Bad News and Quality Reputation among Users of Public Services

About 30% of employees around the world work in public organizations (Hammouya, 1999), and employment tends to increase over time in public organizations dedicated to very relevant services, such as health, education, and social services (Derlien & Peters, 2008). In addition, for decades public organizations have contributed to strengthening economic growth and social welfare in modern societies (e.g., promoting great scientific advances). However, the reputation of public organizations among citizens is not very positive. Mizrahi, Vigoda-Gadot, and Van Ryzin (2010) observed that the image of public organizations in the United States is relatively low. They considered indicators such as the image-reputation of public organizations, quality of public personnel, and trust. Van de Walle (2007) analyzed confidence in public services in 60 countries around the world, using the World Values Study data, and more than 50% of the participants expressed low confidence (not at all or not very much). Given the importance of public services in our societies, these results are somewhat disappointing. Objective quality indicators tend to be very similar when specific public and private companies are compared (Marvel, 2015). Therefore, an important factor to consider seems to be the information citizens receive about the quality reputation of public organizations, and how it is managed by public managers and policy makers.

Organizational reputation is usually defined as the general perception or estimation of an organization across stakeholders or constituents and over time (Wæraas & Byrkjeflot, 2012). One critical challenge to reputation in public services is the quality citizens perceive. The public organization is usually seen as “the ugly sister” (Czarniawska, 1985; Wæraas, 2014) that (compared to private organizations) is rigid, slow, and inefficient, among other pejorative adjectives (Goodsell, 2004). This bias against public organizations is observed even when objective indicators of service quality are similar to those observed in private organizations (Marvel, 2015). This bias is probably related to the difficulties in establishing shared service quality criteria in the public sector (Walsh, 1991). Consequently, bad news seems to be easily transferred to citizens’ opinions and magnified. Wæraas and Byrkjeflot (2012) indicated the existence of five problems in managing reputation in public organizations. These problems can partly explain the pervasiveness of bad news about public organizations in our societies. First, public organizations are connected to political authorities who decide what policies should be implemented. During election campaigns, politicians may be tempted to criticize public services to promote their candidacies or political parties (e.g., the politician makes a commitment to improving efficiency and customer-oriented services). Second, public organizations are usually inconsistent in terms of values, identities, and self-representations (Fuglsang & Rolf Rønning, 2015). Public administrators have to deal with values that are difficult to reconcile (e.g., market orientation and equality), assuming a large variety of identities and audiences that require differential values and complex tasks. This lack of a strong identity can have negative effects on citizens’ perceptions. Third, it is difficult for public organizations to establish emotional bonds with citizens beyond functional services. Although bureaucracies are very helpful in delivering necessary services, they are not the adequate context for the positive emotional relationship that is increasingly expected by citizens. Fourth, public organizations cannot create the uniqueness (the degree to which the organization is distinctively different from rivals) that characterizes successful organizations. Public organizations are part of a larger category (e.g., the national health system), and citizens probably perceive that they are very similar to each other. Finally, achieving and maintaining excellence is a critical obstacle for public organizations because they usually have to make unpopular decisions based on policies prescribed by a superordinate level.

These five problems, described by Wæraas and Byrkjeflot (2012), offer a rich picture that makes it possible to understand the bad reputation of public organizations in terms of service quality and the frequency with which bad news appears in the media. As Grint (2010) pointed out, the appearance of a public organization in the media is often related to a problem or uncomfortable event. By contrast, good news seems less frequent and less interesting to the media. The present research study proposes that the predominance of bad news can be quite relevant in terms of reputation because it has a strong influence on citizens’ perceptions. According to the negativity bias, humans are very sensitive to bad events and news (e.g., Rozin & Royzman, 2001). Disseminating negative information about a public service, even though it is false, may cause a bias in citizens’ assessment, deteriorating service quality perceptions among users of the service. This bias, however, is not as influential in the case of positive information. To test this proposal, Experiment 1 was conducted to examine the differential effects of positive vs. negative initial information on citizens’ quality perceptions of a public service. In addition, Experiment 2 explores the effects of formal (report) vs. informal (rumor) communication on quality perceptions after receiving positive vs. negative initial information.

These two research studies contribute knowledge in at least two ways. First, the investigation of negative information is especially relevant for public organizations. As mentioned above, the public organization tends to be associated with a poor reputation even when the objective quality is the same as what can be observed in private organizations (Marvel, 2015). We test whether negative information plays a role in understanding differences in service quality perceptions of citizens who use the same public service. In other words, although they use the same public service, it is possible that receiving negative information, even if it is false, biases the evaluation of service quality. Second, we examine the degree to which the credibility of the information helps to clarify under what circumstances the negative bias exists. It is reasonable to expect that the influence of negative information only exists when citizens are subjected to credible and apparently well-founded information (formal). By contrast, this bias could be reduced when citizens receive the information through informal means (rumors). This is relevant not only because it makes it possible to advance the knowledge about the conditions for negative bias against public organizations, but also because it is relevant for communication management by managers and policy makers.

Experiment 1

The anchoring and adjustment heuristic refers to estimations people make based on an initial value or information that is adjusted to give a final answer (Epley & Gilovich, 2001). Therefore, different starting points lead to different responses that are biased toward the initial information. For example, the classic experiment by Tversky and Kahneman (1974) described a situation where participants made estimations about the percentage of African countries in the United Nations. Different groups of participants received different arbitrary numbers as starting information. The median of percentage estimates of African countries was 25 and 45 for participants who received 10 and 65, respectively, as arbitrary starting points (Tversky & Kahneman, 1974). This anchoring phenomenon has been demonstrated in numerous contexts, including self-efficacy (Cervone & Peake, 1986), future performance (Switzer & Sniezek, 1991), and negotiation (Schaerer, Swaab, & Galinsky 2015).

The present research study extends the investigation of anchoring to citizens’ quality evaluations of public services and the asymmetry between positive vs. negative information. Taking the negativity bias into account, it has been proposed that the effects of the anchoring and adjustment heuristic should be more accentuated for negative news associated with public services than for positive news. The negativity bias refers to the strong influence of bad events on human behavior. In their very exhaustive reviews, Baumeister, Bratslavsky, Finkenauer, and Vohs (2001) and Rozin and Royzman (2001) concluded that bad events are stronger than good ones in understanding our behavior in different contexts, producing greater, more consistent, or more lasting effects on a number of psychological phenomena (emotions, learning, etc.). Bad events include any harmful or unpleasant event, whereas good events represent any beneficial or pleasant outcome. Baumeister et al. (2001) proposed that this negativity bias is an adaptive mechanism developed during our evolutionary history that has facilitated the survival and success of humans: “A person who ignores the possibility of a positive outcome may later experience significant regret, but nothing directly terrible is likely to result. In contrast, a person who ignores danger (the possibility of a bad outcome) even once may end up maimed or dead” (Baumeister et al., 2001, p. 325). This special sensitivity to bad events and information has been explored in different contexts, including the specific service quality research area. Ofir and Simonson (2001) found that, unless users have low expectations, during consumption they focus predominantly on negative facets of service quality. Choi and Mattila (2008) also concluded that users react particularly badly when they believe that the service organization could have easily prevented a service failure.

Combining this negativity bias with the anchoring phenomenon in public services, it is reasonable to expect that the bias toward the initial information about the public service (anchor) would be more accentuated for negative news than for positive news. In other words, when users of a public service receive bad news about the service in question, their reactions in terms of service quality perceptions are especially negative. However, the magnitude of the reaction is not symmetrical, as far as positive reactions are concerned, when users receive positive information about the public service. Therefore, we propose the following hypothesis:

H1: The effect of anchoring on users’ quality evaluations of public services will be greater for initial negative information than for positive information.

Method

Participants. A total of 105 passengers of a commuter public train transportation organization located in Spain (52.4% were women) participated in this experiment. Ages ranged from 18 to 55 years, with a mean of 30.83 (SD = 9.85). One of the researchers asked for their collaboration in the surroundings of one of the train stations. To be eligible for the experiment, the passenger had to frequently use the train line going to this station (at least once a month). Participation was voluntary and anonymous.

Figure 1

Stimulus (Graph) for Participants in the First Experimental Condition. Experiment 1.Note. High scores (range from 0 to 100) are indicative of better quality.

Procedure and materials. Passengers were randomly distributed into three conditions. Participants assigned to the control condition (N = 35) answered a short questionnaire evaluating the quality of the commuter public train service they had used to arrive at the station where they were recruited for the study. Passengers assigned to the first experimental condition (N = 35) observed, first, a graph showing the results of a fictitious study (see Figure 1). The graph contained a comparison of the levels of quality of commuter public train services in different European countries. The commuter train services located in Spain had the best score. After the participants had observed this graph, they answered the same questionnaire about the quality of the public commuter train service that they had used to arrive at the station where they were recruited for the study. After answering the questionnaire, the researcher informed participants that the data in the graph were false. Finally, participants assigned to the second experimental condition (N = 35) followed the same procedure as those in the first experimental condition. However, the graph showed the results of a fictitious study assigning the worst score to the public commuter train services located in Spain (see Figure 2).

Figure 2

Stimulus (Graph) for Participants in the Second Experimental Condition. Experiment 1.Note. High scores (range from 0 to 100) are indicative of better quality.

The same measure of quality was used for all participants. A general assessment was obtained by asking: “Overall, what is the level of quality you perceive in the commuter service train that you have used?” Following this stem, there were three 7-point scales ranging from very low to very high, from awful to excellent, and from very poor to very good (alpha coefficient = .96). Similar measures can be found in the service quality literature (e.g., Spreng & MacKoy, 1996). High scores were indicative of higher quality.

Manipulation check. To assess whether the manipulation worked, we compared the two graphs (see Figures 1 and 2) to a convenience sample of 20 participants. Ages ranged from 26 to 59 years, with a mean of 40.16 (SD = 10.43), and 75% of them were women. Participants evaluated the service quality of Spanish commuter trains using a scale ranging from 1 (strongly disagree) to 7 (strongly agree), and considering the comparison with the other countries. More specifically, all of them evaluated the degree to which the quality of commuter trains is better in Spain than in the other European countries, based on the information shown in the two graphs. The perception of service quality based on the first graph (Figure 1) (M = 6.70, SD = 0.80) is much better, T₍₁₉₎ = 14.88, p < .01, than the evaluation based on the second graph (Figure 2) (M = 1.25, SD = 0.91). All participants were also asked to indicate in which of the two graphs the Spanish commuter trains obtained the best service quality results and in which of the two graphs they obtained the worst results. All of them (100%) considered that the first graph (Figure 1) described a research study where Spanish commuter trains obtained the best service quality results. They also agreed (100%) that the second graph described a research study where the Spanish commuter trains obtained the worst service quality results. In general, these results confirmed that graphs 1 and 2 were able to show positive vs. negative evaluations of the service quality of Spanish commuter trains, respectively.

Results

Descriptive results for the three conditions were the following: control condition (M = 4.47, SD = 0.95), first experimental condition (M = 4.57, SD = 1.20), and second experimental condition (M = 3.66, SD = 1.03). The analysis of variance (ANOVA) showed the existence of significant differences among the groups participating in this experiment, F_{(2, 102)} = 7.75, p < .01. Bonferroni post-hoc tests confirmed H1. Service quality perceptions of participants in the second experimental condition were statistically lower than in both the control condition (p < .01) and the first experimental condition (p < .01). By contrast, there were no significant differences between the control condition and the first experimental condition (p > .05). As expected, the anchoring effect was observed when users received initial negative information about the public commuter train service (second experimental condition), with their service quality perceptions declining significantly (see Figure 3).

Figure 3

Means for the Conditions in Experiment 1.

Experiment 2

Citizens can receive information about public services in a formal way. One example is the external evaluation of quality represented in the graphs used in the first experiment. However, citizens can also receive information through informal communications. The most popular informal form of communication is probably the rumor. Rumor is defined as an “unverified proposition for belief that bears topical relevance for persons actively involved in its dissemination” (Rosnow & Kimmel, 2000, p. 122). Unlike news and formal reports, rumors communicate information that has not been verified (Bordia, DiFonzo, Haines, & Chaseling, 2005). Taking into account the basic laws of rumors, they tend to multiply when the information is important for the people involved and when there is ambiguity about the issue (Allport & Postman, 1947). These two conditions can easily arise in relation to public services. The public sector often delivers services that are very important for the society and citizens (e.g., education, health). In addition, and given the multiple and complex stakeholders of public organizations, administrators usually project ambiguity toward their audiences (Carpenter & Krause, 2012).

As mentioned above, the initial information (anchor) citizens receive acts as a starting point, impacting their subsequent estimations of the quality of public services. The goal of the second experiment is to examine this anchoring bias for formal communication (e.g., report from an external quality evaluation) vs. rumors. It is reasonable to expect that the anchoring phenomenon would be greater for formal communication than for rumors. Because rumors are improvised and not verified, they should be less credible than formal communication. Considering attribution theory (Settle & Golden, 1974), information from formal communication should describe – from the perspective of the citizen – the actual characteristics of the public service, accurately increasing the credibility of the information provided, whereas rumors are more easily attributed to the opinions/interests of the people involved. Previous studies agree with this argument. In the stock market context, DiFonzo and Bordia (1997) found that news in a newspaper format was significantly more credible than published rumors, and published rumors in turn were more credible than unpublished rumors. In the investigation of traveling information, Chung, Lee, and Han (2015) reported that the magnitude of the path coefficient (in structural equation modeling) from formal communication to credibility doubles the one linking informal communication to credibility. Despite their lower credibility, in some specific contexts rumors can play a relevant role in influencing behavior and decision-making. This is the case of stock market dynamics (da Cruz & Gomes, 2013; DiFonzo & Bordia, 1997), where the need to make quick decisions about buying and selling cannot wait for confirmed news, and rumors make it possible to make sense of uncertain situations. Nevertheless, in most service contexts, where this time pressure is not present, the impact of formal communication should be greater than that of rumors. Therefore, the following hypothesis is proposed for empirical testing:

H2: The effect of anchoring on users’ quality evaluations of public services will be greater for formal communication than for rumors.

Method

Participants. A total of 172 users of an administrative service of a public university located in Spain (69.8% were men) participated in the second experiment. Ages ranged from 18 to 29 years, with a mean of 22.51 (SD = 2.53). One of the researchers asked for their participation in the university facilities. To be eligible for the experiment, the user had to use the administrative service frequently (at least once a month). Participation was voluntary and anonymous.

Procedure and materials. Users were randomly distributed into four conditions. Participants assigned to the experimental condition 1 (N = 47) were subjected to an anchor with positive formal communication. First, they observed a graph showing the results of a fictitious study (see Figure 4). On this graph, there was a comparison of the levels of quality of the administrative services in different public universities located in Spain. The focal administrative services (used by participants) had the best score. After participants had observed this graph, they answered a questionnaire about the quality of the administrative services of the focal university, considering their own experience. After answering the questionnaire, the researcher informed participants that the data on the graph were false.

Figure 4

Stimulus (Graph) for Participants in the Experimental Condition 1. Experiment 2.Note. High scores (range from 0 to 100) are indicative of better quality.

Participants assigned to the experimental condition 2 (N = 38) were subjected to an anchor with a positive rumor. First, users read a written fictitious rumor: “For weeks, rumors have circulated insistently about the excellent quality of administrative services at the University. According to these rumors, students and teachers have indicated that these administrative services provide very good service and show great efficiency in service delivery”. Once participants had read the rumor, they answered a questionnaire about the quality of the administrative services of the focal university, considering their own experience. After answering the questionnaire, the researcher informed participants that the rumor was false.

Participants assigned to the experimental condition 3 (N = 39) were subjected to an anchor with negative formal communication. They followed the same procedure as in experimental condition 1. However, the graph showed the results of a fictitious study where the administrative services at the focal university had the worst score (see Figure 5).

Figure 5

Stimulus (Graph) for Participants in the Experimental Condition 3. Experiment 2.Note. High scores (range from 0 to 100) are indicative of better quality.

Finally, participants assigned to the experimental condition 4 (N = 48) were subjected to an anchor with a negative rumor. They followed the same procedure as in experimental condition 2, but the rumor talked about the bad quality of the administrative services at the focal university: “For weeks rumors have circulated insistently about the bad quality of the administrative services at the University. According to these rumors, students and teachers have complained about poor attention and inefficiency in the provision of administrative services in this university”.

The same measure of quality was used as in Experiment 1, but adapted to the administrative services. A general assessment was obtained by asking: “Overall, what is the level of quality you perceive in these administrative services of the university?” Following this stem, there were three 7-point scales ranging from very low to very high, from awful to excellent, and from very poor to very good (alpha coefficient = .94). High scores were indicative of higher quality.

Manipulation check. To assess whether the manipulation worked, we examined the degree to which the message included in experimental conditions 2 and 4 is perceived as a rumor. We also tested whether the formal communication vs. the rumor differ in the credibility of the information they offer. To do so, a total of 14 participants answered a short questionnaire after reviewing the two formats used in the experiment: formal communication (for participants it was called Annex A) vs. rumor (for participants it was called Annex B). Ages ranged from 24 to 52 years, with a mean of 37.07 (SD = 9.86), and 71.4% of them were women. All participants (100%) agreed that the information in experimental conditions 2 and 4 refers to a rumor (Annex B). They also agreed (100%) that the information included in experimental conditions 1 and 3 (formal communication, Annex A) has more credibility than the information delivered in experimental conditions 1 and 3 (rumor). Participants also indicated the degree to which each format for offering information is more credible than the other (formal communication vs. rumor). To this end, we considered two items: “Information in Annex A is more credible than information in format B” and “Information from format B is more credible than information from format A”. As mentioned above, Annex A referred to the formal communication format, whereas Annex B referred to the rumor format. We used a scale ranging from 1 (strongly disagree) to 7 (strongly agree). Results indicated that formal communication (M = 6.79, SD = 0.43) is much more credible to participants, T₍₁₄₎ = 28.34, p < .01, than the rumor format (M = 1.14, SD = 0.36). In general, we confirmed that information contained in experimental conditions 2 and 4 is perceived as rumor and has less credibility than the formal communication from experimental conditions 1 and 3.

Results

Descriptive results for the four conditions were the following: formal positive communication, experimental condition 1 (M = 4.58, SD = 0.95); positive rumor, experimental condition 2 (M = 4.70, SD = 1.13); formal negative communication, experimental condition 3 (M = 3.86, SD = 1.15); and negative rumor, experimental condition 4 (M = 3.78, SD = 1.15). A 2 x 2 analysis of variance was computed to test the independent effects of information (positive vs. negative) and communication (formal vs. rumor), as well as their interaction. The main effect of information was significant, F_{(1, 168)} = 23.92, p < .01, indicating that quality perceptions were worse for users who received initial negative information than for those who received initial positive information. By contrast, neither the main effect of communication, F_{(1, 168)} = 0.01, p > .05, nor the effect of the interaction, F_{(1, 168)} = 0.33, p > .05, was significant. Thus, H2 was not confirmed. The anchoring effect of type of information (negative vs. positive) was significant, but it occurred with the same intensity for both formal communication and rumors (see Figure 6).

Figure 6

Means for Experiment 2

The current research study has two main goals. First, it examines the asymmetry of the effects of positive vs. negative information on the anchoring bias – regarding citizens’ quality perceptions of public services. The results indicate that users of public services are more sensitive to negative initial information (anchor) than to positive information. In fact, quality perceptions of participants receiving initial negative information were significantly worse than quality perceptions of users who did not receive any initial information (control group) and quality perceptions of participants receiving initial positive news about the service. By contrast, and compared to the control group, receiving initial positive information did not improve quality perceptions. Second, this research examines whether the anchoring bias is more accentuated in formal communication than in rumors. According to the findings, there are no significant differences between formal communication and rumors. Although the anchoring effect is revealed once again, its magnitude is the same for both formal and informal means of communication. In the remaining paragraphs, the theoretical and practical implications of these results are discussed.

Theoretical Implications

One of the challenges in the investigation of public services is to understand their bad reputation among citizens. As mentioned above, the particular characteristics of public organizations (criticism of public services during political campaigns, unpopular decisions by public administrators, etc.) can facilitate the predominance of bad news in the media (Grint, 2010). However, once the bad news is spread, other mechanisms can magnify its negative impact on citizens’ quality perceptions. The current research study supports the existence of a combination of anchoring and negativity biases that partly explains the pernicious effect of bad news. Coinciding with the anchoring heuristic (Epley & Gilovich, 2001), initial information acts as a starting point that has an influence on subsequent user perceptions of quality in public services. Nevertheless, this is just one part of the question. The anchoring effect emerges only for bad news, confirming the negativity bias and the special sensitivity of humans to bad events (Baumeister et al., 2001). Bad news easily produces a negative marker associated with the public service that is transferred to citizens’ quality perceptions.

This phenomenon is observed in participants who use the service in question frequently (at least once a month). Thus, they are not citizens who are unfamiliar with the functioning of the service and evaluate the public organizations based only on news. On the contrary, the participants have certain direct knowledge and frequent contact with the service providers. Even so, they are influenced by the bad fictitious news, which deteriorates their perceptions to a level below the quality evaluations of a control group that does not receive any initial information.

Because of the high credibility of formal communication (Chung et al., 2015; DiFonzo & Bordia, 1997), it was hypothesized that the anchoring effect would be greater for this formal communication than for rumors, but, unexpectedly, there were no significant differences between the two means of communication. Previous research efforts suggest that the impact of rumors on decision-making can be relevant in situations involving time pressure. For example, in stock market dynamics (da Cruz & Gomes, 2013; DiFonzo & Bordia, 1997), quick buying and selling decisions usually have to be made without credible formal reports. However, the present research study shows that rumors can have the same impact as formal communication in situations where users evaluate the quality of a public service. Previous research studies confirmed that the anchoring bias exists even when the information presented to participants is arbitrary (Tversky & Kahneman, 1974), but it is relatively surprising that a rumor can produce the same bias in frequent users of public services as formal communication. In the specific case of bad news, the mere existence of a rumor creates an alert that helps to prevent the bad quality of the public organizations. Tentatively, these findings may be interpreted based on Mintzberg’s (1994) characteristics of information management, which assume that rumors can be very useful in producing fresh and quick information and warnings about potential problems. Consequently, rumors influence people and their perceptions to the same degree as formal communication.

Practical Implications

Public administrators have a difficult challenge related to the management of information. Although positive information can be perceived as useful, the results of the current research study warn about focusing only on spreading good news. On the contrary, the main effort of administrators and other people responsible for public services should be directed toward the management and avoidance of bad news in different ways. For example, it is well known that public administrators are usually forced to make complex and sometimes unpopular decisions that are frequently transferred to the media. This decision-making should be managed in a way that allows sense-making and a correct interpretation of the reasons underlying the decisions, reducing the potential negativity of the news. A proactive attitude is required to offer the right interpretation of decision-making before another actor processes and communicates it in a biased manner. To this end, participation of citizens in public organizations can be helpful (Renedo, Marston, Spyridonidis, & Barlow, 2015). Additionally, special efforts can be made during critical periods (e.g., political campaigns) when public services can be scrutinized to magnify their negative facets. Before these important periods, public administrators can prepare a specific plan (e.g., contacting political candidates) to prevent the public service from exaggerated bad news. Finally, public managers should be aware that rumors could have the same negative effects on the reputation of the quality of public services as formal communication. The best way to avoid negative rumors is probably by offering information. Rumors are especially easy to spread when people do not have access to information from those responsible for services. A proactive attitude is useful again, preparing specific communication plans designed to quickly counter unfounded rumors and avoid ambiguity (Carpenter & Krause, 2012).

Limitations and Future Research

Although the manipulation checks worked and the results were congruent with the hypotheses, there are other mechanisms beyond anchoring and negativity bias that could play a role, opening new doors for future research. The possible influence of the presence of researchers during the evaluation process of public services by participants is especially noteworthy. The mere presence of a researcher could stimulate social desirability and a tendency to answer according to the nature of the information presented by the researcher. With this in mind, additional research can be carried out in a laboratory context where participants receive the information through a computer and without the presence of a researcher during the evaluation of service quality. Another area where future research can provide additional insights is the consideration of active control groups (e.g., Temple & Ellenberg, 2000). In our first experiment, we included a passive control group (without previous information), whereas for the second experiment we did not consider a control group because we concentrated on the differences between formal vs. informal communication. However, the incorporation of active control groups in future research studies (e.g., receiving real information about the public organization being evaluated) could offer a richer picture of the impact of negative news on the reputation of the public sector.

The present research study supports the idea that citizens are especially sensitive to bad news about the public sector, negatively impacting the reputation of public service quality. In addition, the anchoring bias is found not only in formal communication, but also in rumors. These biases (negativity and anchoring) help us to understand the bad reputation of public services and suggest to public administrators that formal communication and rumors should be managed proactively to avoid biased news that can produce an unfair negative reputation associated with the public sector.