1
A study on the influence of using an infographic
versus a text on attitude and voting intention
Carlotte Teunissen – 10572902
Afstudeerproject Politieke Communicatie en Journalistiek Communicatiewetenschap
Universiteit van Amsterdam Sanne Kruikemeier
05-06-2016 Word count: 7000
2 Abstract
This study investigates the effect of presenting information as an infographic (versus as a text) on attitude towards a party and voting intention. Also, this research aims to determine whether being exposed to an infographic (vs. to textual information) has a more positive effect on voting intention for a certain party when someone has a better attitude towards that party. The goal is to close the gap that currently exists in the research on the effects of infographics in the political communication and to advise developers of political campaigns how to create their campaigns possibly different in order for these to be more effective. First, a pre-test was held to determine which topic was most suitable to use in the information and secondly, an experiment consisting of two conditions was held. Half of the participants (N = 150) looked at an infographic displaying information about D66 (a neutral, Dutch party) and their cost reductions and the other half viewed the text with approximately the same content. All
information that was shown was in Dutch. Support of non-support was found that information presented as an infographic positively affects attitude towards a party and voting intention more than when that same information is designed as a text. Furthermore, it could not be confirmed that higher levels for attitude towards a party increased the positive effect of presenting information as an infographic on the voting intention.
Keywords: voting intention, attitude, infographic, experiment, text
Introduction
The last couple of years it has become increasingly difficult for parties to keep their voters committed to them, especially when it comes to those with low levels of political knowledge (Claassen, 2007). For example, research has shown that in the last two months of the
3
certain yet who they would vote for, while another seventeen percent mentioned that the party they were intending to vote for could still change before they would actually cast their vote (Lundberg & Payne, 2014). These citizens, also known as ‘floating voters’, can be recognised by their lack of attachment to a specific party (Van der Brug, 1999).
Due to this disloyalty political campaigns are now an even more important resource for political parties to attract voters. Therefore, they are looking for new ways to do so. One way to do this is by making party information more easily understandable. In recent years, applied researchers have for example shown that illustrations can be effective for this.
Lewalter (2003) states that text supported by illustrations serves as a stimulus for the learning process. It has also been proven that illustrations increase the recall of the presented
information by the viewers (Kools, van de Wiel, Ruiter, & Kok, 2006).
A form of illustrations that is often used in politics is the infographic (Çifçi, 2016). Infographics are an unusual form of illustrations. Often they are presented vertically and they are essentially, just like the word itself proposes, a composition of ‘information’ and
‘graphics’ (Thompson, 2015). Bouquin and Brown Epstein (2015) explain how the use of infographics enables the audience to understand certain information better by displaying it graphically. According to Krauss (2012) infographics also make it easier to apply and recall the information people are exposed to, because of the fact that they contain both visuals and text. Vanichvasin (2013) found similar results and in addition, his study showed that
infographics stimulate appeal and retention. In summary, infographics are beneficial for a better understanding, recall, appeal, retention and the applying of the indicated information. Furthermore, implementing these illustrations is not only popular in politics but also in other professions. In the field of health communication, researchers found that infographics make it easier for people to understand their own medical records (Arcia et al., 2015). In journalism infographics are also used to explain complex topics better and to publish so called ‘big data’
4
(Smit, de Haan, & Buijs, 2014). Moreover, this type of visuals is considered to be in general the most efficient digital form of communicating information (Smiciklas, 2012). However, it must be taken into account that infographics are not the key to achieving success: the impact depends on its’ quality and the way the content is presented (Kos & Sims, 2014).
Despite this popularity of infographics, research on the effect of using them in political communication is unfortunately scarce. Most of the studies focus either on using textual information in the most effective way or on the use of infographics in other fields (of communication). It is safe to say that we do not know much about using infographics in political communication yet and that more research is desired. The exact same goes for research that focuses on both attitude towards a party and voting intention. In addition, many studies in political communication can be found that focus on these three topics separately but none combine the effect of infographics, attitude towards a party and voting intention in one study. The objective of the current research is to put these topics together to expand the scientific knowledge and to close this gap of information. This can be seen as important, due to attitude (towards a party) and voting intention being important motivations for someones’ voting behaviour (Lundberg & Payne, 2014). Since voting behaviour is an important facet for political parties, tools that help influence this are definitely worthy of investigating.
Furthermore, textual information has been used in political campaigns for many years already, while infographics are relatively new and its’ possibilities have so far not yet been explored widely enough. New studies might gain more insight into whether using infographics makes these campaigns more effective.
Carrying out a research on this topic can thus be of great value, because it could potentially be helpful for developers of political campaigns and therefore also for the political parties they are recruited by. If infographics would appear to be more effective than using text for influencing the voting intention of citizens and/or their attitude towards a party it is wise
5
for parties to use this type of visual aspects for their future campaigns. Especially with the increasing amount of floating voters nowadays this is of great value to political parties in their preparations for upcoming elections, as it might offer them a way to enlarge their chances of attracting voters by using infographics. For these reasons, the aim of this study is to
understand the effect of the way information is presented on someones’ attitude towards a party and through that the effect on someones’ voting intention. Thus, the following research question will be examined:
To what extent does the way information is presented, either as a text or as an infographic, affect someones' attitude towards a political party and voting intention towards that party.
To make sure that the relationships between the variables of the research question is clear, the conceptual model below was added:
In the next paragraph, the theoretical framework, the hypotheses for the research question will be explained. To understand the relationship between the concepts multiple theories and articles will then be mentioned. These also support the hypotheses. For example, the Elaboration Likelihood Model (Petty & Cacioppo, 1986) will be explained, as well as the Value-Attitude-Behaviour model (Vaske & Donnelly, 2010). Furthermore, a pre-test will be held to determine which political topic will be most suitable to use in the experiment. After deciding on this subject, the actual experiment will follow. Analyses will then show whether the hypotheses mentioned in the next paragraph are valid, which will be described in the conclusion. Finally, the implications and limitations of this study will be discussed.
6 Theoretical Framework
The way information is presented
It has become increasingly important for our daily lives to discover how people process information. Researchers have therefore become more interested in examining this as well and a great example of one of those researchers that has spent her time on this is Annie Lang. In one of her papers she describes the Limited Capacity Model (Lang, 2000). This theory implies that people do not have a never ending capacity of storing new information; there is a limit to the amount of impressions people can process at once or in a short time of period. It is stated that the extent to which people are motivated to process information depends on whether they have enough opportunity to do so and whether the way the information is presented triggers them to absorb it. Also, she explains how most information is processed automatically by the recipient, since people do not have enough capacity to process much information through the energy consuming central route. However, this is not what, for example, the people who create political campaigns strive for. They aim for their parties’ goals to be recalled and for the campaign to have a significant impact. Thus, it is important for them to find ways to present their information in such a way that the voters are triggered to absorb it and they can be affected by it.
The Elaboration Likelihood model describes how the automatic (also known as heuristic) processing of information is actually a very superficial way (Petty & Cacioppo, 1986) of processing. This means that people who use the heuristic route when they are informed about something do not necessarily remember the content (for very long)
afterwards. Most importantly, researchers found that due to that trivial way of processing it is unlikely that peoples’ intentions or attitudes will be changed. It therefore makes sense that the creators of political campaigns mostly strive for controlled processing (Lang, 2000) or
7
processing via the central route (Petty & Cacioppo, 1986). This way their audience will be easier to influence and they will remember and understand the information better.
Further research has found that in order for people to understand information
processing through that central route is a must. Using visuals has appeared to be effective to achieve this (Kools, van de Wiel, Ruiter, & Kok, 2006). Bouquin and Brown Epstein (2015) also describe in their research that if people aim for their audience to understand certain information better, the use of infographics could potentially be a great tool for this. They also noted that using infographics is especially effective for marketing purposes. This means that it is possibly also very suitable for political campaign developers to use this type of illustrations, as it happens that political campaigns can actually be seen as marketing campaigns (Soberman & Sadoulet, 2007).
As I mentioned before, not a lot of research on the impact of using illustrations or infographics has been done in the field of political communication. However, in health communication it has been examined more often. Due to the fact that the field of political communication shows similarities to other forms of communication, theories developed by, for example, health communication researchers are more than once applied to political communication (Lang, 2006). Arcia et al (2016) concluded that infographics appeared to be more appealing than text. Moreover, using them turned out to be helpful for people to understand their own medical records better and they helped increase how engaged people were to the topic. When it comes to learning from television news, using visuals was also found to be contributing (Graber, 1990). Also, Park, Jeon and Sullivan (2014) concluded in their study that using visuals has a positive effect on brand attitude and consumer behaviour, such as purchase intention. Due to this, the first hypothesis of this study is:
Hypothesis 1: An infographic (vs. textual information) leads to more positive attitudes towards a political party and higher voting intentions.
8 Attitude towards a party and voting intention
Up until now, we have talked multiple times in this study about what political campaigners aim for. It is also important to take into account for what reasons these campaigns are created. Naturally the main goal for political parties is to attract people to vote for them. However, the act of actually voting is part of a process, during which the attitude towards a party and voting intention play a major role. Therefore, these two aspects will be examined while being
influenced by the way the political information is shaped (as an infographic or as a text). A theory that focuses on these two political facets is the Theory of Reasoned Action. The founders of this theory, Fishbein and Ajzen, found that someones’ behaviour can be predicted by his or her intention to engage in that behaviour (Singh, Leong, Tan, & Wong, 1995). They also came to the conclusion that attitude towards a party is the principal, though not the only, motivation for intention. This means that attitude influences intention and that intention determines behaviour (Himmelweit, Humphreys, & Jaeger, 1984).
Other researchers, Vaske and Donnelly (2010), have written about a model that largely supports the Theory of Reasoned Action; the Value-Attitude-Behaviour Model. It describes how attitude and norms influence behavioural intentions and how behavioural intentions eventually influence behaviour, for example voting (Echabe & Garate, 1994). They also mention that both attitude and intention are prone to change, for example much more than values are. This can be seen as an important finding for the field of political communication, as it means that finding out how attitude (towards a party) and (voting) intention can be changed most effectively can be of great value to the process of the development of political campaigns.
Roccato and Zogmaister (2010) describe the Italian elections of 2001 and the way the Implicit Association Test was used. This is a tool to measure attitudes, which appeared to be correlated with voting behaviour. As the previous mentioned models, such as the Theory of
9
Reasoned Action and the Value-Attitude-Behaviour Model, describe, it is considered that attitude influences voting intention, which in turn influences voting behaviour. Therefore, we could conclude that the Implicit Association Test is also a suitable measurement to indicate someones’ voting intention, once again proving that attitude towards a party affects voting intention. Thus, if someone has a positive attitude towards a certain party it is more likely that he or she has the intention to vote for this party than when he or she has a negative attitude towards that party.
Furthermore, Powell, Boomgaarden, de Swert and de Vreese (2015) discovered with their study that using images is the most effective for changing behaviour, while using text is considered the best option for changing opinions and therefore attitudes. This means that if you would want to influence both attitude (towards a party) and (voting) intention by using one tool or way, using an infographic and therefore combining the best of both worlds would be the most efficient thing to do. Due to all these findings, the second hypothesis of this study is:
Hypothesis 2: Exposure to an infographic (vs. to a text) has a positive effect on voting intention towards a party through higher levels of attitude towards that party.
Method
When it came to picking a design for the current research I had to keep in mind what exactly I was going to investigate. Attitude (towards a party) and voting intention are two variables that are not easy to examine, since people themselves often already have trouble explaining theirs’ explicitly when you directly ask them. Also, it was an absolute must that the results would be applicable to all political parties and political communication. Otherwise the results of this study would only be relevant for the political party that was used. Randomisation would be very beneficial for this. Furthermore, I wanted to make sure that the only thing influencing the
10
attitude and voting intention could be the way the information was presented. For this reason, the research would have to take place in a controlled environment to make sure the internal validity would be high. Since I examined the difference in the (positive) effect of using an infographic in comparison to using a text I needed both an experimental group and a sort of control group (the text). Finally, the most important goal was to find causal correlations, so that relevant conclusions could be drawn. Due to all these reasons, it was clear that in order to be able to answer the research question correctly executing a true experiment would be the most appropriate.
Participants and sample
This research has been carried out in collaboration with two other researchers. Because of this, and keeping the standard demands of fifty participants per researcher in mind, we strived to use at least 150 Dutch participants for this research. That goal was achieved. In order to reach enough potential participants we have used a convenience sample, meaning that the greater part was approached by email or through the social medium Facebook. The
participants answered questions about multiple aspects, such as recall and political interest. However, only the questions that were relevant for the current study were used in the upcoming analyses. Hence, from now on only the questions about attitude towards the party and voting intention will be mentioned.
There is not necessarily aimed at a specific age group. Nonetheless, most of them were between eighteen and thirty years old. This appeared to be in line with the expectations, since most of the participants were Facebook-friends of the researchers, who are students. When it was noticed that a participant under the age of eighteen had taken part in our experiment he or she was removed from the data file and not included in further analysis. There are two reasons for this. First of all, minors do not have the right to vote, which means that they are not very
11
relevant for this research on voting intention. Secondly, children under 18 need their parents to give permission for their participation and that was not possible (to check) with the online convenience sample in this study.
Research design
This experiment has a single factor between-subjects design with two levels, namely ‘infographic’ and ‘text’. The ‘infographic’ condition is the experimental condition and the ‘text’ condition is the control condition. After being randomly assigned (through setting this up in the online survey program Qualtrics) to one of the two conditions participants have first answer questions, then were exposed to the manipulation and then answered questions again. Finally, a debriefing was held, explaining the participants that the information they had just processed should not be considered as the truth. This was to prevent the participants from possibly being harmed by the information.
As mentioned before a true experiment was performed, since participants were randomly assigned to one of the two conditions. This was crucial for making sure that the results can be generalised to a larger population than just the participants used. Thus, this improved the external validity of the research. A between-subjects design has been chosen, which means that all of the participants were exposed to only one of the two conditions and not to both. Due to this, there is no chance of order or test effects. Also, when participants are exposed to both conditions it becomes more likely that they figure out what exactly is being researched. This might influence their answers and this is in any case harmful for the internal validity of the research.
Procedure
12
the manipulation. Among these were questions about demographic information, such as age, gender and level of completed education. For example, ‘What is the highest level of education that you have completed?’ was asked. Furthermore, a question about voting intention was asked, ‘How likely are you to vote for D66 in the future?’, which could be answered by means of a 7-point-Likert scale. They were also asked how likely they were to vote for the other large Dutch parties. However, this was solely for the purpose of making sure that they would not be primed before the manipulation would take place. Therefore, their answers to these questions were not used in further analysis. After answering these first few questions, they were exposed to the manipulation and afterwards they answered more questions, about voting intention and attitude towards the party. For example, they were asked again how likely they were to vote for D66 in the future, which was also answered by way of a 7-point-Likert scale. The last question was a manipulation check, asking whether the participant had seen an infographic or a regular text. After answering this last question, the manipulation check, a debriefing took place.
Pre-test
To enlarge the chances of finding relevant results it was important that the right stimulus material was used for the experiment. Therefore, before the actual research took place, a pre-test was held with twenty participants. Also during this small experiment the participants were selected by a convenience sample. Participants were not checked for demographic information or other aspects, because it was not of any relevance in this case. During this pre-test
participants were exposed to multiple infographics, each about a different political related topic. Afterwards they answered the following question: ‘How likely do you think it is that the infographic that you have just seen will influence your opinion about the topic in the
13
infographic that was considered as most influential for changing their opinion or attitude and thus the one with the highest mean was used during the actual experiment. Therefore, the choice was made to use the infographic about D66’s plans for cutting down expenses towards Europe-related topics (M = 3.95, SD = 1.54). On the next page this infographic is shown, as well as the corresponding text.
Experimental stimuli
In the experiment two different groups of people were compared; participants who were exposed to the message processed as a ‘text’ and participants who were exposed to the message processed in an ‘infographic’. The online survey program Qualtrics assigned all of them randomly to one of both conditions. Two different kinds of stimuli and therefore two conditions were used: an infographic, the experimental condition, and the (regular) text, the control condition. Both stimuli contained the same message about the same topic, so that the content of the message could not influence the results of the experiment. This benefited the internal validity. The only difference between the conditions was the way the message is presented. Based on the topic that was chosen after performing the pre-test, the stimuli looked as follows:
14
As the infographic and text demonstrate D66 was chosen as the presented party. This decision was made for multiple reasons. First of all, the experiment was carried out in The Netherlands with Dutch speaking voters being the participants. These people will most likely know more about Dutch politics than about any other countries’ politics, making it a logical choice to use a Dutch party as the subject of the stimulus material. This also means that the questions in the survey and the stimulus material were in Dutch. Secondly, it was important that the party could not be linked very specifically to either the left or right wing. The party had to be as neutral as possible, so that the attitudes that the participants had towards the party, prior to the experiment, would limitedly influence their answers. Choosing a very right winged party like the PVV could for example cause such strong resentment for voters of very left winged parties that it could negative influence the way they responded to the experiment. Since it was desired for the internal validity that religion would not become a confounding variable, the neutral parties like CDA, CU and SGP could also not be used, leaving D66 as the ideal choice.
Then, in order to check whether the manipulation succeeded, a manipulation check was executed. Due to the two ordinal variables that were used, ‘the way the information was presented according to the perception of the participant’ and’ the way the information was actually presented’, a crosstabs table was conducted. After performing this, the Chi-square could be determined; X² = 111.10, p < .05. This meant that the two variables together appeared to have a significant result. Thus, the conclusion could be drawn that the manipulation succeeded.
Data analysis
One independent variable was analysed; ‘presentation of the message’ (infographic vs text). Because of the fact that one, categorical, independent variable and two dependent variables at
15
interval level were measured a mediation analysis was carried out. First, a randomization check and a manipulation check were performed. Then a factor analysis and a reliability analysis followed, prior to doing the actual mediation analysis, to test the coherence between the variables and after this a scale was created.
Measures
Four different statements were answered by the participants that together test for the
mediating variable ‘attitude towards D66’. These statements asked for their attitude towards D66, for the extent to which they found D66 credible, whether they thought D66 had good party statements or not and whether they liked D66 or not. Before a scale could be created with these four items, a factor analysis and reliability analysis were performed. Creating a new scale was desirable, because the scores of the participants on the new scale would be more reliable than their scores on the separate variables/items.
The principal component factor analysis with Varimax rotation helped determine whether it was allowed to create the new scale. One factor was used and the total variance explained was 67.49%. There was only one component with an eigenvalue of 1.00 or higher; in this case 2.699. In addition, the scree plot also showed only one component on the left of the turning point. The component values of the four items were all positive and higher than .45, as table 1 below shows. This means that all four items measured correctly for the mediating variable ‘attitude towards D66’.
16 Table 1
Component Matrixa
Component 1 After seeing this message, my attitude towards D66
is… - Negative : Positive .829
After seeing this message, I think that D66 is… -
Unreliable – Reliable .822
After seeing this message, I think that D66 has… -
Bad party statements : Good party statements .820 After seeing this message, I think that D66 is… -
Not nice : Nice .816
Note: Extraction Method: Principal Component Analysis. a. 1 components extracted.
The reliability analysis checked the internal consistency of the items and thus whether the four items were reliable enough to construct a new scale (‘AttitudeD66’) of. First of all, the analysis showed a Cronbach’s alpha of .83. Keeping in mind that a Cronbach’s alpha of .80 would be sufficient, the conclusion could be drawn that the analysis appeared reliable. Also, the Cronbach’s Alpha could not be improved by deleting one of the variables, as Table 2 shows. There were no negative correlations between the items and also none of the items would become more reliable than the initial Cronbach’s alpha if it would be deleted. This meant I could carry on with further analyses and the new scale ‘AttitudeD66’ (M = 4.51; SD = .91) was created, with a minimum score of 2.00 and a maximum score of 7.00 Table 2
Item-Total Statistics
Cronbach’s Alpha if Item Deleted After seeing this message, I think that D66 is… -
Unreliable – Reliable .789
After seeing this message, I think that D66 is… -
Not nice : Nice .798
After seeing this message, I think that D66 has… -
Bad party statements : Good party statements .791 After seeing this message, my attitude towards D66
17
Besides creating a scale, the means and standard deviations were also analysed, as can be seen in Table 3 below. By interpreting this table, we can come to the conclusion that the average age of the participants was around 28 years old. Also, when it comes to the completed education of the participants the mean was 5.62. In this study this number lays between the fifth answer option ‘Higher General Secondary Education or pre-university education’ and the sixth answer option ‘Higher Vocational Education’. This aligns with the expectation that most of the participants would be students. These generally have not completed a bachelor or master program yet but have finished their higher high school education.
Table 3
Means and Standard Deviations for Participants
Variable Mean (M) Standard Deviation (SD)
Gender 1.51 0.50 Age 28.49 13.19 Education 5.62 0.98 Presentation 1.49 0.50 VotingintentionD66 -1.23 1.63 AttitudeD66 4.51 0.91 Results
Prior to doing the actual analysis for testing the mediation, a few other steps have been taken. First a manipulation check was performed, that proved that the 158 participants were
successfully assigned to the two conditions. After that, a randomization check was executed with the demographical variables age, gender and level of education as dependent variables and the variable that measures whether participants viewed an infographic or a text as the factor. The one way variance analysis (ANOVA) used for this showed no significant results for the demographic variables gender (p = .63) and level of education (p = .06). Nonetheless, it appeared that the variable age (p = .03) was significant and therefore was used as a
18
confounding variable with the analyses that followed after this randomization check.
Moreover, a scale was created with the variables that tested for the mediator ‘attitude towards D66’, after the reliability and factor analysis pointed out it was justified to do so. Also, a frequency table that shows the standard deviations and means of the variables used was conducted. This table (table 3) can be found at the end of the Method section.
The for mediation and moderation analyses developed program PROCESS (Hayes, 2013) was used to perform the mediation analysis. This analysis is an alternative for a
‘regular’ linear regression analysis and tests whether there is a causal relationship between the variables in this study. The analysis mostly shows the values of B and the level of
significance. The B must not be mistaken for the Beta, since the B measures for
unstandardized variables, while the Beta measures for standardized variables. This is different from the linear regression analysis, that does show the Beta values.
The variable for voting intention towards D66 was used as the outcome variable (Y), the variable for the presentation of the message (either as an infographic or text) was used as the independent variable (X) and the variable for attitude towards D66 was used as the M Variable. As was mentioned earlier, the variable ‘age’ turned out to be a confounding variable. Therefore, this variable was used as a covariate.
The analysis showed that the way the message was presented (the independent variable) had a Beta of -.06 and a p = .67 for predicting the attitude towards D66 (the mediator). The covariate ‘age’ had a Beta of -.01 and a p = .25. Since both p-values weren’t significant (in both cases p > .05), both the independent variable as the covariate did not appear to predict the mediator.
Also, the analysis showed that the way the message was presented (the independent variable) had a Beta of -.01 and a p = .96 for predicting the voting intention towards D66 (the dependent variable). The mediator, ‘attitude towards D66’, had a Beta of -.01 and a p = .96
19
for predicting this dependent variable. The covariate ‘age’ had a Beta that was smaller than .01 and a p = .90. Likewise, in this case none of the p-values were significant, meaning that none of the variables predicted the dependent variable.
Finally, the analysis showed that the way the message was presented (the independent variable) had a Beta of -.01 and a p = .96 for predicting the voting intention towards D66 (the dependent variable. The covariate ‘age’ had a Beta of .01 and a p = .89. Since both p-values were not significant, both the independent variable as the covariate did not appear to predict the dependent variable. When it came to determining whether mediation had happened, I looked at the indirect effect of X (‘the presentation of the message’) on Y (‘voting intention towards D66’) via the mediator (‘attitude towards D66’). It appeared that the conference interval laid between -.05 and .05. The conference interval exists out of two bounds and it predicts for 95% that the effect is between those two values. With the PROCESS analysis a conference interval with the value ‘0’ between both bounds means that the mediation did not succeed and is not significant, while the opposite goes when both bounds remain beneath or beyond the value ‘0’ (Hayes, 2013). The indirect effect was smaller than .01 with a standard error of .02. Thus, the conclusion can be drawn that it is for 95% certain that the effect lays between -.05 and .05, which means that the mediation did not succeed.
Conclusion & Discussion
The current study had two different purposes. First of all, the aim was to examine whether the way information is presented leads to a more positive attitude towards a party and a higher voting intention of the audience. In particular, the focus laid on the difference in impact between using a text or an infographic as a way to present information. Secondly, the
experiment was performed to determine whether exposure to an infographic (vs. to a text) has a positive effect on voting intention for a party through higher levels of attitude towards that
20 party.
These aspects are considered as important, because it is useful for developers of political campaigns to know in what way to design them. This is relevant for them and especially for the party that has recruited them, because attitude towards a party and voting intention are considered as important predictors of someones’ actual voting behaviour. Similarly, infographics are an increasingly more common way to transmit information to an audience but there has not been done a lot of research on their effect. This study aimed to close this gap. However, unfortunately this study has been unable to find evidence that proves that the expectations described earlier in this paper are valid.
The first hypothesis stated that in order to improve the attitude towards a party and the voting intention of the audience with certain content it is more effective to present this
information as an infographic than in a text. Support of non-support was found for this
hypothesis. This means that there were no significant results found that prove that the attitude towards a party and voting intention are more affected when the information is presented as an infographic instead of as a text. These results do not coincide with Lang (2000; 2006), Bouquin and Brown Epstein (2015), Arcia et al. (2016), Gaber (1990) and Park, Jeon and Sullivan (2014), whose papers all describe that using images or visuals, such as infographics, can be seen as a successful way to change someones’ attitude towards a party and (voting) intentions in comparison to using text.
Nevertheless, these findings could be explained by the results that Powell, Boomgaarden, de Swert and de Vreese (2015) found with their study. They came to the conclusion that in order to change opinions and attitudes using text would be most effective, while using images is the best choice when the goal is to change someones’ behaviour. As I described earlier in this study it has been proven that intention can be a predictor of behaviour (Singh, Leong, Tan, & Wong, 1995). Although intention might be linked to behaviour, it
21
seems that it is indeed not certain whether using images is the best way to influence that intention.
The second hypothesis asserted that someones’ attitude towards a certain party, in this case D66, strengthens the (expected positive) effect of the way the information is presented on someones’ intention to vote for that party. In this case, there was again support of non-support found for the hypothesis. There was no significant mediation and thus it can be concluded that it cannot be said for certain that presenting information as an infographic (vs as a text) has a positive effect on someones’ intention to vote for a party through higher levels of attitude towards that party. Nevertheless, this does not necessarily mean that this is not true, it just cannot be proven with the data from the current study. These findings do not correspond with Himmelweit, Humphreys and Jaeger (1984), Vaske and Donnelly (2010), Echabe and Garate (1994) and Roccato and Zogmaister (2010), who found indications that attitude (towards a party) is a predictor of voting intention (towards that party).
A reason for this result could be that attitudes in general are not easy to change and require an intensive way of processing information in order to be influenced, as the already discussed Elaboration-Likelihood-Model states (Petty & Cacioppo, 1986). It could be the case that infographics solely ask for a superficial way of processing and do not require the central route to be followed, making it likely that it could be difficult for the information provided in these infographics to be of a great impact. Possibly, using a text for this does require people to use the more energy consuming central way of processing information and this could
stimulate them to change their attitudes.
There are a few reasons that could explain the insignificant effects that were found. First of all, the research was solely focussed on the party D66. Utz (2009) found in her study that a politician who responds to the comments of users on his or her social media account(s) is liked more by those users, in comparison to one who does not react to those comments.
22
Besides, she noted that this effect seemed to be stronger for voters who supported the left-wing and politicians who supported the right-left-wing. Although the reasons for choosing the neutral party D66 for this research were strong and they seemed to be suited for this type of research, the findings of Utz (2009) might also be applicable in the current study. This means that using a left-winged or right-winged party could lead to different effects than when a neutral party is used and therefore might contribute to the production of significant results. A recommendation for future research is to not only use a neutral party as the subject of the infographic and text but to also implement two additional conditions where the focus is on a left-winged party and a right-winged party. The effect of the orientation (either left-winged or right-winged or neutral) of the party can then be easily determined by analysis.
Secondly, no pilot study was performed before the actual experiment. Usually this is held to determine whether you are asking clear questions and to be sure that the subjectivity of the researcher does not play a role. It basically functions as an opportunity to improve your questions prior to doing the experiment. This pilot study could also have pointed out another limitation; questions could only be answered with the 7-point-Likert scale and there was not any room for participants to give suggestions or alternative questions. In conclusion, the lack of a pilot study in this research means a missed opportunity for feedback from participants. Thirdly, it could be put into question whether theories of other fields of
communication are suited to be applied to political communication after all. Several researchers, such as Lang (2006), are convinced that it is definitely possible. Although it might be an option for other studies, it could very well be the case that when it comes to the research on the effect of infographics those theories are not appropriate. Because of the fact that the amount of research on this topic is relatively low, a more decent option would be the development of new theories in the political communication. However, naturally this will take a generous amount of time, energy and money and it therefore might take a while before these
23
can be established. Up until then, the best solution is to keep investigating whether Lang (2006) and her colleagues are right about the appliance of theories from other fields of communications on political communication through trial-and-error.
Furthermore, most of the participants were students between the age of twenty and thirty. According to Spaniol et al. (2011) the brain of (young) adults of the age up until thirty years old responds differently to visuals than people who have already passed that age. In general they feel attracted more to colour and pictures. For this reason, it would make sense if younger adults would respond in another way to infographics and text than older adults. As the one way variance analysis in this study showed, the variable ‘age’ turned out to be a confounding variable. Therefore, a recommendation for future research is to divide the participants into, for example, three different groups, categorised by age. This way it will be easier to determine whether infographics possibly are of a greater influence on younger adults than text, or vice versa. Such results could be valuable for parties, because it enables them to find tools that are suited for their audience and its’ corresponding age category.
All in all, it remains difficult to pinpoint why this research showed insignificant results. However, inevitably the conclusion must be drawn that with the current study there could not be found support that presenting (campaign) information as an infographic
influences someones’ attitude towards D66 and his or her voting intention towards D66 in a more positive way in comparison to when text is used. Also, no evidence was found that using an infographic for presenting information has a positive effect on voting intentions for a party through higher levels of the attitude towards that party. Finally, as I already explained in the introduction of this study, there is no doubt that the gap that exists in research that focuses on infographics in the field of political communication is real and I hope that in the (near) future researchers will find the inspiration to put their time and effort into attempting to close this gap.
24 Reference list
Arcia, A., Suero-Tejeda, N., Bales, M.E., Merrill, J.A., Yoon, S., Woollen, J., & Bakken, S. (2015). Sometimes more is more: Iterative participatory design of infographics for engagement of community members with varying levels of health literacy. Journal of the American Medical Informatics Association, 23(1), pp. 174 – 183.
http://dx.doi.org/10.1093/jamia/ocv079
Bolt, A & Bruins, W. (2012). Effective Scientific Writing. Amsterdam: VU University Press.
Bouquin, D., & Brown Epstein, H. (2015). Teaching data visualisation basics to market the value of a hospital library: An infographic as one example. Journal of Hospital
Librarianship, 15(4), pp 349 – 364. http://dx.doi.org/10.1080/15323269.2015.1079686 Chang, C. (2000). Political advertising in Taiwan and the US: A crosscultural comparison of the 1996 presidential election campaigns. Asian Journal of Communication, 10(1), pp. 1 - 17. http://dx.doi.org/10.1080/01292980009364772
Çifçi, T. (2016). Effects of infographics on students achievements and attitude towards geography lessons. Journal of Education and Learning, 5(1), pp. 154 – 166. http://dx.doi.org/10.5539/jel.v5n1p154
Claassen, R.L. (2007). Floating voters and floating activists: Political change and information. Political Research Quarterly, 60(1), pp. 124 – 134.
http://dx.doi.org/10.1177/1065912906298594
Echabe, A.E., & Garate, J.F.V. (1994). Private self-consciousness as moderator of the importance of attitude and subjective norm: The prediction of voting. European Journal of Social Psychology, 24(2), pp. 285 – 293.
25
Graber, D.A. (1990). Seeing is remembering: How visuals contribute to learning from television news. Journal of Communication, 40(3), pp. 134 – 155.
http://dx.doi.org/10.1111/j.1460-2466.1990.tb02275.x
Han, C., Hoskins, B., & Boon-Yee Sim, J. (2014). The relationship between civic attitudes and voting intention: An analysis of vocational upper secondary schools in England and Singapore. A Journal of Comparative and International Education, 44(5), pp. 801 - 825. http://dx.doi.org/10.1080/03057925.2013.780874
Hayes, A.F. (2013). Introduction to mediation, moderation and conditional process analysis. New York: The Guilford Press.
Himmelweit, H.T., Humphreys, P., & Jaeger, M. (1984). How voters decide (2nd ed.) Milton Keynes, United Kingdom: Open University Press.
Kools, M., van de Wiel, M.W.J., Ruiter, R.A.C., & Kok, G. (2006). Pictures and text in instructions for medical devices: Effects on recall and actual performance. Patient Education and Counseling, 64, pp. 104 – 111.
http://dx.doi.org/10.1016/j.pec.2005.12.003
Kos, B.A. & Sims, E. (2014). Infographics: The new 5-paragraph-essay. Laramie: Rocky Mountain Celebration of Women in Computing.
Krauss, J. (2012). Infographics: More than words can say. Learning & Leading with Technology, 39(5), pp. 10 – 14.
Kress, G. (1998). Visual and verbal modes of representation in electronically mediated communication: The potentials of new forms of text. In Snyder, I & Joyce, M (Eds), Page to Screen Taking Literacy Into the Electronic Era (pp. 53 - 79). London - New York: Routledge. http://dx.doi.org/10.4324/9780203201220_chapter_3
Lang, A. (2000). The limited capacity model of mediated message processing. Journal of Communication, 50(1), pp. 46 – 70. http://dx.doi.org/10.1093/joc/50.1.46
26
Lang, A. (2006). Using the limited capacity model of motivated mediated message processing to design effective cancer communication messages. Journal of Communication, 56, p. 57 – 80. http://dx.doi.org/10.1111/j.1460-2466.2006.00283.x
Lee, Y. & Chang, C. (2011). Missing ingredients in political advertising: The right formula for political sophistication and candidate credibility. The Social Science Journal, 48(4), pp. 659 – 671. http://dx.doi.org/10.1016/j.soscij.2011.04.001
Lee, E., & Kim, Y.W. (2015). Effects of infographics on news elaboration, acquisition, and evaluation: Prior knowledge and issue involvement as moderators. New media & society, 1, pp. 1 – 20. http://dx.doi.org/10.1177/1461444814567982
Lewalter, D. (2003). Cognitive strategies for learning from static and dynamic visuals. Learning and Instruction, 13, pp. 177 – 189.
http://dx.doi.org/10.1016/s0959-4752(02)00019-1
Lundberg, K.B., & Payne, B.K. (2014). Decisions among the undecided: Implicit attitudes predict future voting behavior of undecided voters. PLoS One, 9(1), pp. 1 – 11. http://dx.doi.org/10.1371%2Fjournal.pone.0085680
Park, H.H., Jeon, J.O., & Sullivan, P. (2014). How does visual merchandising in fashion retail stores affect consumers’ brand attitude and purchasing intention? The International Review of Retail, Distribution and Consumer Research, 25(1), pp. 87 – 104.
http://dx.doi.org/10.1080/09593969.2014.918048
Petty, R.E., & Cacioppo J.T. (1986). The elaboration likelihood model of persuasion. Communication and Persuasion, 19, pp. 1 – 24.
http://dx.doi.org/10.1007/978-1-4612-4964-1_1
Powell, T.E., Boomgaarden, H.G., de Swert, K., & de Vreese, C.H. (2015). A clearer picture: The contribution of visuals and text to framing effects. Journal of Communication, 65(6), pp. 997 – 1017. http://dx.doi.org/10.1111/jcom.12184
27
Roccato, M., & Zogmaister, C. (2010). Predicting the vote through implicit and explicit attitudes: A field research. Political Psychology 31 (2), pp. 249 – 274.
http://dx.doi.org/10.1111/j.1467-9221.2009.00751.x
Singh, K., Leong, S.M., Tan, C.T., & Wong, K.C. (1995). A theory of reasoned action perspective of voting behavior: Model and empirical test. Psychology and Marketing, 12(1), pp. 37 – 51. http://dx.doi.org/10.1002/mar.4220120104
Smiciklas, M. (2012). The power of infographics: Using pictures to communicate and connect
to your audience. Indiana: QUE Publishing.
Smit, G., de Haan, Y., & Buijs, L. (2014). Visualizing news: Make it work. Digital Journalism. 2(3), pp. 344 - 354. http://dx.doi.org/10.1080/21670811.2014.897847 Soberman, D. & Sadoulet, L. (2007). Campaign spending limits and political advertising. Management Science, 53(10), pp. 1521 – 1532.
http://dx.doi.org/10.1287/mnsc.1070.0717
Spaniol, J., Voss, A., Bowen, H.J., & Grady, C.L. (2011). Motivational incentives modulate age differences in visual perception. Psychology and Aging, 26(4), pp. 932 – 939. http://dx.doi.org/10.1037/a0023297
Thompson, C.M. (2015). Creating “visual legacies”: Infographics as a means of interpreting and sharing research. Communication Teacher, 29(2), pp. 91 – 101.
http://dx.doi.org/10.1080/17404622.2014.1001761
Utz, S. (2009). The (potential) benefits of campaigning via social network sites. Journal of Computer-Mediated Communication, 14(2), pp 221 – 243.
http://dx.doi.org/10.1111/j.1083-6101.2009.01438.x
Van der Brug, W. (1999). Floating voters or wandering parties? The Dutch national elections of 1998. West European Politics, 22(1), pp. 179 – 186.
28
Vanichvasin, P. (2013). Enhancing the quality of learning through the use of infographics as visual communication tool and learning tool. Proceedings ICQA 2013: The
International Conference on QA Culture: Cooperation or Competition 2013 (pp. 135 – 142). Bangkok: Offset Plus.
Vaske, J.J, & Donnelly, M.P. (2010). A value-attitude-behavior model predicting wildland preservation voting intentions. Society & Natural Resources, 12(6), pp. 523 – 537. http://dx.doi.org/10.1080/089419299279425
29 Appendix – vragenlijst experiment
Q1: Ik ben een… Man Vrouw
Q2: Wat is uw leeftijd in jaren? … jaar (zelf invullen)
Q3: Wat is de hoogst genoten opleiding die u hebt afgerond?
1. geen opleiding
2. lagere school/ basisschool
3. middelbaar voortgezet onderwijs (MAVO, VMBO)
4. middelbaar beroepsonderwijs en beroepsbegeleidend onderwijs (MBO)
5. hoger algemeen en voorbereidend wetenschappelijk onderwijs (HAVO, VWO, Gymnasium)
6. hoger beroepsonderwijs (HBO)
7. wetenschappelijk onderwijs (Universiteit)
Stemintentie - afhankelijke variabele
Q4: De volgende Tweede Kamerverkiezingen vinden plaats op 15 maart 2017. Hoe groot is de kans dat u gaat stemmen bij deze verkiezingen?
30 2. Klein 3. Enigszins klein 4. Neutraal 5. Enigszins groot 6. Groot 7. Zeer groot
Q5: Hoe groot acht u de kans dat u in de toekomst op [partij] gaat stemmen?
1. Zeer klein 2. Klein 3. Enigszins klein 4. Neutraal 5. Enigszins groot 6. Groot 7. Zeer groot
Partijen waarvoor bovenstaande vraag wordt ingevuld: GroenLinks
PvdA (Partij van de Arbeid) SP (Socialistische Partij)
CDA (Christen-Democratisch Appèl) D66 (Democraten 66)
ChristenUnie
31 50Plus
PvdD (Partij voor de Dieren) PVV (Partij voor de Vrijheid)
SGP (Staatkundig Gereformeerde Partij)
Politieke interesse - moderator
Q6: In welke mate vindt u uzelf geïnteresseerd in politiek?
1. Zeer ongeïnteresseerd 2. Ongeïnteresseerd
3. Enigszins ongeïnteresseerd
4. Niet geïnteresseerd, niet ongeïnteresseerd 5. Enigszins geïnteresseerd
6. Geïnteresseerd 7. Zeer geïnteresseerd
[BLOOTSTELLING STIMULUS: infographic of tekst]
Attitude ten opzichte van presentatie boodschap – mediator
[De volgende vragen hebben betrekking op de vormgeving van de boodschap die u hebt gezien.]
Q7: De vormgeving van de boodschap was mooi.
32 2. Mee oneens
3. Enigszins mee oneens 4. Neutraal
5. Enigszins mee eens 6. Mee eens
7. Zeer mee eens
Q8: De vormgeving van de boodschap was aantrekkelijk.
1. Zeer mee oneens 2. Mee oneens
3. Enigszins mee oneens 4. Neutraal
5. Enigszins mee eens 6. Mee eens
7. Zeer mee eens
Q9: De vormgeving van de boodschap was duidelijk.
1. Zeer mee oneens 2. Mee oneens
3. Enigszins mee oneens 4. Neutraal
5. Enigszins mee eens 6. Mee eens
33 7. Zeer mee eens
Persuasion message - afhankelijke variabele
[De volgende vragen hebben betrekking op de inhoud van de boodschap die u hebt gezien]
Q10: Ik vond de boodschap overtuigend.
1. Zeer mee oneens 2. Mee oneens
3. Enigszins mee oneens 4. Neutraal
5. Enigszins mee eens 6. Mee eens
7. Zeer mee eens
Q11: Ik vond de boodschap geloofwaardig.
1. Zeer mee oneens 2. Mee oneens
3. Enigszins mee oneens 4. Neutraal
5. Enigszins mee eens 6. Mee eens
34 Q12: Ik vond de boodschap helder
1. Zeer mee oneens 2. Mee oneens
3. Enigszins mee oneens 4. Neutraal
5. Enigszins mee eens 6. Mee eens
7. Zeer mee eens
Q13: Ik vond de boodschap doeltreffend
1. Zeer mee oneens 2. Mee oneens
3. Enigszins mee oneens 4. Neutraal
5. Enigszins mee eens 6. Mee eens
7. Zeer mee eens
Q14: Ik vond de boodschap krachtig
1. Zeer mee oneens 2. Mee oneens
3. Enigszins mee oneens 4. Neutraal
35 5. Enigszins mee eens
6. Mee eens 7. Zeer mee eens
Recall message - mediator
Q15: Had de boodschap een duidelijk onderwerp? Er was geen duidelijk onderwerp.
Er was wel een duidelijk onderwerp, maar ik weet niet meer wat. Er was wel een duidelijk onderwerp, namelijk ...
Q16: Werd er een politieke partij genoemd in de boodschap? Er werd geen politieke partij genoemd in de boodschap.
Er werd wel politieke partij genoemd in de boodschap, maar ik weet niet meer welke. Er werd wel politieke partij genoemd in de boodschap, namelijk …
Q17: In de boodschap werd aangegeven dat er bezuinigd moest worden. Waren er specifieke onderdelen waarop bezuinigd moest worden?
Er waren geen specifieke onderdelen waarop bezuinigd moest worden.
Er waren wel specifieke onderdelen waarop bezuinigd moest worden, maar ik weet niet meer welke.
36
Q18: Was er een onderdeel waarop duidelijk het meest bezuinigd moest worden? Er was geen onderdeel waarop duidelijk het meest bezuinigd moest worden.
Er was wel een onderdeel waarop duidelijk het meest bezuinigd moest worden, maar ik weet niet meer welke.
Er was een onderdeel waarop duidelijk het meest bezuinigd moest worden, namelijk ..
Stemintentie - afhankelijke variabele
Q19: Na het zien van deze boodschap, bent u meer geneigd om te gaan stemmen op 15 maart 2017? 1. Helemaal niet 2. Niet 3. Waarschijnlijk niet 4. Neutraal 5. Waarschijnlijk wel 6. Wel 7. Helemaal wel
Q20: Zoals eerder is uitgelegd, vinden de volgende Tweede Kamerverkiezingen plaats op 15 maart 2017. U hebt net informatie gelezen over D66. Na het zien van deze boodschap, bent u meer geneigd om op D66 te stemmen bij de volgende Tweede Kamerverkiezingen?
1. Helemaal niet 2. Niet
37 4. Neutraal
5. Waarschijnlijk wel 6. Wel
7. Helemaal wel
Attitude ten opzichte van thema boodschap - afhankelijke variabele
Q21: Nu u deze boodschap gezien heeft, wat is uw houding ten opzichte van meer bezuinigingen op de Europese Unie?
1. Zeer negatief 2. Negatief 3. Enigszins negatief 4. Neutraal 5. Enigszins positief 6. Positief 7. Zeer positief
Attitude ten opzichte van politieke partij - mediator
Q22: Na het zien van deze boodschap vind ik dat D66 … is. 1. Ongeloofwaardig
2. … 3. … 4. … 5. …
38 6. …
7. Geloofwaardig
Q23: Na het zien van deze boodschap vind ik dat D66 … is. 1. Niet leuk 2. … 3. … 4. … 5. … 6. … 7. Leuk
Q24: Na het zien van deze boodschap vind ik dat D66 … heeft.
1. Slechte standpunten 2. … 3. … 4. … 5. … 6. … 7. Goede standpunten
39
Q25: Na het zien van deze boodschap, is mijn houding tegenover D66 … 1. Negatief 2. … 3. … 4. … 5. … 6. … 7. Positief Manipulatiecheck
[Voor het beantwoorden van de volgende vraag is het belangrijk dat u weet wat een
infographic is. Een infographic is een vorm van communicatie waarbij tekstuele informatie visueel aantrekkelijk wordt weergegeven, bijvoorbeeld door middel van afbeeldingen.]
Q26: Hoe was de boodschap van D66 over bezuinigingen op de EU, vormgegeven? In de vorm van een gewone tekst.
