RSCH Infographics, a better medium than plain text for increasing knowledge

Infographics in social networks present more elaborate proposals than simple text. In this study we analyze the general patterns of Twitter infographic messages to determine its main characteristics. Subsequently, we conducted a test to compare learning differences when young people use infographics instead of text as a source of information. Our findings suggest that while infographics are more effective than text for digitally mediated learning in young people, an analysis of age and gender subgroups showed that this effectiveness was only statistically significant for older people

information from a significantly narrower spectrum of sources through social media and email, compared to a search baseline (Nikolov, Oliveira, Flammini & Menczer, 2015).
Actually, young people use old strategies --e.g. learning by doing or solving problems--but in new learning environments such as Youtube tutorials (Scolari, 2018). As indicated in Scolari (2018:85), young people receive many transmedia stimuli, which go beyond simple reading, but in informal learning they continue to use the same classic learning modalities (learning by doing, playing, imitating, solving problems, examining or teaching to others). On the other hand, the negative effects of multitasking have been reported and the debate on the adequacy of the definition of "digital natives" is still open (Sharpe, Beetham & De Freitas, 2010;Kirschner & De Bruyckere, 2017).
This work is divided into two parts. In the first part, we have made a terminological effort to define infographics based on empirical data from Twitter. Then, we present a detailed study of visual elements transmitted through Twitter that combine text and image, and consequently we establish manually a gradation of nine types of communicative elements that go from the pure text to the pure image, establishing intermediate levels (4,5,6) as infographics (figure 2). As we shall see, the infographic is defined as a boundary object (Star & Griesemer, 1989) that combines effectively text and image: it is a multimedia visual language of ecdotic character with the property of recursion (see Morera-Vidal, 2017, 2017 and references therein), and with the ability to become a communicative meeting point between different cultures and generations, thanks to the synergetic combination of image and text.
Secondly, a pre-post test study was carried out to compare the effectiveness of the infographic versus the plain text (without image), in a questionnaire about a scientific topic, such as global warming. Thereby we verify if the infographics are more effective than texts for young people's learning (under 30) in an experiment which is digitally mediated. Although it remains a complex theoretical issue, for simplicity we will assume here a functionalist definition of learning (De Houwer, Barnes-Holmes & Moors, 2013), so that if after the presented stimuli the subjects respond correctly we will consider that they have learned. Finally we compare the results obtained in young people with those in adults, in order to check whether or not there is a generational effect that could be associated with the "digital natives" concept (Kirschner & De Bruyckere, 2017).

En prensa
Antoni Hernández-Fernández Francesc Morera-Vidal RSCH temporary digital culture is essential, since they have become one of the most relevant activities, encompassing both tools and signs (Vygotsky & Luria, 1994), acting as mediator in the transmission of digital information. Previous works have suggested that infographics are not only more effective but also more permanent in the minds (Ozdaml, Kocakoyun, Sahin & Akdag, 2016).), and become standardized forms (Star & Griesemer, 1989;Star, 2010;Bowker et al, 2015) of the digital media that enable coordination (or collaboration) across the communicative frontiers between different communities (Engeström et al, 2015).
Previous studies have shown that the design of the structure of the visual message is crucial. This infographic message can be divided into formal groups with a semantic nature combining text and image, so the reader may tend to follow visual paths that had not been considered by the designer of the infographic (Siricharoen, 2013). In this sense, Ware (2008) warns about two models of perception in the act of vision: "a) Bottom-up: information drives pattern building and b) Topdown: attentional processes reinforce relevant information". So, the reading of the infographic message is done in two levels: one direct level, so-called "insight", in which the reader realizes the type of visual message that has before it and, subsequently, a more detailed reading level where the information is accessed and ends up obtaining an interpretation of the full message (Ware, 2008).
Infographics have a marked intertextual character (Bloome & Egan-Robertson, 1993) and its processing goes beyond reading the written texts that they usually include: it is a communicative amalgam between image and text that transcends the mere image or text. This transversal conception of infographics is typical of transmedia narratives and multichannel languages, as Valero Sancho (2017) pointed out (figure 1), curiously without noticing the potential of Star's ecological approach (Star, 2010;Star & Griesemer, 1989;Timmermans, 2015). In Valero Sancho's (2017) approach, the main idea is that in visual communication each of the five fields represented in figure 1 (Communication, Calculus, Computing, Art and Other disciplines) intersects with the others. Then, relevant partial intersections emerge between disciplines.
Under this theoretical framework, the infographic is considered as central in visual communication and needs the contact of all the disciplines involved. While one could e.g. imagine that an infographic could be done without the help of computers (i.e. manually drawing it) and/or min-In short, we carry out a transdisciplinary approach with the aim of bringing advances in visual communication closer to the intercultural and intergenerational studies of digitally mediated communication. In what follows, we start by discussing the relevance of the study of infographics in digitally mediated communication.

Infographics as boundary objects in social networks
Surprisingly, there is no consensus yet on the definition of infographics (Siricharoen, 2013). In one hand, the infographic is not the mere conjunction of text and image: the so-called infographics are descriptive or narrative stories which aim to be visual alternatives to traditional texts or audios, which propose effective solutions in multicultural communication, increasingly improved as a consequence of digital development in visual communication. In fact, infographics are optimal for social networks and digital platforms, taking advantage of all the opportunities offered by these technologies in a natural way (Morera-Vidal, 2018).
Dick (2020), like Cairo (2012) and other authors understand the concepts of "infographic" and "data visualization" as interchangeably. We argue that data visualization is a category included in infographics. The narrative aspect of the infographic (explanation) can include elements of visualization (description). Although both products can display data, the infographic representation can contain visualizations. On the contrary, visualization cannot.
On the other hand, if we define boundary objects as representational forms that can be shared between different communities, with each community holding its own understanding of the representation (Star, 2010;Timmermans, 2015), the creation and management of boundary objects is a key process in developing and maintaining coherence across intersecting social worlds (Star & Griesemer, 1989), what is usual on the Internet, where individuals from different communities share content. Classically, according to Vygotsky (1962Vygotsky ( , 1978 two qualitative leaps occur in the individual over his development: the first one takes place when the subject acquires the oral language, while the second takes place when the subject acquires the written language. We could speculate that, with the advent of visual language, we are facing a third qualitative leap that will become increasingly important in the context of digitally mediated societies.
For what said above, a detailed analysis of communication through infographics in con-

En prensa 4
Infographics, a better medium than plain text for increasing knowledge

RSCH
imizing the influence of the other five theoretical fields represented in figure 1, the opposite is true in practice, and neither Art, Computing, Calculus, mass-media Communication, nor of course the cultural and social component (Other in figure 1) can be ignored. This approach may seem reductionist, and perhaps it groups under 'Other' too many fundamental components of visual communication (i.e. educational and socio-cultural), but exemplifies the interrelation between very diverse disciplines, something typical of ecological approaches and boundary objects (Star & Griesemer, 1989;Star, 2010;Bowker et al, 2015). As a consequence, infographics could be considered boundary objects, specifically intercultural digital "standardized forms" (Star & Griesemer, 1989), because technically they are elements of common digital communication across dispersed groups.

Approach and methodology
We face two studies in this work. The first (part one) is theoretical, although based on empirical evidence: to explore and describe infographics as visual statements of effective communication in social networks (here, Twitter). We use qualitative and quantitative methods that have previous-ly been found in visual elements (Valero Sancho, 2012;Morera-Vidal, 2017). This characterization allows us to define and locate technically the infographic in the digitally mediated communication. After that, in the second part we explore if infographics are better in digitally mediated learning than the simple text. A pre-test and post-test experiment is carried out to verify the learning effectiveness of infographics with respect to plain text in young people (under 30), comparing the results with adults (over 30 years).

Part one: describing infographics
A corpus of Twitter visual statements has been collected for analysis, following Valero's model (Valero Sancho, 2001). Twitter is one of the most active social networks, whose conciseness (maximum 140 characters) and immediacy make it a communication environment where all types of symbols, labels and images coexist with the text in an unprecedented way (Tascón & Abad, 2011). The collection of samples was done between February 2012 and April 2016, and was limited to static files that users attach as visual elements and that have the usual extension in image files (jpg, png or gif; tiff and bmp are not allowed in Twitter, and dynamic .gif files could not be included in Twitter until October 2015). They are onepage full-view files in a single visual scan without the need to press additional buttons or tabs. Samples were collected indiscriminately, according to the still controversial defining standards of the concept of infographics (Cairo, 2012;Dick, 2020;Morera-Vidal, 2017;Siricharoen, 2013).
In this semantic and qualitative analysis, to specify which visual statements could be considered fully as infographics we follow the theoretical precedents of Wileman (1993). We use a scale of relationship ( Figure 2a) between the text and the image of nine levels based on Molitor, Ballstaedt & Mandl's (1989).
Finally, 1382 visual statements were compiled, at an approximate rate of 1.2 daily files that rule out the accumulation of visual elements of the same trending topic. Thus, no file was rejected because it was not clearly considered as infographics, such as images without textual elements or text in which only the typeface or color font varied (Figure 2). Given the characteristics of this social network, all the infographics were shared by young users (considered being under 30 years of age in this study), whether they were followers of the author or that they were followed by him. Therefore, there could only be a thematic bias in the selection of visual elements.

RSCH
We consider that the infographics corresponds to levels 4, 5 and 6 on that scale. In these cases, the text and the image work closely to complement the information and both are linked to form a superior communication object.
In Figure 2 (

En prensa 6
Infographics, a better medium than plain text for increasing knowledge RSCH items.

Part two: infographics as an effective way of learning
We proposed an investigation to compare the infographic with the text as an effective way of learning, with a sufficient sample to obtain external validity (Wimmer & Dominick, 1996). For this purpose, a pre-test / post-test questionnaire was prepared with a control group in a digital context of normalized conditions, in a specific topic (in our case "global warming"). Indeed, a text (control group) and infographics with the same textual information but complemented with images (experimental group) was supplied. Figure 3 shows the test development scheme.
The study was carried out with the online So-GoSurvey platform. During the questionnaire the participants could not go back. There is a maximum time of 5 minutes for each of the phases, with a total time of 20 minutes. Following Ziefle (1998), it was considered that it could be done comfortably in 12 and 15 minutes. The methods are explained in Suplementary Information (SI). The questions (both in pre-test and post-test) and possible answers (in bold those considered correct) are summarized in table 1 in Suplementary Information (SI).
The statistical analysis was performed with the software SAS v.9.3. and R package sjstats (Lüdecke, 2021). There was a total of 403 partici-pants, of which 108 abandoned the process in the first two levels (Survey # 000 or Survey # 001), 200 formed the experimental group that observed the infographics (Survey # 003) and 95 the control group (which read the texts, Survey # 002). Participants randomly chose the route (Infographics or Text) and the preliminary Chi Square Test ruled out that neither age, nor gender, nor level of education (more than 82% in all groups had university studies) or previous experience in this kind of surveys influenced early abandonment ( Table  2 in SI).
A total of 295 subjects finalized the test, 115 under 30 years (57.5%) in the group of infographics (200 in total) and 43 (45.2%) under 30 in the control group (95 in total). In the experimental group, 46.5% identified as women and 52% as men (1.5% preferred not to indicate their gender) whereas in the control group 42.1% identified as women and 56.8% as men (1% did not indicate their gender). In the sample "under 30", 49 identified as men (42.6%) and 66 as women (57.4%) in the experimental group, and 16 as men (37.2%) and 27 as women (62.8%) in the control.
The abandonment of the post-test is statistically more common among users (young and old) who do not answer the pre-test questions correctly (see SI). Previous experience in knowledge about image treatment was not a determining factor in the early abandonment of the questionnaires. Although the residence factor has not influenced this study as we will see, it can be said that more than 80% of the sample resides in Barcelona and its surroundings, so the cultural environment is mostly that of the Mediterranean area of Catalonia, in southern Europe (82.5%) and the rest come from other European and American countries (17.5%). In short, although the topic of the study can be considered quite transversal and independent of cultural factors (being a scientific topic related to climate) it could be argued that the study sample shares a large part of socio-cultural elements and is, in this sense, homogeneous.

Infographics and learning
To assess the effectiveness of infographics in learning regarding the text, the intervention was stratified and the Cochran-Mantel-Haenszel statistic (CMH) was calculated. In figure 4 a case in which infographics varied much more than the text the correct answer to one of the questions is presented (Q2). The same tendency happened for all questions, in young and old people, regardless of gender, except in question 4 (see Figure 5).  ANOVA test shows that while there are no significant differences in test results between groups in young users (Infografics vs Text) in the pre-test (Df = 1, Sum Sq.= 1.0, Mean Sq.=1.0061, F=2.013, p=0.156), but there are in the post-test (Df = 1, Sum Sq.= 4.0, Mean Sq.=3.961, F=7.287, p=0.007**), so that the group to which the infographic is applied significantly improves learning (see Table 3a, 3b and 4 in SI, and figure 6), which is not the case in the control group that only received texts as learning material. On the other hand, an analysis of age and gender subgroups using the Cochran-Mantel-Haenszel test showed that this effectiveness was only statistically significant for older men, under this more strict test (see Table 5 in SI).
However, in the Likert type questions the evolution is similar in men and women, without sig-nificant differences by age or gender, although young men respond slightly better to infographics. Figure 6 shows this evolution in the responses (pre and post test) of young men and women. Indeed, infographics are more effective than texts in learning in all participants, when considering the full sample (see Tables 3a, 3b, 4 and 5 in SI). Other factors such as variations in the user location, previous experience or level of studies of the respondents did not significantly influence the results of each subgroup.

Conclusions
Classical studies on the learning and reading of visual statements (Levin & Divine-Hawkings, 1974), as well as the relationship between text and image in learning processes (Levin, 1989) and hypertextuality (Horn, 1998), anticipated that visual language is an essential part of communication, what is especially relevant in the context of the current digital social networks in which we live. On the other hand, efforts to understand learning processes and their dynamics currently challenge the traditional definitions of learning and education (Kumpulainen & Erstad, 2016), and are undoubtedly one of the most important challenges of the Big Data era. We should have always a clear definition of an element when we introduce it in the teaching-learning processes. In the first part of this work, after a detailed study of visual elements transmitted through Twitter we have discovered that it is not possible to date to distinguish the infographics automatically. Hence we still need human intervention, through semantic classification, to differentiate the infographic from other visual communicative levels (Figure 2). This fact maintains subjectivity in digitally mediated experiments with infographics, which poses a challenge to Evidence-Based Teaching (EBT) approaches, where the scientific method is intended to be applied (Groccia & Buskist, 2011).
In the second part of this work (pre-post test study), assuming a functionalist definition of learning (De Houwer, Barnes-Holmes & Moors, 2013), we show that infographics are more effective than texts for digitally-mediated learning. But when we compare the results obtained in young people and adults we don't find that the so-called "digital natives" (Prensky, 2001) score substantially better than adults (tables 4 and 5 in SI), providing evidence that the conception of 'digital natives' might be a pedagogical myth (Kirschner & De Bruyckere, 2017). Infographics can therefore improve the learning of everyone, including young people (see Tables 3a, 3b and 4 in SI), but an analysis of age and gender subgroups using the Cochran-Mantel-Haenszel test showed that this effectiveness was only statistically significant for older men ( Table 5 in SI).
The fact that age is not a relevant factor in the abandonment of the performed test ( Table 2 in SI) is significant and goes against some retrograde conceptions that tend to assume that young people are more inclined not to finish tasks or to abandon them prematurely, especially if these require an effort. The initial dropout rate in our test neither depended on age, nor on other factors such as gender, level of education or previous experience. Even so, the test has been carried out in a digitally mediated way, which could have been a determining factor for young people not to leave early. The percentage of dropout is greater in the users who had responded wrongly, which could be a consequence of discouragement because they consider that have not learned enough or do not finally master the subject. However, the abandonment in the post-test is higher in the older than in the young people, and more so in men than in women.
The latter allows us to hypothesize if young people, when faced with digitally mediated tests, tend to finish more this type of test than older people, but more studies are needed to corroborate this. This could be due to a greater adaptation to the digital context, mediated by the screen, so that young users could take more as a game the test and have a greater tendency to finish it, even though they are aware that their answers might not be correct. The fact that older men are users who tend to leave the test earlier, being aware that they are not responding correctly, may be because they receive greater social pressure not to recognize their mistakes. Do they prefer to abandon the test to acknowledge their ignorance? This psychosocial aspect of masculinity should be explored in more detail, but in our case it has been another factor to consider for future work.
The public and massive character of Twitter has made it impossible to differentiate the shared elements only by young users. Accordingly, future statistical studies are necessary to verify whether young users share more or less infographics (or other elaborated visual contents beyond the mere image) than the older users. For instance, a lower exposure of young users to infographics could explain their slightly worse results in the test, although this should be checked properly. Studies on Twitter's potential in learning have focused on written rather than visual language (see Rosell-Aguilar, 2018 for a review) overlooking the potential of the shared image in the networks. Figure 6: Evolution of the responses of the young users in the pre-post test study (odd questions) of the experimental group (infographics) and control (text). "Strongly agree" and "agree" are considered correct answers. The ANOVA test shows that there are no significant differences between groups in the pre-test (Df = 1, Sum Sq.= 1.0, Mean Sq.=1.0061, F=2.013, p=0.156), but there are in the post-test (Df = 1, Sum Sq.= 4.0, Mean Sq.=3.961, F=7.287, p=0.007**).

En prensa 10
Infographics, a better medium than plain text for increasing knowledge

RSCH
The same should be verified regarding the slight gender differences that we found: men (young and old) seem here to respond slightly better than women to the visual language of infographics ( figure 6), in the line of some studies that point to better masculine capacities in the processing of visual information (Shaqiri et al, 2018). A recent report from European Commission (2018) confirms that the gender gap in relation to internet usage exists only among individuals over 55 years old, and such gap has completely closed among those aged 16 to 25 and 25 to 54 years. Gender differences for the type of uses of the internet are also narrowing. In our test these slight differences between genders have been found in all age groups. There are different studies that conclude that sex differences in general intelligence are negligible (i.e. non-verbal and verbal ability across development), although sex differences persist in certain, specific cognitive domains such as in visuo-spatial ability (Toivainena, Papageorgioub, Tostoc & Kovas, 2017;Savage-McGlynn, 2012;Kersey, Braham, Csumitta, Libertus & Cantlon, 2018;Shaqiri et al, 2018).
According to the results of the 2015 PISA program about 15-year-old students (PISA, 2015), those students who access the Internet, chat or use social networks outside school score slightly higher than other students, and on average across OECD countries, girls score 29 points higher than boys. Could these differences in PISA be due, as a factor, to a pre-eminence of educational materials in text form with respect to visual elements such as infographics (which would favor boys slightly)? Let us recall that Li (2014) suggests that women generally score higher than men on verbal memory tasks, object location or recognition as well as verbal memory tasks, with differences in cue use and spatial memory (Jones & Healy, 2006), while the opposite is true for spatial mental rotation and navigation tasks.
Our findings suggest that infographics are boundary objects (Star & Griesemer, 1989) that help people to learn better than through plain text. Yet infographics are not statistically very much present in Twitter: in the sample taken in this study, only 192 visual elements of the 1290 considered were strictly infographics, which is a relatively low percentage (14.9%). Even more, if we consider the visual elements that are shared daily in other social networks, in our sample infographics probably would be overrepresented, since the infographics would represent a much lower percentage of the total shared visual content, given the preponderance of photography in all networks, especially in Instagram or Pinterest, and the tendency to create images that are mere text endowed with a colored background and a basic typography, in Facebook or Twitter. Leisure predominates in the social networks of young people, so the elements that are shared do not seem to be oriented to learning.
However, more field studies are needed with even larger samples. Although the survey was universally distributed in English, geographic areas in Asia and Africa were not accessed; thus, the limitations of this research are of a geographical nature with restrictions due to language or Internet access. In future phases of this research, it could be limited in specific areas for greater control of the data and people surveyed. Other future lines of research are oriented to the study of digital infographics as a bitmap message. Finally, our results should encourage educators to introduce more infographics as teaching materials, alternative or complementary to texts, whether in digitally mediated or traditional contexts.

En prensa 14
Infographics, a better medium than plain text for increasing knowledge

Suplementary information: Experimental methods in pre-post test study
The surveys in pre-post test study are the following: Survey #000: brief explanation of the questionnaire and identification of the general data of the subjects (age, gender, place of residence, level of studies and previous experience in this kind of topics) Survey #001: (Pre-test) test of ten questions about "Global warming" with four possible answers closed and a single valid correct answer.
Survey #003: 1 screen with text (the same as in the infographics) about "Global warming".
Survey #004: (Post-test) Final test with the same questions as in the Survey #001.
The questions (both in pre-test and post-test) and possible answers (in bold those considered correct) are summarized in table 1.
In questions that could be answered using a Likert scale of 4 responses (odd questions, Q1, Q3, Q5, Q7, Q9) and in questions with a specific correct answer (even questions, Q2, Q4, Q6, Q8, Q10), it was possible to evaluate the evolution in men and women in stating the correct answer in both groups and according to age . Tables 3a and 3b shows the evolution of these answers of young users and one way ANOVA test. As can be seen in Tables 3a and 3b, in addition to the early abandonment mentioned above there is another set of the sample who does not respond to the posttest. This abandonment of the post-test is statistically more common among users (young and old) who do not answer the pre-test questions correctly.