Does it pay to avoid speaking straight about controversial issues? Impact of argumentative ambiguity on the perception of the speaker

Abstract

Purpose

The goal of the research was to check whether the sender communicating in an ambiguous manner can gain the benefits in three aspects: perception of his or her image, evoking agreement with issue stand and intention to support him or her in the election. Impact of ambiguous messages was compared with the impact of messages consistent or inconsistent with participants' opinions.

Design/methodology/approach

Two experiments were conducted. Participants were randomly divided into three groups and each of them was presented with (1) a message supporting or (2) opposing given proposition or (3) an ambiguous message not revealing the ultimate opinion of the sender. Participants' initial views on the issues were measured. In experiment 1 an expert message concerned the building of the nuclear power plant was presented. In experiment 2 it was a politician's message about introducing the guaranteed number of parliamentary seats for women.

Findings

The results suggest that a strategy of argumentative ambiguity applied by senders may be beneficial if the point is to avoid recipients' objections. However, the consequences of ambiguity for the evaluation of sender's credibility and voters' intentions can be seen to be negative.

Originality/value

This research was looking for the reconcile of the contradictory results of previous research, which may have their source in the various operationalization of ambiguity. It was focused at an argumentative ambiguity (i.e. presenting different points of view without declaring support for any of the options).

Citation

Koniak, P. and Cwalina, W. (2022), "Does it pay to avoid speaking straight about controversial issues? Impact of argumentative ambiguity on the perception of the speaker", Journal of Communication Management, Vol. 26 No. 1, pp. 84-97. https://doi.org/10.1108/JCOM-11-2020-0154

Ambiguity is natural, although rarely noticed, part of human communication. As Keysar (2007, p. 71) states, “Most people, most of the time, think that what they say is pretty clear. Ambiguity is not routinely noted when people normally communicate.” Sources of ambiguity are so numerous that some ambiguity is virtually guaranteed (Bavelas et al., 1990; Bello, 2005; Forgas and Cromer, 2004). According to Bavelas et al. (1990, p. 21), ambiguous message “say nothing while saying something”, and what characterizes it are: “self-contradictions, inconsistencies, subject switches, tangentializations, incomplete sentences, misunderstandings, obscure style or mannerisms of speech, the literal interpretations of metaphor and the metaphorical interpretation of literal remarks.” From a rhetorical perspective, Muzzillo (2010) distinguishes seven rhetorical devices that create ambiguity: renaming a referent, purposeful manipulations of information, omissions (e.g. a syllogism with an unstated premise), unexpected reversals of meaning, comparative descriptives, stretches (or exaggerations) and pragmatic (e.g. idioms or jargon).

Ambiguity may be “poor” communication by some standards, but in many situations, “equivocation is a good solution to a bad situation” (Bavelas et al., 1990, p. 159). It can be used strategically in both organizational (Eisenberg, 1984; Gioia et al., 2012) and political contexts (Page, 1976; Tomz and Van Houweling, 2009). According to Eisenberg (1984; Eisenberg and Witten, 1987), ambiguity is a relational variable which arises through a combination of source, message and receiver factors. Strategic ambiguity fosters the existence of multiple viewpoints in organizations, and it is commonly found in organizational missions, goals and plans. “It allows for multiple interpretations to exist among people who contend that they are attending to the same message – i.e. perceive message to be clear” (Eisenberg, 1984, p. 231). Then, ambiguity may be used strategically to foster agreement on abstractions without limiting specific interpretations – to promote unified diversity.

In politics, Page's emphasis allocation theory of ambiguity (1976) assumes that candidates strategically select unclear messages to divert their voters' attention from particularly controversial political problems and make them focus on issues or goals on which there is more agreement (e.g. country development, economic growth or women's rights). They consciously present their positions as ambiguous and do not make much effort to develop it. On the other hand, they do not usually have enough resources that would help them discuss their views in a precise and detailed way. As Page (1976) notes, voters are not interested in such detailed messages either, because analyzing and understanding them would require a lot of effort. As a result, ambiguous message can effectively produce more agreement than clear message (Carmines and Gopoian, 1981; Tomz and Van Houweling, 2009).

Nowadays, growing attitude polarization in society as well as preferring by citizens their echo chambers are serious problem (e.g. Baldassarri and Bearman, 2007). Ambiguous messages can become an opportunity to present a point of view different from this preferred by particular citizens, especially when discussed topic is raising strong divergences in public opinion and evoking extreme assessments. As a result, ambiguous message can become helpful in reaching social consensus, or at least mitigation of ideological conflicts. For this reason, it is important to explore the consequences of using such strategy of communication.

Consequences of the ambiguity of messages for the evaluation of the sender

Although there is a lot of research on the communicative effects of using ambiguity, only a few studies focus on its influence on the communicator's evaluation and agreement with his or her message. This effect can be important for the future persuasion efforts. The speaker may use ambiguity strategically, to avoid negative audience responses which might damage future persuasive attempts. While the goal of persuasion is an agreement based on understanding, the use of strategic ambiguity suggests that in route to that end, one facilitating step might be to create a temporary agreement based on an unclear understanding of the speaker (Williams, 1980). In political or public context, ambiguity can also be a politician's or other public person's strategy of trying to gain support, for example by facilitating the projection of receivers' own views on him or her.

The results of the research conducted so far on eliciting the agreement of recipients with ambiguous messages and on the perception of the communicator's image are often contradictory. In early studies, Zimbardo (1960) and Manis (1961) found the assimilation effect in the judgments of ambiguous statements. When responding to ambiguous messages, recipients perceived the communicator's message as more similar to their own attitudes than when they received unambiguous messages that they essentially disagreed with. Furthermore, they judge the ambiguous messages as more moderate opinion about the issue than when they are responding to the unambiguous messages (pro or contra).

In a series of experiments Williams (1980; Goss and Williams, 1973; Williams and Goss, 1975) found that ambiguous statements about issues evoking disagreement among audience positively affect the speaker's character ratings (e.g. reliability, honesty and honorableness) but the ambiguity had no effect on competence ratings. Moreover, in the case of issues evoking disagreement among the audience, the ambiguous message produced significantly more agreement than did the clear message. Williams and Goss (1975) suggest that ambiguity tends to suppress incongruent meanings, so that listener calls up only congruent interpretations. In consequence, the speaker is judged to be more favorable than if he was identified with a clearly stated disagreeable message. Then, ambiguity can help prevent losses in credibility. It is a kind of “character insurance.”

However, the strategic use of message ambiguity can have a significant drawback (Dulek and Campbell, 2015). Ambiguity is one feature of powerless or low-intense language (Hamilton, 1998). A person using the powerless style exhibits a relatively large number of hedges, abstract terms, inconsistencies, and/or declarative sentences and, in consequence, messages exhibiting the linguistic cues of powerlessness produce relatively low ratings of speaker power (Bradac and Mulac, 1984). According to Hamilton (1998), one of the instances of ambiguity is a situation, when a speaker intentionally avoids opinionated language. He defines opinionated language as “degree to which the source accepts (or rejects) a particular idea and the degree to which source accepts (or rejects) receivers depending on whether they agree or disagree with this idea” (Hamilton, 1998, p. 112). Results of experiments and meta-analyzes conducted by Hamilton (1998; Hamilton and Stewart, 1993; Hamilton and Thompson, 1994) provide evidence of “charisma sequence”, in which ambiguity lowers the audience's perception of the speaker's dynamism, which negatively affects the assessment of her competence and ultimately lowers his or her credibility.

It seems that the contradictory results obtained in the research may, among others, have their source in various operationalization of ambiguity (e.g. single statement vs message). We are focusing on the specific form of message ambiguity, namely – argumentative ambiguity, which we define as presenting opposing viewpoints of the problem (pro and con arguments) without declaring support for any of the options, i.e. without declaring which side of the issue is preferred, etc. (Cwalina and Koniak, 2007). This is a message in which each of the arguments or statements is clear, but the entire message cannot be clearly interpreted. The argumentative ambiguity is neither equated to the two-sidedness of the message nor the message with omitted conclusion (e.g. Hovland et al., 1963). With two-sided argumentation, one side of the issue is more or less explicitly favored – arguments for this side are stronger or more numerous, the conclusion of the message supports this side, etc. In a message with the omitted conclusion, the receiver is able to infer this conclusion based on the structure of the argumentation. In an argumentatively ambiguous message, neither the conclusion nor the suggestive structure of argumentation is present – neither side of the issue is favored; both receive equal attention.

To explore the effectiveness of argumentatively ambiguous message, we need to compare it with messages in which the sender's view is clearly stated. In research presented here our control messages are messages with exactly the same arguments as those presented in ambiguous messages, but with the explicit conclusion that the sender is favoring or opposing discussed proposition. However, looking at the message issue stand as a pro or con is only one of the encodings of this stand. In addition to this descriptive encoding, this issue stand can be also encoded evaluatively as compatible or incompatible with the receiver's views (Wyer et al., 1991). For this reason we need to take into account participants' initial attitudes toward the message issue – to compare the effect of the ambiguous message with the unambiguously pro- and counterattitudinal message.

Pro- and counterattitudinality of message are the variables used in persuasion studies (Clark et al., 2008). However, we focus on assessing the sender and agreeing with him or her rather than changing attitudes toward the object of the message. Results of previous studies suggest that at least in some conditions the source of the counterattitudinal message is evaluated more unfavorably than the source of the proattitudinal message (Brehm and Lipsher, 1959; See and Petty, 2006). It should be also noted than in the context of a comparison between ambiguous and unambiguous messages we are interested in a primary aspect of attitudinal discrepancy: whether the receiver's and communicator's standpoints are at the same or opposite side of the pro-con continuum (Lange and Fishbein, 1983), not in the level of this similarity or discrepancy (e.g. Kaplowitz and Fink, 1997).

Based on the analysis of the impact of ambiguous and unambiguously pro- and counterattitudinal messages on the perception of the image of speakers and on agreeing with their messages, we predict that:

H1.

Speaker communicating ambiguously will be evaluated more positively than speaker communicating a view opposite to the participants' view, and more negatively than speaker communicating view aligned with those of the participants.

H2.

Participants will be much more likely to agree with speaker if the message is ambiguous or compatible with their views than when it is not compatible.

To verify these hypotheses, we conducted two studies. In both of them message concerned controversial issue was presented. By controversial issue we mean an issue with a lot of disagreement existing in society, where substantial parts of society are holding opposing attitudes toward this issue. Our studies were conducted in Poland, and we chose two issues where social attitudes were diversified: building of the nuclear power plant and introducing the guaranteed number of parliamentary seats for women. The idea of building the first Polish nuclear power plant is a recurring theme in Polish politics, with a substantial amount of both opponents and proponents of such solution. For example, in 2018 36.7% of surveyed Poles were supporting the building of the nuclear power plant, while 43.3% were opposing this (Szaniawski, 2018). Similarly, underrepresentation of women is politics is often raised in Polish media, especially during the election time. 2019 election resulted in a quota of female representation in Poland's parliament equal to 28.47% for lower house and 24% for the upper house (Szczęśniak, 2019). At the same time the propositions of changing this state of affairs by introduction of legal regulations are sources of social disagreement – e.g. 36% of respondents were accepting proposals of introduction of the guaranteed number of places on voting lists for women while 53% were opposing such solution (CBOS, 2010).

Study 1

Method

Participants

A sample of 343 participants (M_age = 39.07 years, SD = 14.38, 51.6% female) of the Ariadna Polish Research Panel (https://panelariadna.pl/) were recruited to participate in exchange for credit points (members of Ariadna can collect points by participating in surveys and exchange them for prizes). A sensitivity analysis conducted with G*Power (Faul et al., 2007) showed that our sample was sufficient to detect effects of f < 0.17 with a power of 0.80 for between-subjects ANOVA. The Research Ethics Board of SWPS University of Social Sciences and Humanities approved the research procedures for experiments.

Stimulus material and procedure

Participants were randomly assigned to one of the three experimental groups in which the manipulated message concerned the nuclear power plant (pro vs ambiguous vs con). Prior to the presentation of the stimulus material, all participants were asked to express their own opinions on the construction of a nuclear power station in Poland (they were choosing between three options: “I am for”, “I am against” and “I have no opinion”). This question was placed between questions about other topics, which were used as buffer questions. Only individuals who initially reported a pro- (N = 146) or con- (N = 197) attitude toward this issue proceeded with the experiment, as only those attitudes allowed for a clear definition of attitude-consistent vs attitude-inconsistent message.

Participants were randomly divided into three groups and each of them was presented with: (1) a message containing a two-sided argumentation and supporting construction of a nuclear power station; (2) a message containing a two-sided argumentation and opposing construction of a nuclear power station; or (3) an ambiguous message containing only pro and con arguments, without revealing the ultimate opinion of the sender. Each message contained three pro arguments (focused on lowering prices of the electricity, reducing fossil fuels' exploitation and high safety standards) and three con arguments (focused on the risk of the radioactive contamination, problems with storage of the radioactive waste and uranium price fluctuations).

In the pro and con messages, the sender's ultimate position was stated straightforwardly at the beginning [ “I am in favor (against) this type of power stations”] and at the end of the message [“Given these points – in my opinion building of the nuclear power plant is a good (bad) solution”], with the argument consistent with the author's position presented as the second, and the inconsistent one as the first. In all messages, the pro and con arguments were preceded by the “on the one hand …” or “on the other hand …” phrase. In order to control the influence of the sequence of the presented arguments, the ambiguous message was presented in either pro and con or con and pro version. Preliminary analyzes did not show any significant influence of the order of argument presentation; therefore, this factor was omitted in the subsequent analyzes. Authorship of the message was prescribed to Professor Jerome Friedman, nuclear physic, Nobel Prize winner (fictional person, actually). Stimuli materials for our both studies are available at https://doi.org/10.5281/zenodo.4587203.

For the manipulation check purposes, after exposure to the message, participants answered how they perceived the ambiguity of the sender's issue stand (very ambiguous-very unambiguous). Next they indicated their level of agreement with the sender (strongly disagree-strongly agree) and answered questions concerning sender's competence (very incompetent-very competent, definitely not an expert-definitely is an expert, r = 0.74, p < 0.001). All scales ranged from 1 to 101.

Results

Manipulation check

Two-way ANOVAs were performed in design: 3 (message: pro vs ambiguous vs con) × 2 (participant's attitude: pro vs con), where the dependent variable was the perception of the ambiguity of the sender's issue stand. The results indicate [F(2, 337) = 31.99, p < 0.001, η_p² = 0.160] that the ambiguous message was evaluated as more ambiguous (M = 46.34, SD = 24.91) both compared with the message in favor of nuclear power plants [M = 65.92, SD = 25.30; F(1, 337) = 33.79, p < 0.001, η_p² = 0.091] and with the message against them [M = 70.15, SD = 25.92; F(1, 337) = 52.11, p < 0.001, η_p² = 0.134]. There was no significant difference between perceived ambiguity of the pro and con messages (F < 1.20). The effect of the participants' attitude and the interaction between this attitude and the message were not significant (Fs < 1).

Evaluation of sender's image

Two-way ANOVA with evaluation of the sender's competence as a dependent variable showed no significant effects of the message or the participant's attitude (Fs < 1). Hypothesis 1 states that speaker communicating ambiguously will be evaluated more positively than speaker communicating a view opposite to the participants' view, and more negatively than speaker communicating view aligned with those of the participants. The interaction effect between participant's attitude and position in the message was significant (F(2, 337) = 13.77, p < 0.001, η_p² = 0.076). However, an a priori comparison conducted by contrast analysis showed that the impact of the message on the evaluations of the sender's competence was other than hypothesized. In fact sender was perceived as more competent when his message was aligned with participants' attitude, but when he was speaking ambiguously or contrary to the participants' attitude, evaluation of his competence was lower. The proponents of the building of the nuclear power plant evaluated the competence of the author of the pro message (M = 73.05, SD = 21.26) higher both than the competence of the sender of the con message [M = 56.83, SD = 24.64; F(1, 337) = 11.15; p < 0.001; η_p² = 0.032] and the ambiguous one [M = 63.58, SD = 19.42; F(1, 337) = 4.81; p = 0.029; η_p² = 0.014]. However, the differences between evaluation of the image of the sender in the last two groups were not significant (F = 2.37, p = 0.125). Analogically, the opponents of the building of the nuclear power plant evaluated the competence of the author of the con message (M = 73.23, SD = 21.83) higher both than the competence of the sender of the pro message [M = 55.39, SD = 23.85; F(1, 337) = 16.65; p > 0.001; η_p² = 0.047] and the ambiguous one [M = 60.78, SD = 21.02; F(1, 337) = 11.69; p < 0.001; η_p² = 0.034]. The differences between evaluation of the image of the sender in the last two groups were not significant (F = 1.87, p = 0.173).

Agreement with the message

Two-way ANOVA, where the dependent variable was the degree of agreement with the sender's message showed that proponents of the nuclear power plants (M = 65.70, SD = 24.30) were more willing to agree with the sender that the opponents [M = 58.35, SD = 28.05; F(1, 337) = 7.61; p = 0.006; η_p² = 0.022]. This effect was qualified by the interaction of the message with the participants' opinion about this issue [F(2, 337) = 22.00; p < 0.001; η_p² = 0.115]. Hypothesis 2 states that participants will be much more likely to agree with a politician's message if the message is ambiguous or compatible with their views than when it is not compatible. The results of a priori comparison supported the expectations, but only partially. Proponents of the nuclear power plants agreed more with a message if this message was compatible (M = 75.51, SD = 24.16) than if it was incompatible with their views (M = 54.33, SD = 26.23) – F(1, 337) = 14.35; p < 0.001; η_p² = 0.041). Furthermore, the level of agreement with an ambiguous message (M = 66.32, SD = 20.41) was similar to an agreement with a compatible message [F(1, 337) = 3.42; p = 0.065; η_p² = 0.010 – although it should be noted that this difference is close to the conventional level of significance], but it was higher than agreement with an incompatible message [F(1, 337) = 5.64; p = 0.018; η_p² = 0.016].

Opponents of the nuclear power plants also agreed more with a message if it was compatible (M = 72.02, SD = 27.88) than if it was incompatible with their views (M = 43.45, SD = 30.07) – F(1, 337) = 32.19; p < 0.001; η_p² = 0.087). However, for this group, level of agreement with an ambiguous message (M = 57.21, SD = 23.33) was both lower than agreement with a compatible message [F(1, 337) = 12.47; p < 0.001; η_p² = 0.036], and higher than an (dis)agreement with an incompatible message [F(1, 337) = 9.16; p = 0.003; η_p² = 0.026].

Discussion

Results of study 1 shows, that argumentatively ambiguous message can help to avoid disagreement and, at least among proponents of the discussed issue, even to be evoking agreement at a level similar to the message clearly supporting point of view of this group of receivers. However, as a side effect, ambiguity can lower the evaluation of the competence of the author of the message. The expert was evaluated as more competent when he was presenting a point of view compatible with attitudes of recipients. Speaking ambiguously or taking a position overtly incompatible with receivers' attitudes was harmful to his image. These results align with observation of the growing tendency to evaluate high only those experts who speak what people want to hear (Nichols, 2017).

In this experiment, we used an expert as the source of the message and evaluation of the competence, the basic dimension of the experts' image, was measured. It's not clear whether other dimensions of the image are also affected by the ambiguity of the message. While ambiguous message can be interpreted as a sign of inability to conclude, it can be also a sign of the hiding the truth. This possibility becomes important when the message comes from the politician, since politicians are arousing more suspicions than other sources (McGraw et al., 2002). When the politician is speaking, people may be more vigilant about trustworthiness than toward other dimensions of the image.

For this reason in study 2 we were testing our hypotheses in situation, when the politician was the source of the message and when different topic was discussed. Moreover, since ambiguity can affect different dimensions of the image in different ways, in study 2 we were using a multidimensional measure of the image. Lastly, with politicians, another aspects of evaluation are becoming important – i.e. intention to support this politician in an election. In study 2 we were asking about this intention and our expectation were:

H3.

Intention to vote for a politician will be stronger when his view is compatible with the participants' views than when it is ambiguous or incompatible.

Study 2

Method

Participants

Participants (N = 238, M_age = 21.06 years, SD = 1.82, 75.2% female) were psychology majors at a university in southeast Poland. Participation was voluntary. The experiment was conducted in small groups of about 15–20 persons each. The experimenter handed the subjects a booklet containing the experimental material, and each participant proceeded at their own pace. A sensitivity analysis conducted with G*Power (Faul et al., 2007) showed that our sample was sufficient to detect effects of f = 0.20 with a power of 0.80 for between-subjects ANOVA.

Stimulus material and procedure

Participants were randomly assigned to one of the three experimental groups in which the manipulated message concerned the introduction of the guaranteed number of parliamentary seats for women (pro vs ambiguous vs con). Prior to the presentation of the stimulus material, all participants were asked to express their own opinions on the idea of guaranteed number of parliamentary seats for women (on a 7-point scale, where 1 is “strongly against” and 7 “strongly for”). Only individuals who initially reported a pro- or con-attitude toward this issue proceeded with the experiment, as only those attitudes allowed for a clear definition of attitude-consistent vs attitude-inconsistent message. The first group included those who in the value scale selected above-the-middle scoring (5, 6 or 7; N = 176); the other group comprised participants who chose the lower (1, 2 or 3; N = 62) (see Nemeth and Endicott, 1976).

Participants were randomly divided into three groups, each of which was presented: (1) a message containing a two-sided argumentation and supporting the proposition of securing a guaranteed number of parliamentary seats for women; (2) a message containing a two-sided argumentation and opposing such proposition; or (3) an ambiguous message containing only pro and con arguments, without revealing the ultimate opinion of the sender. Each message contained one pro argument (“Introducing such a guaranteed number of seats will make women start to be proportionally represented in parliament. This will help to level their political chances”) and one con argument (“Voters should decide who sits in parliament. It is impossible to determine in advance whom and in what proportions is to be chosen”). In the pro and con messages, the sender's ultimate position was stated straightforwardly [“I am for the introduction (against the introduction) of the guaranteed number of parliamentary seats for women”]. In all messages, the pro and con arguments were preceded by the “on the one hand …” or “on the other hand …” phrase. In order to control the influence of the sequence of the presented arguments, the ambiguous message was presented in either pro and con or con and pro version. Preliminary analyzes did not show any significant influence of the order of argument presentation; therefore, this factor was omitted in the subsequent analyzes.

Authorship of the message was prescribed to Andrzej Nowak, MP. Neither party affiliation nor any ideological cues were provided, which prevented perceiving him through the partisan or ideological lenses. Also, the name “Nowak” is a very common Polish surname, and with 460 members of the Polish lower house, participants could treat the author of the message as a real politician (in fact – he was a fictitious person). After exposure to the message, participants evaluated its author on the 12-adjective, 5-point Leathers Personal Credibility Scale (Leathers, 1992). This scale is measuring three dimensions of image – trustworthiness, competence and dynamism. The trustworthiness index consists of four scales: dishonest–honest, evasive–straightforward, untrustworthy–trustworthy and insincere–sincere (Cronbach's α = 0.86). The competence index also consists of four scales: incompetent–competent, unqualified–qualified, poorly informed–well informed and unintelligent–intelligent (Cronbach's α = 0.80). The third dynamism index scales are: unassertive–assertive, timid–bold, meek–forceful and inactive–active (Cronbach's α = 0.80). For each participant, three composite indexes were calculated. Their values ranged from 1 to 5.

Participants next answered questions (scales ranged from 1 to 7) concerning agreement with Nowak's position (I strongly disagree – I strongly agree) and intention to vote for Nowak (definitely not – definitely yes). For the manipulation check purposes, they were also answered questions about the direction of Nowak's position (strongly against – strongly for) and the ambiguity of his position (definitely unambiguous – definitely ambiguous).

Results

Manipulation check

Two-way ANOVA performed in design: 3 (message: pro vs ambiguous vs con) × 2 (participant's attitude: pro vs con) showed that the sender's issue stand was perceived as intended [F(2, 231) = 48.95; p < 0.001; η_p² = 0.298], with the sender of the pro parity message perceived as more in favor of [M = 5.40, SD = 1.07; F(1, 231) = 34.32; p < 0.001; η_p² = 0.129], and the sender of the con parity message perceived as more against [M = 2.86, SD = 1.50; F(1, 231) = 23.09; p < 0.001; η_p² = 0.091] the idea of guaranteed number of parliamentary seats for women than the sender of the ambiguous message (M = 4.10, SD = 1.22). The difference between the perceived position of the sender of pro and con messages was also statistically significant [F(1, 231) = 97.31; p < 0.001; η_p² = 0.297]. The effect of the participants' views and the interaction between participants' views and the message were not significant (Fs < 1).

Two-way ANOVA showed also [F(2, 232) = 19.97; p < 0.001; η_p² = 0.147] that the ambiguous message was evaluated as more ambiguous (M = 4.93, SD = 1.82) both compared with the message for introducing the guaranteed number of parliamentary seats for women [M = 3.78, SD = 1.79; F(1, 232) = 21.09; p < 0.001; η_p² = 0.083] and with the message against it [M = 3.29, SD = 1.80; F(1, 232) = 34.31; p < 0.001; η_p² = 0.129]. There was no significant difference between perceived ambiguity of the message supporting the parity and the one against it [F(1, 232) = 1.58; p = 0.210]. The effect of the participants' views and the interaction between participants' views and the message were not significant (F < 1 and F = 2.55, p = 0.080, respectively).

Evaluation of Politician's image

Two-way ANOVAs were performed in design: 3 (message: pro vs ambiguous vs con) × 2 (participant's attitude: pro vs con). The dependent variables were three dimensions of a politician's (sender's) image: trustworthiness, competence and dynamism. Significant effects of the message were found for the sender's trustworthiness [F(2, 229) = 4.11; p = 0.018; η_p² = 0.035] and dynamism [F(2, 231) = 3.75; p = 0.025; η² = 0.031], while for the evaluation of politician's competence there was only a statistical trend [F(2, 231) = 2.94; p = 0.055; η_p² = 0.025].

Because no hypotheses were formulated on the existing main effects (the position in the message), they were put to post hoc analyses by the Duncan test. The results showed that the sender of an ambiguous message was evaluated as less trustworthy (M = 2.99, SD = 1.00) than the sender of the message against the introduction of the guaranteed number of women in the parliament (M = 3.36, SD = 0.91, p = 0.022) and sender of the pro parity message (statistical trend: M = 3.28, SD = 0.93, p = 0.061). However, the message against the introduction of the guaranteed number of women in the parliament, irrespective of participants' opinions, promoted the sender's evaluation as more dynamic (M = 3.79, SD = 0.61) compared with the ambiguous (M = 3.49, SD = 0.86, p = 0.019) and the pro parity message (M = 3.51, SD = 0.78, p = 0.023). Post hoc analysis indicated that evaluation of competence of the sender wasn't significantly affected by the position stated in the message.

Hypothesis 1 states that speaker communicating ambiguously will be evaluated more positively than speaker communicating a view opposite to the participants' view, and more negatively than speaker communicating view aligned with those of the participants. However, the interaction effect between participants' initial views and position in the message was insignificant. For this reason, an a priori comparison was conducted by contrast analysis. The results of contrast analysis partially confirmed hypothesized pattern but only in the evaluation of politician's trustworthiness. Proponents of the parity evaluated sender's trustworthiness significantly lower when the message was ambiguous (M = 2.97, SD = 1.05) than when it was compatible [M = 3.33, SD = 0.99; F(1, 229) = 4.38; p = 0.037; η_p² = 0.019], but not when it was incompatible with their views [M = 3.23, SD = 0.96; F(1, 229) = 2.25; p = 0.135]. However, among the proponents of parity the differences between the “compatible” (pro parity) and “incompatible” messages (con parity) were insignificant [F(1, 229) = 0.32; p = 0.572]. At the same time, opponents of the parity evaluated trustworthiness of the sender of the message compatible with their views (M = 3.76, SD = 0.61) significantly higher both than the trustworthiness of the sender of the ambiguous [M = 3.05, SD = 0.89; F(1, 229) = 5.98; p = 0.015; η_p² = 0.025] and incompatible message [M = 3.10, SD = 0.73; F(1, 229) = 4.32; p = 0.039; η_p² = 0.018]. Moreover, the difference between trustworthiness evaluation of the senders of ambiguous and incompatible messages wasn't statistically significant [F(1, 229) = 0.03; p = 0.860].

As for the evaluation of politician's competence opponents of the parity perceived sender of the message compatible with their views as more competent than the sender of the incompatible message [M = 3.76, SD = 0.56 and M = 3.17, SD = 0.62, respectively; F(1, 231) = 5.75; p = 0.017; η_p² = 0.024]. Evaluation of the competence the sender of the ambiguous message (M = 3.41, SD = 0.80) was not statistically different neither from the evaluation of the senders of compatible [F(1, 231) = 2.45; p = 0.119] nor incompatible [F(1, 231) = 1.15; p = 0.284] message. Different pattern of result was found for the dynamism evaluations. Opponents of the parity evaluated sender of the message compatible with their views as more dynamic than the sender of the ambiguous message [M = 3.87, SD = 0.64 and M = 3.37, SD = 0.94, respectively; F(1, 231) = 4.39; p = 0.037; η_p² = 0.019]. However, neither the differences between evaluation of the dynamics of the senders of the messages incompatible (M = 3.47, SD = 0.65) and compatible [F(1, 231) = 2.33; p = 0.128] nor incompatible and ambiguous [F(1, 231) = 0.19; p = 0.663] were not significant. Among proponents of the parity evaluation of the sender competence and dynamism wasn't influenced by the message (all Fs < 1 and Fs ≤ 2.83, p ≥ 0.094, respectively).

Agreement with the position expressed in message

Two-way ANOVA was performed, where the dependent variable was the degree of agreement with the presented MP's position on introducing the guaranteed number of parliamentary seats for women. The results showed that significant were: the position expressed in the message, F(2, 232) = 6.99; p = 0.001; η_p² = 0.057, and its interaction with participants' opinion about this issue, F(2, 232) = 32.44; p < 0.001; η_p² = 0.219.

The results of the post hoc Duncan test showed that participants, regardless of their opinions, agreed with the ambiguous message more significantly (M = 4.74, SD = 1.64) than with the message against introducing the guaranteed number of women in the parliament (M = 4.22, SD = 1.78; p = 0.043). Other differences failed to reach the level of statistical significance (p ≥ 0.186, agreement with the pro message: M = 4.55, SD = 1.79).

In order to verify Hypothesis 2, which states that participants will be much more likely to agree with a politician's message if the message is ambiguous or compatible with their views than when it is not compatible, an a priori comparison was conducted by contrast analysis. The results supported the expectations, but only for proponents of parity. These participants agreed more with a politician's message if their views were compatible (message for parity – M = 5.20, SD = 1.39) than if they were incompatible with the message (message contra parity – M = 3.76, SD = 1.62) – F(1, 232) = 25.08; p < 0.001; η_p² = 0.098. Furthermore, the level of agreement with an ambiguous message (M = 4.79, SD = 1.57) was similar to an agreement with a compatible message [F(1, 232) = 2.19; p = 0.140], but it was higher than agreement with an incompatible message [F(1, 232) = 13.06; p < 0.001; η_p² = 0.053].

Opponents of parity also agreed more with a politician's message if their views were compatible (message contra parity – M = 5.71, SD = 1.45) than if they were incompatible with the message (message for parity – M = 2.39, SD = 1.14) – F(1, 232) = 40.88; p < 0.001; η_p² = 0.150. However, for this group, level of agreement with an ambiguous message (M = 4.63, SD = 1.82) was lower than agreement with a compatible message [F(1, 232) = 5.14; p = 0.024; η_p² = 0.022], and higher than agreement with an incompatible message [F(1, 232) = 23.05; p < 0.001; η_p² = 0.090].

Support for politician

Two-way ANOVA was performed, where the dependent variable was participants' declarations on support for the politician. The results showed that politician's support was significantly influenced by the position expressed in the message, F(2, 232) = 4.88; p = 0.008; η_p² = 0.040, and its interaction with participants' opinions, F(2, 232) = 4.71; p = 0.010; η_p² = 0.039. The effect of participants' opinions was insignificant – F < 1.

The results of the post hoc Duncan test showed that the declared support for the politician, irrespective of participants' opinions on parity, was significantly higher among participants who had been presented the message contra women's guarantees (M = 3.28, SD = 1.66) than among the participants who had been presented ambiguous message (M = 2.73, SD = 1.57; p = 0.047). Other differences failed to reach the level of statistical significance (p ≥ 0.252, support for sender of the pro message: M = 3.03, SD = 1.70).

In order to verify Hypothesis 3, which states that intention to vote for a politician will be stronger when his view is compatible with the participants' views than when it is ambiguous or incompatible, an a priori comparison was made. Contrary to expectations, the results indicated no significant differences between “compatible” (M = 3.24, SD = 1.73) and “ambiguity” (M = 2.85, SD = 1.63) and “incompatible” (M = 3.05, SD = 1.65) messages in the group of proponents of parity (F's < 1.69; p ≥ 0.195). In the group of opponents of parity, support for a politician was markedly higher when their views were compatible with participants' views (M = 4.00, SD = 1.54) than when the message was ambiguous [M = 2.44, SD = 1.42, F(1, 232) = 9.63; p = 0.002; η_p² = 0.040] or incompatible [M = 2.33, SD = 1.46, F(1, 232) = 9.27; p = 0.003; η_p² = 0.038], with no difference between these two last messages (F < 1).

Discussion

Ambiguity seems to be rather harmful strategy when trustworthiness is at stake, especially when confronting with messages clearly supporting receivers' views. Moreover, when receivers were opposing discussed solution politician using ambiguity strategy was perceived as dynamism lacking. Furthermore, it was found that a politician opposing the parliamentary guarantees for women, irrespective of participants' opinions on the matter, was evaluated as more dynamic than a politician supporting such a position. At the same time, ambiguity was not affecting the evaluation of the competence of the source – neither in a positive nor negative way.

When speaking to the proponents of the parity, ambiguity was a good strategy for avoiding theirs disagreement. It was also not so bad strategy handling with opponents of the parity – at least it was not evoking as a strong disagreement as clearly stated view incompatible with theirs views.

Finally, when winning of the support is the goal, ambiguity seems to be completely ineffective strategy. Opponents of the discussed issue are willing to support the politician aligned with their views. And the support of proponents is unrelated to the message.

General discussion

The goal of our research was to analyze the impact of ambiguous messages on the perception of message sender. The analysis covered a message focused on one area that was defined as compatible or incompatible with the participants' views or as argumentatively ambiguous. It became apparent that the impact of ambiguous messages depends on whether the target of influence is the message sender's image, the degree of agreement with his or her opinions or intention to support him or her. In summary, the ambiguity strategy seems to be beneficial and effective in preventing the incitement of receivers' opposition or even in arousing theirs agreement. This result seems to go together with predictions of the emphasis allocation theory of ambiguity (Page, 1976): that candidates strategically electing to speak more ambiguously to distract voters' attention from controversial political issues may reap some benefits across the whole spectrum of the electorate and not just among their staunch supporters. However, ambiguity has negative consequences for the evaluation of sender image. Ambiguity seems to be especially harmful for the main dimension of the image of particular types of sender – competence for the expert and trustworthiness for the politician (see Hamilton, 1998). It also does not guarantee expected benefits, such as winning the support for a politician.

In both of our studied we found systematic differences between the reactions of proponents and opponents of the discussed issues. When assessing agreement with the message, proponents were more willing to treat ambiguous message similarly as the message clearly consistent with their views. Opponents were rather reluctant when confronted with ambiguity and declared the highest support for the source with views compatible with their own opinions. This result may suggest that merely being “against” required greater caution and restraint. Or, speaking differently, that merely being “for” resulted in the more shallowing processing of the presented message, as the “for” attitudes are weaker than “con” attitudes (Bizer and Petty, 2005).

Strength of the attitudes can be the potential moderator of the effectiveness of the strategy of ambiguity. As Page (1976) and Carmines and Gopoian (1981) suggest, the effectiveness of ambiguous messages depends to a large extent on the issues. The issue used in this experiment was not as much polarizing as, for example, abortion. Attitudes toward different issues, while still being pro or con, can also differ in strength. Attitude strength-related phenomena, like attitude involvement, attitude extremity or attitude certainty can be one of the factors worthy of further exploration in the context of the effectiveness of the ambiguity strategy.

The limitations of our studies should be addressed. Many of the effects observed here are not causal (given that attitude variable was measured, not manipulated). It should be also noted that politician presented in our study 2 was lacking the advantage of strong voter identification or support. Such political objects are nothing unusual in a multi-party system, especially when the whole political system is going through changes and new political entities are trying to get noticed and trying to convince voters to support them (see Cwalina and Drzewiecka, 2019). However, what is bad for small and new players can be beneficial for bigger and widely known politicians. Our results speak to the effectiveness of the ambiguity strategy with relatively new and non-polarizing issue advanced by a little known politician. However, it is possible that the strategy of ambiguity is much more beneficial when the politician is known very well, has a strong image and is supported by substantial parts of the electorate. These attributes of political communicator (strength of the image, level of support, recognition by voters, etc.) and their influence on the effectiveness of the ambiguity strategy, as well as the properties of the raised issues and attitudes toward them, should be the object of further research.

To conclude – the strategy of ambiguity seems not to be the best option when the goal is the defense of the image of the speaker or building support for him or her. However, the results obtained for influence of ambiguity on the agreement with the message seem to suggest that this strategy can be used for reaching a more important goal – mitigation of ideological conflicts in society. Ambiguous messages can be the way of preparing citizens for contact with points of view different from theirs own. This strategy could be helpful for avoiding anger of people hearing arguments they do not support and for holding back rejection of information not convenient for them.