PSC) refers to a process of self-focused attention defined by “a general awareness of the self as a social object that has an effect on others” (Fenigstein, Scheier, & Buss, 1975, p. 523). In the context of expressing an opinion, those high in PSC tend to be relatively more aware of their impact on others, whether the impact is aesthetic (e.g., “I’m usually aware of my appearance”) or more socially based (e.g., “I usually worry about making a good impression”). The awareness of the opinions of others is greater among those who score higher on the PSC Scale.
Snyder’s (1974) Self-Monitoring Scale is a measure of individual differences in “self-observation and self-control guided by situational cues to social appropriateness” (p. 526). In an opinion context, high self-monitors should be relatively more aware of the opinions of others, and more specifically, they should use this information to compare and contrast the views of others with their own. The Marlowe–Crowne Social Desirability Scale (SDS; Crowne & Marlowe, 1960) tests avoidance of disapproval (e.g., “I am always courteous, even with people who are disagreeable”). Those who seek to avoid the disapproval of others are more likely to engage in one of two behaviors when faced with the prospect of possessing a minority opinion. In an attempt to completely avoid social condemnation, some may offer a majority opinion to save face. On the other hand, volunteering a dissenting view is an option, though it is also an open invitation for others to disapprove of our views, and possibly—as an extension—to disapprove of us. Bearing this in mind, those who score high on the SDS may have to cut through these social fears before offering a minority response. This extra step translates into extra time and larger response latencies for the minority.
We chose these specific personality measures because they all capture the internal differences that may affect the way social pressure is internalized. Those who are more cognizant or fearful of negative reactions from others are more likely to consider the consequences of sharing opinions that could be unpopular. These personality measures allow for the possibility that individuals will vary in their internalizations and considerations of majority information. Those who most strongly fear the social repercussions of their public expressions may hesitate more when expressing a minority view as compared to those who don’t care about others’ perceptions of their views.
Research Questions and Hypotheses
The first aim of this study was to replicate one of Bassili’s (2003) reported MSE studies, in which respondents offered private judgments of objects presented to them as digitized images on a computer screen. Majority and minority groups were determined by counting the number of “like” and “dislike” responses for each object. If more “like” responses than “dislike” responses were recorded for a given object, then the majority view was “like” and vice versa. This will be referred to as the “object level” in conceptualization and analysis.
In Bassili’s (2003) work, those in the minority took longer to express their views. There was a positive relationship between the size of the majority and the average minority response latency. In calculating the discrepancy between the mean majority and mean minority response times at the object level (e.g., mean response time for those who “liked” apple pie as compared to the mean response time for those who “disliked” apple pie), we define the “minority slowness value” (MSV) as the difference between the two. A larger MSV indicates a greater group-level hesitancy when minority viewpoints are expressed (in relation to the average response latency for the majority viewpoint). To return to the apple pie example, drawing on the data collected for the current study, the mean majority response time (i.e., “like”) was 1.06 seconds, while the mean minority response time was 1.33 seconds. In this case, the MSV would be 0.27 seconds.
We use Bassili’s (2003) term, “minority slowness effect” (MSE) to refer more specifically to the correlation between the MSV and the discrepancy between the size of the majority group and the size of the minority group (see Figure 1). From this point forward, we will refer to this discrepancy between the size of the majority group and the size of the minority group as the “majority discrepancy.” For example, in the present study, 198 respondents were asked to offer a judgment (“like” or “dislike”) for apple pie. A total of 177 respondents chose the “like” response, while 21 chose the “dislike” response. In this case, the majority discrepancy would be 156 (i.e., 177 – 21). Once the MSV and majority discrepancy are calculated for each object, these numbers can be used in a correlation analysis that will produce an object-level MSE.
Prediction of Response Latencies Relative to Size of Majority (Minority Slowness Value).
Although it may at first seem strange that individuals would hesitate when expressing minority viewpoints to a computer (with little possibility of having their answers revealed to others beyond the investigators), Bassili’s research (2003) suggests that even in the absence of a social context, response latencies seem to correlate with social considerations (i.e., relative size of the majority and minority).
Hypothesis (H1): At the object level, the response latencies for minority viewpoints will be greater than for majority viewpoints.
Hypothesis (H2): The MSV will be positively correlated with size of the majority discrepancy.
Prediction of Minority Slowness Effect.
If response hesitancy is primarily driven by a “quasi-statistical” awareness of the relative size of the majority and the minority, then individual respondents will most likely display similar hesitancy for expressing a known minority opinion. On the other hand, it is possible that the MSE is—at least in part—an outcome measure of some antecedent personality difference, and such a finding would alter the interpretation of causality. Instead of positing that minority response latencies will increase in a relatively similar manner across all respondents in accordance with the majority discrepancy, the prospect that individual characteristics are moderating the relationship between minority status and response latencies is an intriguing alternative explanation.
In this study the “individual level” refers to the effects of minority status for each individual respondent, with the number of majority and minority responses and mean response latencies for those majority and minority responses used as independent variables. To calculate a value similar to the discrepancy between the majority and minority size at the object level, a variable was created (i.e., the majority discrepancy at the individual level) to quantify the difference between the number of majority responses and the number of minority responses offered by each respondent. In other words, the number of minority responses for each individual was subtracted from the number of majority responses.
Research Question 1 (RQ1): Will the MSE hold at the individual level as it does at the object level?
Research Question 2 (RQ1): How will differences in personality traits moderate the MSE at the individual level?
It is intriguing to see the MSE operating for responses to everyday, mundane objects. Indeed, most of the objects used in Bassili’s (2003) study carried relatively little social baggage. Respondents were probably not deeply concerned with stimuli such as puppies and ice cream. Still, if the MSE is to gain further merit, it should also operate when more vital issues are used as stimuli. In the interest of extending this line of research, objects that most would consider to be relatively more controversial (e.g., politicians, budget deficits) were added to the pool of stimuli. Since these “political” objects symbolize the possibility of more conflict when considering a future interpersonal interaction, they will most likely cause respondents to take more time to think about their answers. Although this will slow down overall response latencies for both minority and majority responses—and therefore, introduce more statistical noise—examining the effect when more socially significant objects (henceforth referred to as “political” as opposed to “nonpolitical” objects) are placed before respondents is essential.
If normative considerations are salient, respondents will likely weigh the social relevance of their personal responses against their environment (i.e., the climate of opinion). It is also likely that objects in the political domain are more subject to constant social construction and thus cognitions would be more diffuse, requiring greater retrieval time. In terms of the study at hand, such additional considerations will likely lead to a greater response latency. 1 Therefore, it seems respondents will be more likely to hesitate when faced with the prospect of offering a minority response to a political object (as opposed to those offering a majority viewpoint). Furthermore, it is probable that the discrepancy between the majority and minority response times (MSV) for these political objects—and therefore the MSE—will be greater than for nonpolitical objects (see Figure 3).
Minority Slowness Effect: Political Versus Nonpolitical Objects.
Hypothesis (H3): The MSE will be greater for political objects when compared to nonpolitical objects.
Range, Mean, Standard Deviation, and Reliability for Personality Scales Used in Analysis.
Note: N = 184. The low reliability score for self-monitoring indicates that tests using this measure should be viewed accordingly.
By replicating Bassili’s (2003) results at the object level, examining the role of personality measures as predictors of the MSE and extending the study to include political as well as nonpolitical objects, we aim to uncover a better understanding of some of the most basic processes involved in opinion expression. Specifically, what type of person—in terms of personality attributes—displays the MSE to the largest degree? And how does this effect operate when political objects are used as stimuli?
Method
Students from introductory and upper-level undergraduate classes were recruited for this study. A total of 198 students completed the computer-based trial, and a total of 184 students participated in both the written personality inventory and the computer-based portions of the study (Nfemale = 106, Nmale = 78; Mage = 20.96 years, SD = 2.34). As compensation for their participation, students were awarded extra credit in their courses. In order to maintain a sufficient level of respondent anonymity, participants were asked to provide only the last five digits of their social security numbers as means of identification for both the written portion of the study and the computer-based section. These identification numbers were then used to ensure that participants received extra credit for their participation and for matching of survey responses with the results of the computer-based trial.
Participants were first asked to fill out a pretest of survey items, including the previously discussed personality measures. These scales were reverse scored where appropriate and averaged to arrive at a single measure for each of the sampled personality attributes. This made it possible to include each of these scales as a variable that could be used in analysis of individuals’ displayed MSE.
For the second portion of the study, participants entered a computer lab and sat at separated machines. They were asked to offer judgments of the objects that appeared on the computer screen. In order to obtain relatively consistent results, the participants were asked to respond “as quickly and accurately as possible.” After receiving both verbal and written instructions, respondents viewed 75 different digitally scanned cards drawn from a board game called Personal Preference. The computer-based trial was conducted using SuperLab Pro software. SuperLab Pro allows for random presentation of experimental stimuli, records participants’ responses to each object included in the trial, and records the response latency for each answer (i.e., the amount of time elapsed between the presentation of the image on the screen and the participant’s choice of the “like” or “dislike”). Using a standard computer keyboard, the respondents entered “1” if they liked the object in question and “0” if they disliked the object. As soon as a judgment was offered and recorded, a new object appeared on the screen. The mean response time across all subjects and objects was 1.52 seconds (N = 15,325, SD = 0.39 seconds).
Measures
The creation of experimental categories (majority and minority) was based on the total number of majority and minority responses for each object. In other words, if an individual expressed “dislike” for puppies, he or she was placed in the minority category for that object, as the majority of respondents indicated that they liked puppies. Because the intent of this study was to investigate the MSE at both the object (i.e., the various pictures respondents were shown and asked to respond to) and at the individual level, these experimental groups were then extrapolated to two different files.
For the object-level file, the majority and minority were determined by the responses from participants. The group sizes were then calculated and compared to create the majority discrepancy value. The majority discrepancy was used in conjunction with the MSV in calculating the MSE (i.e., the correlation between the majority discrepancy and the MSV). For example, in offering judgments of spiders (one of the stimulus objects included in the study; see Table 2 for a complete listing of objects and response latencies), the mean majority response time was 1.13 seconds, and the mean minority response time was 1.21 seconds. Therefore, the MSV was 0.08 seconds. This MSV for spiders was then used in conjunction with the majority discrepancy for spiders in calculating the overall, object-level MSE. The mean MSV for all stimuli at the object level was 0.21 seconds (n = 75, SD = 0.33). In other words, at the object level, minority responses were about two tenths of a second slower than majority responses.
Response Latencies for Majority and Minority Responses and Comparisons.
Note: N = 198. In conducting independent samples t tests, equal variances were not assumed.
p < .10. **p .05.
Pearson Correlations Between Majority Discrepancy (MD), Minority Slowness Value (MSV), and Response Times.
Note: N = 75.
p < .05. **p < .01.
For the subject-level file, each respondent was used as a row in the data file. Since each respondent’s answers were recorded and coded for majority/minority status, a majority discrepancy—similar, though not identical to the object-level majority discrepancy—was calculated for each individual. By subtracting the number of minority responses from majority responses, an index of each respondent’s frequency of majority responses was created. This majority discrepancy was calculated for each respondent, and that value was used in correlation with each respondent’s individual MSV (i.e., the mean majority response time for each participant subtracted from the mean minority response time for each participant) to calculate each respondent’s MSE. At the individual level, the calculated MSV was 0.27 seconds (n = 198, SD = 0.34), meaning that individuals’ minority responses were almost three tenths of a second slower than their majority responses. The personality measures were also included for each participant in the subject file. Separate “political” and “nonpolitical” MSVs were also calculated. These numbers separated respondents’ majority and minority judgments into two additional categories, creating a 2 X 2 matrix of response options.
Results
The first goal of the study was to replicate the MSE as reported by Bassili (2003). Bassili’s analysis consisted of comparing majority and minority response latencies with t tests at the object level and calculating the correlation between the majority discrepancy and the MSV. In replicating the t tests, 32 of the 75 objects used in the computer trial yielded significantly larger response latencies for those who answered with a minority response at the p < .05 level, with an additional six objects exhibiting the same directional difference at a slightly less significant level (p < .10). In contrast, only six objects displayed significantly faster minority response times (p < .05). A paired-samples t test was used to compare the mean minority response time with the mean majority response time at the object level. In support of the first hypothesis, minority responses at the object level were significantly slower than majority responses. The mean difference was 0.21 seconds, SD = 0.33, t(74) = 5.48, p < .01.
In support of H2, the correlation between the majority discrepancy and the MSV was found to be significant (r = .249, p < .05, two tailed). That is, there was a significant relationship between the majority discrepancy (i.e., relative to group sizes) and the MSV (calculated as the difference between the mean minority and majority response times), though the relationship was a bit weaker than Bassili’s (2003) findings (r = .52, p < .001, two tailed). Further analysis yielded the strongest correlation between the majority discrepancy and the mean majority response time (r = –.405, p < .001, two tailed). This negative relationship is logical—in that the larger the majority, the faster those in the majority were to respond (i.e., lesser response latencies)—and indirectly supports the overall existence of the MSE. With respect to the majority discrepancy and the mean minority response time, the relationship was not found to be significant (r = –.120, p = .30, two tailed). The negative relationship between the majority discrepancy and the minority response time was quite surprising, as it is in direct opposition to Bassili’s reported findings. Overall, the effect held. As the size of the majority increased, majority responses were found to be relatively faster.
Individual Analysis
The data were also analyzed at the individual level, swapping the respondents for objects in the data file in order to assess the relationship between the individual-level MSV and the individual-level majority discrepancy. In order to answer the first research question, which deals with the individual-level existence of the MSE, a correlation analysis of the MSE at the individual level was conducted. The correlations at the individual level were similar to those at the object level. To calculate the MSV at the individual level, the mean majority response time was subtracted from the mean minority response time for each respondent. In total, 171 out of 198 respondents had a positive MSV, which means that only 27 respondents had faster mean minority response times as compared to their mean majority response times. The mean difference between the minority response times and the majority response times was M = 0.26 seconds, SD = 0.34. In other words, the average respondent took almost three tenths of a second longer in offering minority answers as opposed to his or her own majority answers across all objects. The overall MSV at the individual level was found to be significantly different from zero, t(197) = 10.90, p < .001.
As previously mentioned, an individual-level majority discrepancy value—similar to the object-level majority discrepancy value—was created. This individual-level measure captures the difference between the number of majority responses and the number of minority responses offered by each respondent. In other words, the number of minority responses for each individual was subtracted from the number of majority responses. This is conceptually similar to the discrepancy between the majority and minority size at the object level, though it is not exactly the same measure, as it pertains to each individual rather than each object. As seen in Table 4, this individual-level majority discrepancy was also significantly correlated with the individual-level MSV, though the relationship was not strong (r = .212, p < .01). While this is slightly weaker than the object-level correlation, it does indicate a tendency by those who offer more majority responses to offer slower minority responses than those who offer more minority responses. Here we begin to see a trend of reticence in which those who are less prone to offer minority responses are more likely to hesitate when they do offer a minority response. Whether this arises from an underlying personality difference will be explored in further analyses.
Pearson Correlations Between Majority Discrepancy (MD) and Minority Slowness Value (MSV) at the Individual Level.
Note: N = 198.
p < .05. **p < .01.
Personality Measures as Moderators
Because this effect still seemed a bit mysterious, a more meticulous method of analysis seemed necessary. It is a given that certain attributes vary from person to person, so it makes sense that respondents would differ in their display of the MSE. Why are certain individuals more susceptible to this effect? In a regression context, the individual personality variables yielded the results displayed in Table 5. Of the five independent variables tested, two were significant at the p < .05 level (WTSC and FNE) and two were found to be marginally significant (p < .10) predictors of the MSV across all participants (SDS and PSC). Results reveal that those who are more likely to actively withhold personal opinions (i.e., WTSC) and those who are more fearful of being viewed negatively (i.e., FNE) are more likely to measurably hesitate when offering minority viewpoints. Those who are more concerned with social acceptance (i.e., SD and PSC) are also more likely to hesitate, though this relationship appears to hold only marginal significance. All of the observed predictors were positively related to the individual-level MSV (i.e., the mean minority response time minus the mean majority response time). 2
Linear Regression Weights for Personality Predictors of Minority Slowness Value (MSV).
Note: N = 184.
Political Versus Nonpolitical Objects
Because some topics carry more social importance than others, it seems likely that reactions to the more important or “political” objects (e.g., tax cuts, the NRA [Nonresident Alien]) would differ from reactions to everyday or “nonpolitical” objects (e.g., playing golf, wide ties). There were 47 nonpolitical objects and 28 political objects in the study. 3 For the nonpolitical objects, the mean minority response time was 1.49 seconds (SD = 0.29 seconds) while the mean majority response time was 1.32 seconds (SD = 0.23 seconds). For the political objects, the mean minority response time was 2.00 seconds (SD = 0.49 seconds), while the mean majority response time was 1.74 seconds (SD = 0.30 seconds).
It was predicted in H3 that the MSE would be stronger for the political objects. The “split file” command was used in SPSS to separately investigate the correlations between the object-level majority discrepancy and the object-level MSV for both political and nonpolitical objects. As predicted, the MSE was stronger for the political objects (r = .422, p = .025) than for the nonpolitical objects (r = .253, p = .086), though the relationship between the majority discrepancy and the MSV for nonpolitical objects was only marginally significant. 4 The difference between these correlations did not reach the level of significance (z = 0.76, p = .22), though this is almost certainly due to the very small sample sizes of these correlations (n = 47, for nonpolitical objects; n = 28, for political objects.)
Mean Response Times (RT) for Majority and Minority Responses to Political and Nonpolitical Objects.
Note: N = 13,800. Response times are in seconds>.
Object-Level Pearson Correlations Between Majority Discrepancy and Minority Slowness Value (MSV) for Nonpolitical and Political Objects.
Note: N = 75.
p < .10. **p < .05.
Conclusions
The results of this research suggest that the MSE is driven by many social forces, including personality traits, the size of the majority and the minority, and whether the object in question is “political” or “nonpolitical.” The overall effect—as put forth by Bassili (2003)—held, and a more detailed explanation of how and why the effect operates was uncovered. It was also revealed that those who score higher on measures of social self-consciousness (e.g., FNE, SDS, etc.) are relatively more likely to display the MSE at the individual level.
Such results echo recent findings regarding the variability among individuals in terms of their susceptibility to social forces (e.g., Matthes et al., 2010). Those most likely to hesitate when expressing a minority viewpoint are the same individuals who fear negative evaluation at the hands of others. They are also more likely to self-censor their own opinions. People who most strongly seek the approval of others and worry the most about their public profile are also more likely to delay their responses in the face of majority pressure.
While a more detailed understanding of the overall effect was unearthed, the basic operation of the MSE is far from completely understood. As the setting for this experiment was somewhat artificial, it is possible that these recorded judgments—and their corresponding latencies—are not truly reflective of opinion expression in the real world. Furthermore, respondents in this study were forced to offer a judgment and allowed only two simple alternatives. It stands to reason that some respondents may have offered opinions only to complete the task, with little connection existing between the recorded answer and their actual opinion. Recent research has also revealed the importance of having a “no response option” as an important and realistic choice (e.g., Hayes, 2007), though the inclusion of such an option would certainly complicate the handling of response latencies.
It may seem a bit counterintuitive that response latencies should function as a result of perceptions of majority/minority status in such a private setting as described previously. However, when considering that most of our attitudes and beliefs are based—at least in part—on social exchanges with others (i.e., we do not have firsthand knowledge of most things), it makes sense that the social norms associated with various objects will be present to some degree when we are presented with various stimuli. This seems to be the case even when the objects in question are nonpolitical and carry little social relevance.
The existence of the MSE presents the possibility that our means of expression are deeply socialized. They may be so deeply ingrained that, in fact, we may bring social considerations to bear even when there is no social environment to consider. This research creates questions about the basic-level processes underlying information consumption, intrapersonal communication, and especially opinion expression. Is it possible to completely leave our social worries behind when there is no social environment? Maybe some objects are so tangled in controversy that we cannot think of the object itself without also dragging out the debate.
Work in psychology (e.g., Bargh & Chartrand, 1999) and neuroscience (e.g., Craig, 2002) certainly hints at the possibility that many of our actions and reactions are nearly automatic. Presented with a given stimuli, our brain floods with associations of the present and past. More specifically, Bargh and Chartrand have argued that our brains are set up for “continual automatic evaluation of one’s experience,” yet we often have little conscious awareness of this evaluative process (p. 462). Craig’s work in neuroscience has shown that much of human behavior might be explained by how the brain is wired. Different areas of the brain are said to “light up” in different situations, especially when we are faced with physical pain or emotional stress.
With this “automaticity” in mind, it seems logical that response latencies would show signs of social influence even when the social context is not present. People typically do not wish to be perceived as followers or blind sheep. However, it seems that much of what we say and do is based on processes that are less than deliberative. It makes sense then, when considering that much of our input (i.e., listening, learning, making sense of the world) occurs in social situations, to acknowledge that much of our subsequent output (i.e., reacting, expressing opinions, hesitating to do so) has a strong social element to it, even if we are not as cognizant of such an element.
Zajonc (1980; Zajonc & Markus, 1982) wrote extensively on the overall nature of thought in terms of how ideas enter the mind (e.g., reacting to various stimuli, automatically, or in a controlled manner) and how they interact with emotions and with thoughts already held. According to Zajonc (1980), stimuli may elicit a nearly instant affective response in individuals. As issues or ideas are given further consideration, the initial affective response may diminish or be reversed, but it is still the peg on which the initial considerations are hung. However, within the context of the climate of opinion, it may be true that this affective response cannot be separated from consensus information related to an issue or object. While preferences may need no inferences, it may also be the case that our preferences are inextricably tied—either in conflict or consistency—to the norms associated with the object at hand.
This trend toward an “automatic” view of expression, coupled with the power of a possibly mainstreaming media, can lead to some unnerving questions. How many of the ideas in our head are truly our own? It seems possible that majority/minority information is simply passed along with little regard given to its origins. As we seek to maintain balance in our social bonds and in our own heads, we must constantly compromise our own ideas to better fit with the expressions of others. But if so many of our expressions are normatively shaped, is collective opinion then only a jumble of conformist declarations?
Future researchers would do well to experimentally manipulate the majority/minority information available to the respondents. In other words, an image—along with information about the percentage of others who liked and disliked that image—could be presented to respondents. In this way, there would be no question in the respondent’s mind about his or her majority or minority status. The percentages could be altered to show strong majorities (90% vs. 10%) slim majorities (55% vs. 45%), dead heats (50% vs. 50%), and values that could easily be construed as relatively even (51% vs. 49%). In this way, the effect that the presence of majority/minority information has on subjects in terms of their response latencies could be better understood.
Another interesting extension of this research would be to use a “think aloud” protocol in which the researcher asks the respondents to vocalize as much of their thought process as possible while they are offering judgments for various objects. With this method, the content of the thought process could be exposed, and the prevalence of various types of utterances (e.g., normative information, affective responses) could be recorded.
Obtaining a better understanding of the most basic elements of public opinion is a goal that can and should daunt researchers for the foreseeable future. While the overall processes involved in opinion reception, contemplation, and expression are quite complex, a better understanding of these mechanisms will aid in the continuation of a productive and beneficial field of research.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.