Right-wing face, left-wing faces: The matching effect in the realm of political persuasion

Abstract

We present two studies showing that a matching effect in persuasion can be observed between message content and the shared symbolic representation of the message object. The experimental paradigm was built on the Italian political context where citizens hold asymmetrical symbolic representations about the right-wing and the left-wing (i.e., centered on the leaders in the former, and on the coalition in the latter). Since 2001, both coalitions have focused their persuasive strategies principally on candidate image. Thus, the right-wing coalition is in a communicative congruence condition, whereas the left-wing coalition is in an incongruent condition. Results showed that when the coalitions use these images in corresponding ways, they provoke in the audience an impression of message efficacy, enhancing their persuasiveness.

Keywords:

• Matching effect
• Political communication
• Message efficacy
• Persuasion

Research in the attitude domain has clearly shown that a message is more likely to have a persuasive effect on an audience when its content matches the functional or structural characteristics of the receivers’ attitudes, compared to when there is not such a correspondence. At the functional level, findings highlighted that the more the messages affect the function that is primarily fulfilled by a particular attitude, the more they are persuasive (DeBono, 1987; Lavine & Snyder, 1996; Snyder & DeBono, 1985). On the structural level, the effect of a persuasive appeal based on affect (vs cognition) is stronger when the receivers have an attitude based on affect (vs cognition) (Fabrigar & Petty, 1999). This is what has been called the “matching effect” in persuasion.

In the commercial or political domain, persuasive messages constituted solely by pictures are often employed. In this case, the image is devoted to symbolically representing the “product” to be promoted and to shed a favorable light on it. However, to the best of our knowledge, it is as yet unclear if a matching effect could be observed when the symbolic representations evoked by an iconic message correspond to the meanings characterizing the receivers’ representations of the product. Indeed, in the case of such a match, the image would facilitate a ready product categorization into a familiar system of categories (Cavazza, Serpe, Graziani, & Rubichi, in press) and in this way it would activate in the audience the positive connotation associated to familiarity. In this line, Nelson and Garst (2005) showed that the match between the values evoked in a political message and those traditionally associated with the party source of the communication promotes close attention to the message.

Our paper aims to test if a matching effect could be observed between the message content and the shared symbolic representation of the message object. To reach this goal we conducted our studies considering the Italian political domain, because it is a communicative context “naturally” characterized by persuasive messages realizing different levels of matching between their content and the shared symbolic representation of the political objects.

Political scientists recognize that a diffusive “personalization” process has affected contemporary politics in many Western countries (e.g., Wattemberg, 1995). The personalization implies a process in which the emergence of individual actors within parties and institutions is accompanied by communication strategies centered on them. This process focuses voters’ attention on the leaders’ charismatic skills rather than on the issues, and increases the influence of leaders’ perception and evaluation on vote choice.

In Italy, the personalization process of politics began considerably later compared to the USA. It seems likely that this process originated when the traditional party system suddenly disappeared at the beginning of the 1990s, because of the legal proceedings of “Tangentopoli” (“Bribetown”), which led to the indictment and prosecution for corruption of a large portion of the main majority parties. In 1993 the electoral system was transformed from proportional to majoritarian, and new political formations were generated while the old ones were eliminated. In 1994 Silvio Berlusconi founded a new party (Forza Italia) that was able, in 3 months, to collect the relative majority of ballots (see Appendix). Since then, as regards personalization, the center-right and the center-left coalitions showed an asymmetrical pattern of development.

In particular, the right-wing coalition has had the same leader (Berlusconi). In contrast, the left-wing coalition changed its leader five times in the same period. Furthermore, Berlusconi has been the first Italian political leader to apply marketing strategies to control his image and to become the focus of his party's campaign. He soon became the most visible figure in the political arena, even to those who were not particularly interested in politics.

This asymmetry is also consistent with the emphasis that left-wing and right-wing political discourse puts on different values: the former in favor of “horizontal” values (for instance, equality and loyalty) and the latter for “vertical” values (i.e., success and wealth) (Barnea & Schwartz, 1998; Oishi, Schimmack, Diener, & Suh, 1998).

These two consistent effects lead to different shared symbolic representation of the political “right” and “left”: a representation particularly centered on the leader (borrowing the Oishi et al. definitions, we can call it a “vertical representation”) in the former case, and a representation mainly centered on categories (parties, coalition, and social categories such as “the workers” or “the poor”, i.e., a “horizontal representation”) in the latter one. Nevertheless, starting from the 2001 political election campaign, the left-wing coalition increasingly focused its persuasion strategy on the image of the leader and the local candidates.

Despite the growing salience acquired by leaders’ images during election campaigns, political scientists tend to undervalue the importance of the effects induced by the use of pictures compared to other factors involved (Bailenson, Garland, Iyengar, & Yee, 2006). Several studies in political psychology (e.g., Banducci, Karp, Thrusher, & Rallings, 2008; Budesheim, & DePaola, 1994; Keating, Randall, & Kendrick, 1999) highlighted that the appearance of an individual portrayed in a photo influences how people judge him/her as a political candidate even when individuating personality information is provided. Todorov, Mandisodza, Goren, and Hall (2005) showed that, based on the candidates’ image, people go as far as to formulate inferences regarding their characteristics and particularly their competence. Furthermore, competence evaluation is significantly associated with the real level of success of the candidates in the elections.

Following this line of thought, the two Italian coalitions are in different communicative conditions. The communication of the right-wing coalition is characterized by congruence between the conveyed image and the personalization process, whereas the communication of the left-wing coalition appears as being incongruent.

Our studies tested whether and to what extent message congruency of the two coalitions influences the perception of the message's persuasive effectiveness, and its real persuasiveness. Our main hypothesis was that the match between the evoked meaning of an image and the receivers’ symbolic representation of the coalition makes the message more effective in comparison with a condition of mismatch.

Before testing our hypotheses, we checked in a qualitative pilot study if the asymmetrical personalization process is actually translated into different shared symbolic representations of the Italian political left and right.

Pilot study

Method

The pilot study was based on the verbatim analysis of 10 focus groups, attended by 44 men and 35 women aged between 18 and 70. From the political standpoint, 34 people defined themselves as left-wingers, 39 as right-wingers, and 6 as undecided. The focus groups took place in five Italian cities (Bologna, Avellino, Milano, Bergamo, and Palermo). The participants were homogeneous with regard to their educational level. In each city two focus groups were conducted: a group comprising people with only a basic level of education (corresponding to the US junior high school) and one comprising people who had attained further qualifications (high school diplomas and university degrees). However, each group was mixed with respect to the participants’ political orientation, gender, and profession.

Each session was held by the same experienced facilitator, following Krueger's (1994) method. Conversation was started around the following cue: “When people talk about politics, they often use expressions like ‘those who vote for …,’ ‘the voters of …’. We are interested in understanding what people are referring to when they use such expressions.” Each group engaged in a lively discussion and debate of these issues. The discussions lasted, on average, an hour and a half.1

The reconstruction of the Italian political field was performed by analyzing the participants’ linguistic production, since language reflects the symbolic dimensions that people refer to in order to communicate and make themselves understood (Flick, 1998; Markova, 2003). Textual analysis, in fact, proves to be a particularly valuable tool for researchers attempting to show how ordinary citizens give meaning to politics (Brugidou, 2003). The transcriptions of each focus group were preliminarily classified on the basis of the discourse target by grouping all the textual productions referring to the six main parties (three for the left: Partito Rifondazione Comunista, Democratici di Sinistra, Margherita;2 and three for the right: Lega Nord, Forza Italia, and Alleanza Nazionale) and to the two coalitions (Ulivo for the left, and CdL for the right). The resulting textual corpus was composed of 28,581 words (3928 different words) and was analyzed using the Alceste software (Reinert, 1986). Alceste is regarded as a particularly suitable tool for identifying the symbolic representation of an object through the analysis of the formal structure of a text describing it, since: (a) it reveals the thematic cores (or lexical fields) typically used to talk about it; and (b) it allows one to differentiate lexical worlds, by isolating each world's specific dictionary (Reinert, 1993, 1998)3.

Results

The text's lexical structure

Tables 1and 2 show that the most characteristic words of our lexical fields were the names of the main Italian political entities. However, the representation of the left-wing and right-wing parties and coalitions displayed by our participants differed markedly with respect to their contents. All the lexical fields concerning the right-wing coalition and parties included the names of the leaders (see Table 1). On the contrary, the lexical fields concerning the left-wing coalition and parties did not include leaders’ names. Bertinotti's name was included in the class constructed for his party (Partito Rifondazione Comunista) but with a very low rank. In the Ulivo lexical field, the nickname dark man appeared as a unique identifier for the opponents; this definition subsumed an acknowledgement of Berlusconi's protagonism (see Table 2). For the lexical fields concerning left-wing parties, strong contributions accrued from the terms that referred to these parties’ historical roots (e.g., Catholics, communists, unions) and to the coalition's political anchoring (left). However, strong references of this kind were also present in the class identified for Alleanza Nazionale (fascists, fascism).

Table 1. Lexical fields obtained for right-wing coalition and parties

CdL (right-wing coalition)	χ^2	Alleanza Nazionale	χ^2	Forza Italia	χ^2	Lega Nord	χ^2
Cdl	40.25	An	196.00	Forza Italia	101.71	Lega Nord	188.28
Television+	37.48	Fini (leader of AN)	135.79	Pension+	60.59	North	62.97
Influence/to influence	26.21	Fascist+	54.32	Million+	52.30	Leghisti (Lega supporters)	54.93
To look	19.85	To cry	44.62	Increase	44.70	People from Veneto	39.78
Interest+	19.64	Fascism	44.31	Increased	36.60	Bossi (leader of Lega)	38.67
Novelty	17.70	Young people	38.58	Berlusconi	34.12	Identity	29.75
TV soap	15.02	Order	29.14	Promise+	33.07	People from Southern Italy	24.76
Emilio Fede	15.02	Violen+	18.57	Best	19.67	Foreigners	24.05
Entrepreneur+	14.85	Nostalgia	14.11	Gip/To cheat	15.48	Taxes	23.84
Charism+	14.05	Consisten+	13.74	Vote	15.34	Educ+	19.78
Berlusconi	12.25	Famil+	11.77	Hope	12.50	Racis+	19.78
To defend	12.12	Tradition	10.56			Roma	19.45
Freedom	11.23					To reason	19.01
Success	10.14					Rich+	16.07
The χ^2 value expresses the strength of association between each word and its lexical field. The larger this value, the more important the word is for the statistical construction of the field. Emilio Fede is a very biased television journalist, showing a strong positive attitude toward Berlusconi. Veneto is a north-eastern Italian region.
Individualistic concepts in bold font and categorical concepts in italics.

Table 2. Lexical fields obtained for left-wing coalition and parties

Ulivo (left-wing coalition)	χ^2	Democratici di sinistra (Democrats)	χ^2	Margherita	χ^2	Partito Rifondazione Comunista	χ^2
Ulivo	110.31	Democratici di sinistra	198.02	Margherita	179.27	Partito della Rifondazione Comunista	122.14
Hours	43.20	Communist+	22.45	Catholic+	97.04	Union+	50.28
Experience	21.88	Modern+	19.18	Former	62.65	Sold	33.87
Left	21.19	Fight+	18.52	Italian popular party (Ppi)	56.75	Roots	31.84
Government	17.85	Local party office+	16.86	Center democratic union (Udc)	43.86	Equality	23.95
Dark man	17.78	To believe in	16.86	Christian Democrats (DC)	39.95	Collective	23.95
Crime	17.78	Symbol+	15.40	DC supporters	38.32	Juste+	23.66
Apartment+	17.78	Emilia	13.57	To renew	35.30	Communist party	15.99
Foreign affairs	17.78	Attitude	12.59	Foundations	18.42	Fight+	15.99
Work+	16.72	Young people	12.14	Secular	17.57	Opinion	14.44
		Communist Party	10.44	Origins	17.57	Bertinotti (leader of Partito Rifondazione Comunista)	13.70
		Romagna	9.55			Public	10.48
		Red+	9.42			Communist	10.28
The χ^2 value expresses the strength of association between each word and its lexical field. The larger this value, the more important the word is for the statistical construction of the field.
Emilia and Romagna are two Italian areas governed for many decades by left-wing administrations. DC was the former Democratic Christian party, which governed Italy for nearly 40 years. Its voters were labeled Democristiani (Christian-democrats). In the 1990s it split into several small parties. Among them, Ppi (Italian Popular Party) (a center-left Catholic party now part of the Margherita party) and Udc (Center Democratic Union), a center-right Catholic party.
Individualistic concepts in bold font and categorical concepts in italics.

Frequency in the use of key terms

After completing a dictionary of the terms used by our participants sorted by frequency, we isolated some “key words” referring to the “objects” structuring the Italian political field (parties’ and coalitions’ names, names of political leaders), and then we compared the frequency with which they were used. In this way we identified the political symbolic references that Italians are most familiar with, whatever their attitudes. As concerns left-wing political objects, the most quoted term in this category was the coalition's identifier (Ulivo), followed by the names of the main parties, and finally the names of their leaders (see Figure 1). When our participants talked about the right, they used the name of the coalition leader (Berlusconi) most frequently, followed by the names of the main parties composing it. The name of the coalition was quoted with a rather low frequency (see Figure 2).

Figure 1. Pilot study: Frequency in the use of left-wing objects’ names.

Figure 2. Pilot study: Frequency in the use of right-wing objects’ names.

Discussion

The lexical analysis showed the divided representation of the Italian political field, actually characterized by markedly asymmetric symbolic references: Our participants talked about right-wing and left-wing voters by referring to different objects (the leaders in the former case, the coalition in the latter one). The same was true for the more abstract left-right dimension: The term “left” emerged in reference to the Ulivo coalition, while the term “right” did not characterize the representation of the Casa della Libertà. Globally these results showed that the main elements structuring the left side of the Italian political field ranged from the more general categories (the term “left”, the name of the coalition, categorizing labels such as “communists”) to the names of the political leaders (which appeared with almost negligible frequencies). On the contrary, the right side of the political field was strongly organized around the name of the coalition leader, and much less around the names of the political entities composing it.

Given this asymmetry, we can consider it as the operationalization of one of the variables we wish to include in the matching versus mismatching conditions; that is, the shared symbolic representation of the political left and right coalitions.

Study 1

Overview

The goal of this study was to test the perceived persuasive efficacy of an image-based message of a political campaign, when it matches or not the shared symbolic representations of the two coalitions. We compared the persuasive efficacy judgments of two pictures—(a) the portrait of a man, and (b) the portrait of a heterogeneous group of people—matched with the two Italian coalition symbols: that of the center-left (which has become in the meantime the Unione) and that of the center-right coalition (Casa della Libertà). Thus, on the basis of the pilot study, we defined two conditions of congruent message (i.e., photo of the individual/right-wing symbol, and photo of the group/left-wing symbol) and two of incongruent message (i.e., photo of the group/right-wing symbol, and photo of the individual/left-wing symbol).

Respondents in congruent conditions were expected to perceive the message as more effective and convincing compared to those in incongruent ones. In other words, we predicted an interaction effect between the type of image (individual vs group) and the associated symbol (left: Unione vs right: Casa della Libertà).

Method

Participants and procedure

A total of 160 students of the University of Modena and Reggio Emilia (66 men and 94 women4) aged between 19 and 42 (M = 22.24, SD = 3.29) participated voluntarily in the study.

We operationalized the “vertical representation” of the political right as an image of a (supposed) candidate (a confederate) formally dressed and with an open smile. The “horizontal representation” of the political left was operationalized as an image of a heterogeneous group of people (found on the Web). Before the experimental session, in order to check for the appropriateness of this manipulation, we exposed each picture to 40 university students (outside the experimental sample) and we asked them to estimate the probability that the picture was adopted by the right-wing or by the left-wing coalition for a political campaign. An 11-point scale (from 0 = certainly left to 10 = certainly right) was used for the estimations. The single-person image (M = 6.60, SD = 2.26) was estimated on the “right side” of the probability dimension, whereas the group image (M = 3.95, SD = 2.28) was estimated on the “left side”, F(1, 78) = 27.25 p <.001, η ² =.26.

The congruent message conditions matched the single-person image with the right-wing coalition symbol, and the group image with the left-wing coalition symbol. For the incongruent message conditions, the match combined the images and symbols in the reverse order (single person with the symbol of the Unione and group with the symbol of the Casa della Libertà). In sum, the experimental design was comprised of two factors with two levels for each factor (type of image: individual vs group, and symbol: right-wing vs left-wing).

Participants were told they were taking part in a study testing political communication material for the next political election. They watched one photo depicting a congruent or an incongruent message condition. None of them recognized the people portrayed. After that, a short questionnaire was administered, in which participants had to formulate their evaluation of the image persuasive efficacy. Some additional demographic items (i.e., sex and age) and the political orientation question (on a 5-point scale from 1 = left-wing to 5 = right-wing) were also included as control variables. Results on political orientation were recoded as follow: left (answers 1 and 2) (n = 87), center (answers 3) (n = 22), and right (answers 4 and 5) (n = 51). Finally, the participants were thanked and debriefed.

Measure: Perceived persuasive efficacy

Participants evaluated the persuasive efficacy of the image adopted for political campaign through six items—expressive, convincing, persuasive, positive, effective, contributing to the success of the campaign—on a 10-point scale (from 1 = not at all, to 10 = very much). The items were collapsed into a unique persuasive efficacy index (α =.87).

Results

An univariate ANOVA with type of image (individual vs group) and symbol (right-wing: Casa della Libertà vs left-wing: Unione) as independent factors was performed. The hypothesized interaction effect emerged, F(1, 156) = 7.78, p <.05, η ² =.05, confirming that the right-wing individual (M = 5.55, SD = 1.79) and the left-wing group (M = 5.71, SD = 1.43)—that is, the congruent conditions—were perceived to convey a more effective message compared to the right-wing group (M = 4.70, SD = 1.52) and the left-wing individual (M = 5.17; SD = 1.51)—that is, the incongruent conditions (see Figure 3).

Figure 3. Study 1: Mean perceived persuasive efficacy as a function of kind of image and associated symbol (N = 160).

The ANOVA including the respondents’ political orientation as the third independent factor showed that it did not significantly affect the judgments, F(2, 148) = 0.53, ns, nor did it interact with type of image, F(2, 148) = 0.38, ns, and symbol, F(2, 148) = 1.52, ns. Also the three-way interaction was not significant, F(2, 148) = 0.34, ns.

Discussion

The results of the study confirmed our expectations: When there is a match between the images employed in the electoral poster and the party symbols, the message is judged as more effective and convincing than messages in which images and symbols are associated incongruently, regardless of receivers’ political orientation. More generally, when the transmitted image evokes contents matching the symbolic representation owned by the receivers, it is judged as more effective compared to when this match is not at stake.

Nevertheless, the influence that congruent versus incongruent message has on receivers’ attitude and behavior towards the two coalitions was not directly analyzed in this study. It seems reasonable to test whether the positive evaluation of the congruent messages influences receivers’ behavior (i.e., voting choice) as desired by the source of the message. We carried out this test in Study 2.

Study 2

Overview

Study 1 focused on the receivers’ perception of the persuasive effect of an iconic message. The goal of the present study was to test the actual persuasiveness of congruent–incongruent political communications based on visual imagery in influencing receivers’ behaviors. Specifically, we hypothesized that when receivers are required to vote choosing among previously unknown parties only on the basis of election posters, they will tend to choose the parties represented in a congruent way (i.e., whose image evokes contents in line with the shared representation of the political left and right) more than in an incongruent way. In other words, if voting for a left-wing party Italian participants would prefer a party whose message is conveyed by the image of a heterogeneous group of people, more than by the image of a single candidate. On the contrary, if voting for a right-wing party, receivers would prefer a party represented by a single candidate image over one using the image of a heterogeneous group of people.

Method

Participants and procedure

A total of 254 adults agreed to participate in the study (125 men and 129 women). They were recruited during summertime 2007 in several beach complexes of an Italian tourist city on the Adriatic Sea, which is a massive holiday destination. They were aged between 18 and 68 (M = 27.47, SD = 8.50). The study was presented as a research project on voting decision-making processes. Each participant received a three-page questionnaire. On the first page they were asked to read the following mock election scenario:

Imagine that you have just moved into a new town. As soon as you arrived, you realize that in few days there will be the local elections, and you are required to vote. As a newcomer in town, you do not know anything about the political situation. The only thing you get to know from the billboards placed along the streets is that four parties are in competition: two of the right-wing and two of the left-wing coalition.

On the second page, the election posters of the four parties (called party X, Y, W, Z) were shown and participants had to vote for one of them. We employed the same photos from the first study along with two other photos also portraying a heterogeneous group of people (found on the Web) and another smiling man with a similar formal look (another confederate). A single-person image (supposedly a candidate) and an image of a heterogeneous group of people were attributed to each of the two coalitions (left-wing and right-wing). Thus, for both coalitions participants had the choice between a congruent message (i.e., single person/right-wing coalition and heterogeneous group/left-wing coalition) and an incongruent message (i.e., heterogeneous group/right-wing coalition and single person/left-wing coalition). As in the previous study, participants were asked if they had ever seen the pictures before. None of them recognized the people portrayed.

Sex, age, and political orientation items were included on the last page of the questionnaire.

The answers on political orientation (on a 10-point scale from 1 = left-wing to 10 = right-wing) were recoded in three categories: left (answers from 1 to 4) (n = 120), center (answer 5 and 6) (n = 51), and right (answers from 7 to 10) (n = 79); four participants did not answer the question. Finally, participants were thanked and debriefed.

Measure: Voting choice

The critical dependent variable was the choice of preferred party. After reading the scenario and observing the photos, participants were asked to “vote” for the party they preferred. Within each coalition two pictures were presented: one congruent (single person and right-wing; heterogeneous group and left-wing) and one incongruent (single person and left-wing; heterogeneous group and right-wing) with the political representation of the two coalitions. The presentation order of the photos was balanced across the two political coalitions.

Results

As hypothesized, the chi-squared test performed on the contingency table crossing the type of party image (single person vs group) and the party coalition (left-wing vs right-wing) showed a significant relationship between the two variables, χ²(1, N = 254) = 17.34, p <.001. In particular, the results highlighted that the party portrayed by means of a single candidate was chosen more often when it was on the right (19.90%) than when it was on the left (9.84%), whereas the reverse was true when the parties employed images of the group (right-wing group 26.38% vs 44.88%, see Figure 4).

Figure 4. Study 2. Percentage distribution of choice as a function of party political sides and kind of image.

However, we should note that the political parties represented by the groups were generally more preferred than the ones depicting the single individuals: Of the total, 181 voted for the lists portraying the group, and only 73 participants expressed their preference for the lists picturing the single person. This distribution of frequencies proved to be significantly different from the balanced distribution at the chi-squared test, χ²(1, N = 254) = 45.92, p <.001.

In order to control for the effect of participants’ political orientation, we conducted a log-linear analysis: 2 (party coalition: left-wing vs right-wing) × 2 (type of image: single-person vs heterogeneous group) × 3 (respondents’ political orientation: left-wing vs center vs right-wing). The analysis showed a significant relationship with the party coalition: as one would expect, left-wing participants highly preferred the left-wing parties (89.16% of left-wingers voted for one of the left-wing parties), whereas right-wing participants mainly chose a right-wing party (89.87% of right-wingers voted for one of the right-wing parties) χ²(2, N = 250) = 128.34, p <.001. With regard to the participants belonging to the center, no significant differences were found in terms of preference expressed toward the parties of the two coalitions (43.13% chose one of the left-wing parties and 56.84% chose one of the right-wing parties). The analysis also showed a significant relationship between the participants’ political orientation and the type of image, χ²(2, N = 250) = 8.36, p <.02, due particularly to the choices of left-wing and center participants, who highly preferred the group images (respectively 79.17% and 82.35%), whereas the right-wing participants balanced their ballot more (51.90% preferred the group images). Finally, even if the most frequent choices were observed in the two conditions of congruent symbolic representation, party orientation and respondent orientation (i.e., group image/left-wing party/left-wing respondents, and individual image/right-wing party/right-wing respondents), the three-way interaction failed to reach statistical significance, χ²(2, N = 250) = 2.71, p =.26, ns. This result might be due to the low number of observations in some cells, and for this reason we refrain from interpreting this finding.

Discussion

The results of Study 2 showed that an iconic message is more likely to have a persuasive effect on receivers’ behavior when the image evokes meanings congruent with the shared symbolic representation of the advertised “product.” When receivers are asked to vote for a party on the basis of the electoral posters in a fictitious scenario in the political domain, they tend to prefer “congruent” messages: For Italian participants, examples of such congruent messages are images of heterogeneous group of people for the left-wing parties, and portraits of a single candidate when belonging to the right-wing coalition. The results of the log-linear analysis also involving respondents’ political orientation showed that, in addition to the matching effect between the shared representation of the parties and their message contents, the left-oriented respondents highly preferred the group image. This finding may be due to the fact that a photo portraying a group of people who differ one from the other is likely to trigger a reference to the value of “universalism” in the observers, and the research by Caprara, Schwartz, Capanna, Vecchione, and Barbaranelli (2006) showed that “universalism” is a value normally perceived as more relevant by left-wing voters than by right-wing voters.

It is noteworthy that the context of the judgmental task of Study 2 was quite different from the one of Study 1. In the first study we asked for an absolute judgment of just one photo, whereas in the second study participants performed a comparative judgmental task. Notwithstanding this variation that could induce different information processing styles (Riggle, 1992), converging findings were observed.

Unexpectedly, we found a spread tendency to prefer parties represented by a group of people. These results seem completely in contradiction with the trend of personalization in politics. On the contrary, we believe that it may be due to the specific experimental stimuli we used: the two pictures of groups seemed to have a better aesthetic quality than the ones of the single persons. To test if a possible bias was induced by the different aesthetic quality of the pictures, we carried out a follow-up in which we asked to 30 students of the University of Modena and Reggio Emilia (13 men and 17 women aged between 19 and 43) to judge the quality of the four photographs. Participants formulated their judgments on six items of bipolar adjectives (pleasant/unpleasant; good looking/bad looking; positive/negative; attractive/unattractive; effective/ineffective; nice/nasty). Items were collapsed into an aesthetic quality index (all αs comprised between.82 and.92). Results of the ANOVA with a repeated factor confirmed that the two group pictures were judged as having a better aesthetic quality (M _group1 = 5.78, SD = 0.85; M _group2 = 5.29, SD = 1.13) than the two single-person photos (M _person1 = 4.47, SD = 1.02; M _person2 = 4.34, SD = 1.03) F(3, 27) = 23.83, p <.001.

In sum, our results are in line with those relative to the matching effect in persuasion and take them even further. Indeed, studies aimed at testing the matching effect have focused above all on the correspondence between the characteristics of the message and specific attitudes. However, it was not clear if the same effect could be obtained through correspondence between the symbolic representations evoked by the message and those that characterize the receiver's own representations. Moreover, in previous studies the communicative content used in the experimental paradigm was forwarded through a linguistic code, but persuasive communication is becoming more and more frequently entrusted, either principally or entirely, to visual imagery. In our study the message was put forward through visual imagery referring to representations of Italian political coalitions. As Study 1 clearly showed, the correspondence between the meaning evoked by an image and the symbolic representation owned by the audience induces positive evaluation on persuasive efficacy. Importantly, in the present study we demonstrated the actual persuasive benefits deriving from the congruent images.

General discussion

The results of our studies converged in showing that when the two coalitions use images in a congruent way, they induce in the observers a positive evaluation of persuasive efficacy and they are more likely to influence voting choices. Moreover, the findings were observed involving both university students and adults as participants, thus striving to bypass the student sample bias (Meloen, 1993).

We explained these findings by considering that in Italy the personalization of politics is an asymmetric process, which involves the right-wing coalition more than the left-wing one. More particularly, personalization of political communication was introduced in Italy by Berlusconi in 1994, while the left-wing coalition did not take care of its own persuasive strategies to the same extent. In the following years it became increasingly clear that the left-wing coalition adopted communicative methods very similar to those of their competitors, in particular focusing their messages on the figures of the leader and the candidates, although still today the process of political personalization of the left-wing coalition continues to be much less apparent compared to that of the right-wing one.

Given that both studies were run when the prime minister was supported by the center-left coalition, we believe that the effect observed does not depend on who is currently in power and thus receives greater visibility and media coverage (focusing on the person rather than the parties). Indeed, notwithstanding the prime minister's greater visibility, people preferred a collective image of the left. Therefore, it seems that media coverage and the power role did not influence the results.

Unexpectedly, a generic preference for collective images in political communication emerged. We showed that this was due, at least in part, to the fact that the collective images we used were more aesthetically pleasing. However, we cannot exclude that collective images are generally more pleasing as political messages. Indeed, politics is a collective activity and ideally directed to the advantage of collectivity. Thus our participants might have perceived groups as more appropriate for politics than individuals. Therefore, notwithstanding the personalization trend in European politics, and in particular in the Italian political right, our findings suggested that the popular image of the left is solidly collective, whereas the image of the right is probably more complex, with both individual and collective elements.

One limitation of the experimental paradigm used in the two studies may be that the stimuli employed are quite “minimalist,” since participants simply saw pictures without any context. Obviously, in a more naturalistic frame they collect a set of more complex information. For example, we considered candidates unknown to the receivers. It cannot be excluded that, in case of known candidates, other characteristics (e.g., whether a politician's popular image is positive or negative, or the candidate's gender) are more influential than the matching effect. Nevertheless, our results might apply to the many circumstances in which unknown candidates are really at stake. When the electoral system allows people to vote both for a party and for preferred candidates, candidates have to gain these preferences with their own campaign: for example, this is the case for the great majority of candidates to the European Parliament, and to the local administrations.

Furthermore, in Study 2 we forced our participants to choose among four voting alternatives. In the real world the range of behavioral options may be wider: one can also decide not to vote, for example when the parties of the preferred coalition adopt only mismatching communication strategies. Thus, with the present data we cannot definitely say if the mismatch effect is powerful enough to induce voters to actually switch to a different candidate or simply to favor their abstention. However, we can speculate that it depends on the “ideological climate” characterizing the environment of the political decision. In low ideological contexts voters may be significantly influenced by communication effects like the one documented here, and switch toward the opposing parties adopting a congruent communication; whereas in highly ideological context, such as the Italian political field, mismatching messages may facilitate the choice to abstain. This question remains to be investigated empirically.

Future research should also examine whether the “content matching effect” we found can be generalized to other communicative contexts, carrying out studies in: (a) more ecologically valid settings (e.g., with real and known candidates, or outside the political field), and (b) more controlled environments, in which experimenter is able to directly manipulate the symbolic representation of the message object. Furthermore, we have yet to deeply understand why the kind of matching effect we observed does work. Our suggestion is that congruent messages may facilitate the categorization task and benefit from familiarity.

From a theoretical point of view, the present findings are in line with those of Han and Shavitt (1994), showing that the persuasive effect of individualistic versus collectivistic focused messages is affected by the match with the general individualistic versus collectivistic orientation of the communicative context. We went further by expanding the field of application of the matching effect in persuasion to the area of iconic messages and to the political field.

From an applied point of view, the present data suggest the creation of persuasive campaigns corresponding to the conditions of the political coalition (salience of the leader, salience of the parties and the coalition) and to the symbolic representation that citizens have developed of them. Our results support the idea that an effective communicative strategy cannot be applied to a variety of contexts without an attentive preliminary analysis of those contexts.

It seems that in Italy the left-wing coalition has underestimated the potential effect of messages that are in line with the general trend of personalization politics but incongruent with respect to the popular representation of the coalition. In fact, in 1994 Berlusconi introduced the use of large billboards with his image as the focus of the persuasive message, whereas the portrait of the opposing leader (Achille Occhetto) did not appear. In 2001, the left-wing leader (Francesco Rutelli) adopted the “Berlusconian” strategy by diffusing his pictures. In the last political election (2008), Berlusconi decided to change strategy and did not appear in the campaign material, granting a larger space to (and allowing voters to familiarize themselves with) the symbol of his new coalition (Popolo della libertà), whereas the picture of the left-wing leader (Walter Veltroni) still was the main visual element of the left's campaign material. Political communication that focused on the figure and image of Berlusconi doubtless contributed to the success of the right-wing coalition in Italy. To the best of our knowledge, research studies empirically documenting the comparative effectiveness of these communicative strategies were lacking. Thus, we cannot say if the left-wing leaders and parties had reason to believe that personalized campaigns would be effective for them, but our results showed that simply imitating the strategies of the right is not sufficient to guarantee success.

Notes

¹ The meetings were held in small classrooms or meeting rooms, where a large board displaying all the Italian political party symbols was stuck on the wall behind the moderator. Participation was rewarded with 30 euros.

² The focus groups were held before the merging of Democratici di Sinistra and Margherita into the Democratic Party (2007).

³ Alceste divides the textual corpus into homogeneous elementary contextual units (hereinafter UCE), which approximately match the size of a sentence, based on their lexical profile. Only full words (nouns, verbs, adjectives, and adverbs) are taken into consideration in the analysis, after reducing them to their root. The software cross-tabulates full words with the UCEs, generating a square distance matrix, in which the distance between two UCEs is a function of the number of words they share. A descending hierarchical classification is then performed and classes of words (called lexical fields), which co-occur and are best differentiated from the other classes, are extracted. The strength of association between each word and its class is expressed by the chi-squared value (Bauer & Gaskell, 2000). Once the lexical fields have been identified, we examined their associations with the variables that do not participate in their computation, specifically participants’ educational level and the discussion target. Since education was weakly associated with our lexical fields, it has not been considered in this paper.

⁴ The effect of gender was tested in all the analyses: It did not produce any significant differences on research critical variables.

Appendix

The Italian political situation

Italian politics throughout the post-war period, and up until the 1980s, was always extremely ideological. Italy was the European country with the largest Communist Party and the largest Catholic Party, and featured strong party identification, especially among Communist and Christian-Democratic voters. At the beginning of the 1990s, however, the Italian party system quite suddenly experienced a process of complete destruction. In fact, almost all of the traditional parties that had given stability to the political system for half a century disappeared from the political scene (they either simply dissolved or took on new names). The process was triggered by the legal proceedings of “Tangentopoli” (“Bribetown”), which led to the indictment and prosecution for corruption of a significant portion of the two main majority parties (Christian Democrats and Socialist Party). The system's destruction was greatly accelerated by the fall of the Berlin wall, which neutralized the ideological foundation of the communism/anti-communism opposition. Moreover, Christian Democrats, who had governed the country without interruption for the entire post-war period, experienced a fatal crisis. The party system finally came to a formal end in 1993 when the electoral law transformed the electoral system from proportional to majoritarian, generating new political formations and eliminating old ones.

In January 1994, television tycoon Silvio Berlusconi created a party (Forza Italia) allied with Alleanza Nazionale, an originally neo-fascist party that converted to democracy, Lega Nord, a party with its roots in the rich and affluent north of Italy, and Unione Democratica di Centro, a Catholic center-right party. Since then, Berlusconi has been the center-right coalition leader for 14 years and prime minister three times: 1994–1996; 2001–2006; 2008–the present. The center-right coalition took the names of: Polo della Libertà (Pole of Freedom), then Casa delle Libertà (House of Freedom), and now Popolo della Libertà (People of Freedom).

In 2001 the main opposition parties were joined in the center-left Ulivo (Olive Tree) coalition, whose leader was Romano Prodi, prime minister between May 1996 and October 1998. The coalition included: (a) Democratici di Sinistra, a left-wing party that moved away from communism in 1991; (b) Margherita, a Catholic center-left party; and (c) Verdi, an ecologist party. To the left of Ulivo, there was Partito della Rifondazione Comunista, the largest and most radical Italian communist party, whose leader was Fausto Bertinotti. The center-left coalition was renamed Unione in 2005. The Unione won the political election in 2006 and its leader, Romano Prodi, became prime minister for the second time until April 2008. In this time frame Democratici di Sinistra and Margherita merged into the Partito Democratico (Democratic Party).