H.M. (Heleen) van de Lustgraaf | 1 Having a Top Spot How conceptual metaphors affect relevance perception of information in charts Heleen van de Lustgraaf ANR 828595 Master thesis Communication- en Information Sciences Specialization Communication and Design Faculties Humanities University of Tilburg, Tilburg Mentor: L. van Weelden Second reader: Dr. J. Schilperoord July 2013 2 | Having a Top Spot H.M. (Heleen) van de Lustgraaf | 3 Master thesis HAVING A TOP SPOT HOW CONCEPTUAL METAP HORS AFFECT RELEVANCE PERCEPTION OF INFORMATION IN CHARTS H.M. (Heleen) van de Lustgraaf ANR 828595, University of Tilburg Abstract Conceptual metaphors influence our way of thinking. The present study investigated whether the conceptual metaphors of RELEVANT IS UP, RELEVANT IS RIGHT and RELEVANT IS BIG are activated when looking at web charts. Two experiments were conducted in which the perceived relevance was measured by two means; relevance perception and recall order. The results of the present study suggest that only the conceptual metaphor RELEVANT IS UP affects our perception of relevance. The experiments indicate that the top spot for information is in the upper half of the web chart. Keywords: Importance, relevance, perception, conceptual metaphors, design Most people think that metaphors are mainly used by writers and poets to make their writings more lively, but metaphors are not only linguistic embellishments. They influence the way we structure our thoughts. The essence of a metaphor is understanding and experiencing one kind of thing in terms of another kind of thing (Lakoff & Johnson, 1980). As such, a metaphor is build upon two concepts; a target and a source. The target is the concept the metaphor is about, whereas the source is used to tell something about the target. Consider the famous sentence of Shakespeare; JULIET IS THE SUN. In this sentence Juliet is the target and the sun is the source, because the concept of the sun is used to tell that Juliet is as beautiful and warm as the sun. As such, characteristics of the sun are transferred to Juliet, as being characteristics of her. We structure our thoughts about the target, according to the characteristics of the source. Shakespear’s metaphor only deals with the transfer of certain features from one concept to another. This kind of metaphor can be made with all possible concepts into numerous combinations. We can convert Juliet into Tim and all of a sudden the features of the sun are transferred to him. Alternatively, we can replace the sun with the moon, and then the metaphor states that Juliet is as cold and lonely as the moon. This is not the case with conceptual metaphors. They are universal and grounded in our everyday life experiences in which critical aspects of the target and the source co-occur with one another (Grady, 1997; 1999). Consider for example the metaphor MORE IS UP. Imagine yourself as a toddler playing with boxes. You try to make a tower of the boxes and gradually you notice that the more boxes you use, the higher the tower gets. Based upon numerous of 4 | Having a Top Spot experiences where the concepts of more and up are tightly correlated, the two concepts of MORE and UP get integrated (Ortiz, 2010). MORE becomes UP. Would these conceptual metaphors influence how we process information in a diagram? When you are designing a chart to represent information about a certain topic, the main reason to pick particular information is that it is relevant for the topic and important for the reader to know. Sometimes charts are used for persuasive purposes, like giving multiple arguments for and against a point of view. However, what if certain information becomes more relevant because conceptual metaphors become activated by a certain position or size? This could mean that conceptual metaphors influences the way we process information. If these influences are known we could use them for our own advantage. Before we can examine this, it is crucial to select the conceptual metaphors that are likely to have influence on the process of processing information in a diagram. Furthermore, we should discover how we normally would process a diagram. The diagram chosen for this master thesis is a web chart. This is a spider-like chart design, where the middle box presents the topic of the chart and boxes that contain information about the topic are presented in different sizes and are placed clockwise around the center. See figure 1 for an example. This kind of a diagram is chosen, because it has three clear variables. All the boxes surround the main box in the centre, but they all have a different size, a different vertical position and a different horizontal position. This results into the research question of this master thesis; ‘Do frame position and scale affect relevance perception of information in a web chart?’. Fig.1. Web Chart Metaphors There are three conceptual metaphors that deal with the variables of size, verticality and horizontality: POWER IS UP (Schubert, 2005), GOOD IS LEFT or RIGHT, depending on hand coordination of a participant (Casasanto, 2011) and IMPORTANT IS BIG (Schilperoord, to be published). These metaphors will be discussed in detail below and will be translated to concepts that present relevance rather than power (Schubert, 2005), goodness (Cassanto, 2011) and importance (Schilperoord, to be published). H.M. (Heleen) van de Lustgraaf | 5 Vertical position POWER IS UP is a conceptual metaphor that can be traced back to the experience of laying in a crib and looking up to your parents above you. They took care of you, made decisions for you and gave you love. They were important to you, because they controlled your life. When you got up your feet, people who were taller then you were powerful. Your teacher, your big sister, your parents, other adults; they were all taller then you. Schubert (2005) examined the metaphor POWER IS UP to see if this metaphor indeed is embodied. This means, in this case, that the way we think about something is controlled by how it is presented to us. Research by Goatly (2007) shows that we are confronted with conceptual metaphors dealing with height and power on a daily basis. Sentences like “Your highness”, “being in the upper class” and “having a top job” carry the notion that relevance and power are associated with the concept of height. Schubert (2005) did a study on this metaphor with the goal to examine whether height is indeed connected to power. In his experiment Schubert presented sentences with three different power relations: powerful, powerless or horizontal, see figure 2 for examples. Participants were presented with six propositions with a powerful, powerless and horizontal relation between agent and patient. Then they had to choose a figure that best visualized the notion of the sentence, see figure 3. The results showed a relation between the figure the participants chose and the power relation that the sentence contained. The figure ● in the powerful sentence is favoured to be at the top of the figure and in the bottom for a powerless sentence.. Furthermore, the figures were favoured to be placed horizontally for a horizontal sentence. Powerful “● has influence on ○”, Powerless “● is weaker than ○.”, Horizontal “● pulls ○.” Fig. 2 Relations between agent and patient; sentence examples, Schubert 2005. Fig.3. Figures, Schubert. This notion of power might be related to the concept of importance. A powerful person is an important person. However, these concepts cannot be interchanged, because an important person might not be a powerful person. A man might be important and relevant to his wife, but he might not have power over her. On the other hand, a powerful person is always important and relevant. Hence, when someone has power over you, he is important for you, because he can influence you. From this, it can be concluded that important is a subcategory of the concept powerful. With this in mind the metaphor RELEVANT IS UP can be subtracted from the metaphor POWER IS UP. 6 | Having a Top Spot Horizontal position The study of Schubert (2005) discussed in the previous section also showed an effect of horizontality on power perception. The results show that powerful agents are more frequently given a left position and that powerless agents are more frequently located on the right side. Whether these results will appear in the present study is unsure, because Schubert stated that these effects are caused by causality, which flows from left to right (Maass & Russo, 2003), and hence is not the effect of perceived power. Casasanto (2011) conducted research on horizontal position related to goodness. The results showed that the conceptual metaphor GOOD IS RIGHT or LEFT is indeed present in people’s judgments of what is good. This preference for right or left was congruent with the dominant hand of the participant. Casasanto found that the participants who where left-handed placed the animal which was perceived as ‘good’ in a left column, whereas participants that were right-handed placed it in a right column. The notion of ‘good’ could be related to the concept of importance. It is possible the people see their dominant hand as the most important one of their two hands, because they use it often. Imagine for example that you are being held hostage and to make your family to hurry up with the money, they are going to chop off one of your hands. You get the terrible choice to pick the hand that you would like to keep. Most probably, you would like to keep your dominant hand, because that one is more important to you. From this horrifying story, the conceptual metaphor IMPORTANCE IS DOMINANT HAND SIDE can be extracted. Moreover, following the reasoning that important is relevant, RELEVANT IS DOMINANT HAND SIDE can be subtracted from this metaphor. Size Research by Schilperoord on the metaphor IMPORTANT IS BIG was presented in a lecture of the master course Multimodality and Communication in 20212. In his experiment, inspired by the research of Schubert (2005), participants were asked to choose drawings for sentences like “Grey rules over White”. Between drawings, there was a manipulation of size and position. In the experiment there were two hypotheses about the choice of figure for a powerful sentence like “● has influence on ○”, see figure 4. The first hypothesis is the verticality hypothesis, which claims that people choose the drawing in which the powerful grey subject is above the powerless white subject. The second hypothesis, the size hypothesis, claims that people choose the drawing in which the grey subject is bigger than the white, regardless of their position. The results of the study were in favour of this size hypothesis. Participants chose the figure in which a big circle represented the powerful object and a small circle represented the powerless object in a horizontal presentation more often than the figure in which equal sized objects where presented in vertical position of each other, with the powerful object presented above and the powerless below. This means that a bigger circle next to a small circle is better in visualising a powerful object, than a circle that was positioned above another circle of the same size. H.M. (Heleen) van de Lustgraaf | 7 Size hypothesis: Vertical hypothesis: ●= powerful ○= powerless Fig. 4 Schilperoord´s hypotheses. This finding is congruent with the literature research of Goatly (2007). Sentences like ‘making a big thing out of it” and ‘today is a big day’ carry the notion that importance is expressed by the word ‘big’. These findings suggest a preference of connecting big items to important and relevant items. It is expected that these findings will be reflected in the results of the present experiment. Pro cessing Diagra ms Now we know what kind of effects we might expect from these three conceptual metaphors, it is important to see what we know about how diagrams are processed. Hegarty (2011) states that the visual features of a display influence how that display is remembered. Other cognitive researchers also argue that if representations are informational equivalent, they are not computed the same, due to display features (Larkin & Simon, 1987). Hegarty’s (2011) theory describes how visual displays are understood and which processes there involved in comprehension. This theory is based upon work of Freedman and Shah (2002), and Pinker (1990). An adjusted version of this diagram is shown in figure 5. Conceptual Metaphors Salient Attention External Display Visual Features Domain Knowledge Inference Internal Representation of Referent Fig. 5. Diagram of the different representations (indicated by boxes) and processes (indicated by arrows) that influence the processing of a diagram, with an own addition of the concept of metaphors. Adapted from “The cognitive science of visual-spatial displays: implications for design.” by M. Hegarty, 2011, Topics in cognitieve science, 3, 446-474. The external display is the way the graph is presented on paper. How the reader will remember the display, the internal representation, is influence by its visual features, such as shape, colour, and position. Hegarty’s theory predicts that attention for visual features is directed by a viewer’s goal and expectations or by what is salient in the display. What is perceived as salient by the reader 8 | Having a Top Spot can be manipulated. Hegarty gives the example that people might perceive items that are red as being dangerous and, very relevant for the current study, that items that are higher than other items are perceived as being better. This means that the internal representation and the actual representation of a diagram might differ and that this is caused by what is experienced as salient in a display. Things that are experienced as salient can be grounded in conceptual metaphors. When content is presented with certain salient visual features (e.g., particular information might be presented above other information in a chart), the conceptual metaphor of RELEVANT IS UP might be activated. Based on this conceptual metaphor, the information at the top of the graph might be perceived as more important than the information at the bottom of the graph. So, conceptual metaphors could be the cause for the difference between the external display and the internal representation. This must be interesting for a designer, because it seems there are connections in our lives by which we make sense of things, and these relations influences what we think is important. If people obtain the information resulting from this research, they can use it to their advantage to make information seemingly more relevant by applying the right characteristics. As a results, the viewer may recall them better as well. Meta-analyses (Benoit, 1998; Wood & Quinn, 2003) show that if people know they are being influenced, they are less likely to adapt their opinion. This is called the forewarning of persuasive intent. Logically speaking, when designing a persuasive object, you want to influence people without them knowing. The results of this experiment might aid in doing so, because people are not aware of the conceptual metaphors that organise their lives. If the influences of these metaphors are tested on content, you can design a figure in such a way that it chooses your side in an argumentative piece, but the readers would not know it. They presume they are getting objective information, but the figure can be designed in particular ways making particular content more important and better remembered. So according to Hegarty’s (2011) theory, some information becomes more salient due to visual features. In my thesis I claim that what is perceived as more salient, is caused by conceptual metaphors. This leads to the following hypotheses. H1. Information in boxes will decrease in relevance when the size of the boxes decreases. H2. Left-handed participants will perceive information in boxes on the left side as more relevant and right-handed participants will perceive information on the right side as more relevant. H3. Information in the boxes above the middle box will be rated as more relevant than boxes below the middle. All independent variables (horizontal position, vertical position and size) will be present in one figure. This creates the opportunity to examine interaction-effects. Based upon Schilperoord’s experiment (to be published) that confirmed the size hypothesis, the following hypothesis is created. H.M. (Heleen) van de Lustgraaf | 9 H4. The size of the boxes has a bigger effect on perceived relevance of information than the horizontal or vertical position of the boxes. Experiment 1 The first experiment evaluated the role of Verticality, Horizontality and Size on the perceived relevance of information. The variables of Horizontal position, Vertical position and Size are combined in one web chart, which can be seen figure 6. Participants were presented with this web chart and were asked to give a relevance score for the content in the boxes (for the content see appendix A). ←Big ←Medium Up ↑ ↓ Down ← Small Left ← → Right Fig.6. Web Chart with independent variables Relevance judgments are typically measured with ranking on contrasting scales (Riddle, Horwitz and Dietz, 1966), but a recurring finding from such studies is that the inter- and intrajudge reliability of these kinds of judgments is not very high. Two studies of Cuadra et al. (1966; 1967) show that these judgments can be influenced and manipulated by the kind of instructions the judges are given. This underlines the importance of finding a correct method that reflects the judge’s actual opinion of what is relevant. The terms ‘ranking’ and ‘rating’ are often seen as synonyms of each other, but Katter (1968) shows that this is not the case. Ranking is putting objects or content in a fixed order. The number of slots in which you can rank the content is predetermined. This may results in a distortion of the actual perceived rank. This restricting might result in judges giving information a higher or lower ranking than they would when their choice would be unrestricted. On the other hand, they also could make a distinction between two kinds of content when actually this discrimination is not in line with their judgment. This problem does not occur when judges are asked to rate the content, because no predetermined number of slots is given. Katter (1968) recommends using rating-scales instead of ranking-scales, because the outcome of rating scales reflects the judge’s actual opinion in a more accurate way. In this experiment relevance will be measured with a 7-point likert-scale. This is subtracted from Katter’s rating scales, because it meets the requirements of unrestricted choices and does not force the judges to make distinction. 10 | Having a Top Spot Method Participants Forty-two Tilburg University students and teachers (30 women, 12 men) participated. The mean age was 22 (SD=4.79) years, ranging from 18 to 50 years. The majority (N=26) of the students were from the Humanities faculty. Design The experiment had a 2 x 2 x 3 x 2 x 3 design, with Vertical position (levels: Up and Down), Horizontal position (levels: Left and Right) and Size (levels: Big, Medium and Small) as withinsubjects factors and Topic (levels: Licorice and Fire tornado) and List (levels: List 1, List 2 and List 3) as between-subjects factor. Materials The participants received two forms, see Appendix C. The first presented the web chart and asked the participants to score the information on relevance. The second form was about demographical information. The web chart consisted of 12 boxes with information about one of the two topics, fire tornado or licorice. The boxes differed in size (small, medium and big), vertical position (up and down) and horizontal position (left and right), see Figure 6. The size of the boxes increased percentually from 100% (small box), 125 % (medium box) to 150% (big box), and with this enlargement the font size of the information increased from 7.5 pt. (small box), 9.5 pt. (medium box) to 11.5 pt. (big box). To ensure that content did not influence the results, there were three experimental lists in which the content of the boxes was varied (see appendix B for distribution of content). The number of words of the created items was between 11 and 14 words (M=11.88, SD=1.03), and the number of tokens had a range from 57 to 74 tokens (M=64.9, SD=3.65). See appendix B for the exact numbers and appendix A for the sentences that were used. In addition, the relevance of an information item for the topic of the web chart and newness of this information was controlled for with a pilot-test. In total twenty-four participants (15 women, 9 men) with a mean age of 27 years, ranging from 18 to 49, participated in this pilot-test. The participants rated the relevance of the 12 licorice items and newness of 12 fire tornado items on a 7-point likert-scale, or the newness of the 12 licorice items and relevance of 12 fire tornado items on a 7-point likert-scale. A one-way ANOVA revealed no effect of topic on relevance (F(1, 286)=1,333, p=0.25), but there was an effect of topic on newness (F(1, 286)=13,859, p<0.001). The topic of fire tornado’s (M=5.17, SD=1.67) had a higher overall newness score then the topic of licorice (M=4.37, SD=1.99). After a close examination of the items, it was concluded that one item about the flavour of licorice had an extreme low newness outcome (M=1.42, SD=0.67). For this reason, it was replaced with a new item about licorice and ice cream. A second pilot study was conducted with this new item. In total 12 participants, 7 female and 5 male, with a mean age of 27 years, ranging from 21 to 58, took part in the second pilot study. The participants rated the relevance and newness of the 12 sentences of licorice on a 7-point likert-scale. An analysis showed that the newness score of licorice H.M. (Heleen) van de Lustgraaf | 11 (M=4.94, SD=1.19) had gone up with 0.57 compared to the first pilot study. After implementing the data of pilot study 1 of the fire tornado’s and the data of pilot study 2 of the licorice items, a one-way ANOVA revealed that there was no longer an effect of topic on relevance (F<1), or on newness (F(1, 286)=1,196, p=0.28). It can thus be assumed that any differences between topics cannot be assigned to structural differences between items of the two topics. Tab.1 Mean (M) and standard deviation(SD) of the Relevance and Newness Scores of Pilot study 1 (fire tornado) and 2 (licorice) combined. Relevance Newness M SD M SD Fire tornado (12) 5.17 1.67 4.68 1.54 Licorice (12) 4.94 1.88 4.77 1.34 Note. Relevance score and newness score has a range of 1 (minimum relevance/newness) till 7 (maximum relevance/newness) It is important that the relevance score of the items the same, see table 1, because the distribution of the items is not equal across all independent variables, see table 2. If information itself influences the perceived relevance, the effects of the independent variables cannot be measured. With other words, if an effect is found on the perceived relevance it is unsure which variable caused this effect. Fortunately, the two pilot studies ensure that any found effects are not caused by the information the items give. Tab2. Distribution of Item (N) across Independent Variables Size (Small, Medium, Big), Horizontality (Left, Right) and Verticality (Up, Down). Size Item (N) Horizontality Verticality Small Medium Big Left Right Up Down Type/Root (42) 14 14 14 28 14 28 14 Japan/Giorgio (42) 14 14 14 28 14 28 14 Time/Percentage (42) 14 14 14 28 14 28 14 Names/Kilo (42) 14 14 14 14 28 28 14 Distance/Salt (42) 14 14 14 14 28 28 14 Height/Pressure (42) 14 14 14 14 28 28 14 Trees/Spend (42) 14 14 14 28 14 14 28 Combination/Popular (42) 14 14 14 28 14 14 28 Co-occur/Originate (42) 14 14 14 28 14 14 28 Speed/Hardness (42) 14 14 14 14 28 14 28 Inside/Produced (42) 14 14 14 14 28 14 28 Phenomenon/Producer (42) 14 14 14 14 28 14 28 Total (504) 168 168 168 252 252 252 252 Note. 12 items were used in 6 diagrams, each shown to 7 participants, resulting in 504 items in total. 12 | Having a Top Spot Procedure Participants were told that the purpose of the experiment was to examine how people read diagrams. The participants were randomly assigned to one of two topics and one of the three experimental lists. They were given a survey on paper showing a web chart and were told to give every box a random number from 1 till 12 and that they should exclude the middle box. After this was done, they scored the information items in the boxes on a 7-point likert-scale (1 (minimum relevance) till 7 (maximum relevance)) with the corresponding number. They were asked to score the information according to how relevant this information was for the general topic. Afterwards they filled in a form regarding their demographical information. Results A one-way ANOVA showed no main-effect of Topic on the relevance score (F<1). Therefore, Topic is not taken into further analysis. A one-way ANOVA did reveal that the items had a significant influence on the relevance score (F(11,492)=3.88, p<.001). However, because the pilot studies ensured that Item would not influence relevance, it is assumed that this effect is causes by the unequal distribution of items across the independent variables. Therefore, Item is not taken into further analysis. An ANOVA analysis showed no main effect of Size (F<1), Verticality (F(1,492)=3.43, p=0.07) and Horizontality (F(1,492)=3.02, p=0.08) on the relevance score. The analysis showed no significant two-way interactions between Size and Verticality (F<1), between Size and Horizontality (F(2,492)=1.66, p=0.19) and between Horizontality and Verticality (F<1). The analysis did show a three-way interaction between Size, Verticality and Horizontality (F(2,492)=5.56, p<0.01, η2= .022). An ANOVA separated by Size indicates that this interaction between Horizontality and Verticality is caused by the medium boxes (F(1,164)=4.88, p<.05, η2=.029) and small boxes (F(1,164)=6.33, p<.05, η2=.037), see figure 7 and table 2. If the information in a medium upper box that appeared on the left side is compared to the information in a box that appeared on the right side, the information on the right side had a higher relevance score as compared to the information on the left side (+1.07). For the medium lower boxes the increase is minimal (+0.05), see figure 8. The relevance of the information in a small box responds differently on the movement from left to right. The information in an upper small box decreases in relevance when being shifted from left to right (0.71), but the information in a lower small box increases when being moved from left to right (+0.57), see figure 9. H.M. (Heleen) van de Lustgraaf | 13 +1.07 -0.71 +0.57 +0.05 Relevance Score Fig.7. Changes in the relevance score when medium or small boxes appear left or right. Relevance score has a range of 1 (minimum relevance) till 7 (maximum relevance) 7 5 3 Up 1 Down Left Right Horizontality Relevance Score Fig.8. Two-way interaction of Horizontality, Verticality and Size, focused on the Medium boxes. Relevance score has a range of 1 (minimum relevance) till 7 (maximum relevance) 7 5 3 Up 1 Down Left Right Horizontality Fig.9. Two way interaction of Horizontality, Verticality, focused on the Small boxes. Relevance score has a range of 1 (minimum relevance) till 7 (maximum relevance) 14 | Having a Top Spot Tab.3 Means (and SDs) for the Relevance Score of Size (Big, Medium, Small), Verticality (Up, Down) and Horizontality (Left, Right) Left (252) Big Medium Small Total Right (252) Total (504) M SD M SD M SD Up (84) 4.14 1.63 4.36 1.65 4.25 1.64 Down (84) 4.64 1.53 4.93 1.60 4.79 1.56 Total (168) 4.39 1.59 4.64 1.64 4.52 1.62 Up (84) 3.93 1.55 5.00 1.36 4.46 1.55 Down (84) 4.62 1.56 4.67 1.53 4.64 1.53 Total (168) 4.27 1.59 4.83 1.45 4.55 1.54 Up (84) 4.76 1.57 4.05 1.72 4.40 1.68 Down (84) 4.19 1.74 4.76 1.57 4.48 1.68 Total (168) 4.48 1.68 4.40 1.68 4.44 1.67 Up (126) 4.28 1.61 4.47 1.62 4.37 1.61 Down (126) 4.48 1.61 4.79 1.56 4.63 1.59 Total (252) 4.38 1.61 4.63 1.60 4.50 1.61 Note. Relevance score has a range of 1 (minimum relevance) till 7 (maximum relevance) If we compare the mean relevance score of the variable Horizontality, see table 3, we notice that information shown on the left side of the chart is perceived as less relevant (M=4.38, SD=1.61), than information displayed at the right side (M=4.63, SD=1.60). When looking at the variable Verticality, the scores show that information that appeared in the top of the chart was perceived as less relevant (M=4.37, SD=1.61) then information that was shown in the bottom section of the chart (M=4.63, SD=1.59). The relevance score of the variable Size indicate that the information in the medium boxes is seen as most relevant (M=4.55, SD=1.54), followed by the information in the big boxes (M=4.52, SD=1.62) and the small boxes (M=4.44, SD=1.67). Conclusion The results of Experiment 1 show that there is a three-way interaction of Size, Verticality and Horizontality on relevance score, but not one of the variables had influence on his own. Hence it can be concluded that Size, Verticality and Horizontality (hypothesis 1-3) do not influence the perceived relevance of information. Hypothesis 4 stated that the Size is more influential on relevance than Verticality, but neither one of the variables had influence on its own. Only a combination of the variables did influence the perceived relevance of an item. Information in big boxes was perceived as most relevant if it appeared in the lower right side. The information in medium boxes was perceived as most relevant when appearing in the top right corner, and also increased slightly in perceived relevance when the information was presented at the bottom right corner. Information in small boxes was perceived as most relevant, either in the top left corner or in the bottom right corner. H.M. (Heleen) van de Lustgraaf | 15 When we link all these results back to the theory of Hegarty (2011), we now know that the visual features that make information more relevant are a combination of position and size. It seems that information in medium boxes becomes more salient when it is on the right side. Information in small boxes on the other hand, becomes more salient when it is presented on either the top left corner or the lower right corner. Concluding, the top spot for information to get a higher relevance score is in a medium box in the upper right corner (M=5.00, SD=1.36). Discussion Figure 10 shows the boxes that were positively affected in the perception of relevance score by their position. Only the perceived relevance of the information in medium and small boxes was affected by their horizontal and vertical position. The boxes that are indicated in figure 10 are the places where information perceived a higher relevance score, then when the same information was presented in another box. Whether the participants were left or right handed did not influence the relevance score when the analysis was divided on horizontality. Due to this RELEVANCE IS DOMINANT HAND is changed into RELEVANCE IS RIGHT, because there were only two left-handed participants and in total 40 right-handed participants. RELEVANCE IS RIGHT is found in this experiment, when looking at the mean relevance score and with an ANOVA but only in combination with Verticality and Size. Fig.10. Boxes of which the perceived relevance of the content had benefit from its position and size. Evidence for RELEVANCE IS UP is not found in this experiment. The only box that had benefit from an upper position was the medium box, and this was in combination with a right position. It could be that Size has a neutralizing effect on Verticality, as was found in the experiment of Schilperoord (to be published). Only the effects predicted from the metaphor of RELEVANCE IS BIG were not found. Size had no significant influence, or in any other combination with variables, on relevance. Sperber and Wilson´s (1995) relevance theory states that messages become more relevant when they contribute to an increase of the knowledge of a reader. Experiment 1 did not measure the relevance of information for the reader, but the relevance contribution information had for the main topic. Experiment 1 indeed did not show an effect of newness of the information on relevance score. Nevertheless, it cannot be excluded that some participants might have thought they had to give a rating on relevance for themselves instead of relevance to 16 | Having a Top Spot the main topic. This means that relevance ratings can fluctuate due to different relevance types. To control for this confound, another measurement of importance was used in Experiment 2; recall. Experiment 2 The second experiment evaluated the role of Verticality, Horizontality and Size on the recall of the information in the chart. The same web chart and content were used as in experiment 1. Participants were presented with the web chart and were asked to study and recall the information in the boxes. Multiple researchers (Britton, Meyer, Simpson, Holdredge & Curry, 1979; Freebody & Anderson, 1986) have found that it is more likely to recall main or superordinate information then less important or subordinate information. Their research demonstrated that there is a positive relation between the perception of the importance of the information and the likelihood that this information will be recalled (Gomulicki, 1956; Johnson, 1970). This could mean that if the information in a certain position or box size is more often recalled than the same information presented in another box or position, this is caused by visual features that made the information seem more relevant and subordinate. The findings of Gomulicki (1956) and Johnson (1970) support the hierarchical representation hypothesis. This hypothesis states that important information is represented in the higher levels in the brain, whereas less important information is stored in lower levels. Due to this way of storing information, information that is more important is easier to access then less important information and therefore better to recall (McKoon, 1977). Method Participants Forty-two Tilburg University students (30 women, 12 men) participated for course credit. The mean age was 22 (SD=2.62) years, ranging from 18 to 28 years. The majority (N=40) of the students were from the Humanities faculty. Materials and Design The materials and design used were the same as in Experiment 1. Procedure Participants were told the purpose of the experiment was to examine how people read diagrams and how this influences recall. The participants were seated in front of a computer screen showing the web chart and were randomly assigned to one of two topics and one of the three experimental lists. Participants were granted five minutes to study the content of the chart. After these five minutes the participants were instructed to resolve a brainteaser in 5 minutes, to prevent short-term memory influences. Then, the participants received a recall form. They were asked to write down everything they remembered of the topic they read. The recall form had twelve places to fill in, to remember participants about the number of boxes they had seen. This was done in order to get maximal response. Again, the participants were granted five minutes, H.M. (Heleen) van de Lustgraaf | 17 after which they filled in a form about demographical information. For an example of all forms that were used, see Appendix D. Results The recall of an information item and the recall order were converted into a recall score. If an information item was remembered first, it was assigned a score of 12 points, because there were twelve items to remember. If an information item was remembered second, it was assigned 11 points, etcetera. Erroneous items were included in the recall order and 'stole away' scores. So if the third item was recalled incorrectly, but the fourth item was recalled correctly, the latter was assigned 9 points. In order to evaluate the items on correctness, an evaluation form was designed. The main principle was that a recalled item had to contain three main concepts to be a correct recall, see appendix A for these words. Synonyms and different word-orders were also judged as correct, as long as the semantic meaning did not differ. A one-way ANOVA showed no main effect of topic on the recall score (F(1,231)=1.51, p=0.22). Due to this Topic is not taken into further analysis. It did show a significant effect of Recall order score item on recall score (F(11,231)=3.65, p<.001). However, because the pilot studies ensured that Item would not influence relevance, it is assumed that this effect is causes by the unequal distribution of items across the independent variables. Therefore, Item is not taken into further analysis. An ANOVA showed no main effect of Size (F<1)), and Horizontality (F(1,220)=1.01, p=.32). The analysis did show a main effect of Verticality (F(1,220)=8.04, p<0.01, η2= .04). The upper boxes were recalled faster (M=8.69, SD=2.98) than the lower boxes (M=7.51, SD=2.37), see figure 11. 9 8 7 6 Up Down Verticality Fig.11. The effect of Verticality on the recall order score.. The recall order score has a range from 12 (remembered first) to 1 (remembered last). The analysis showed no significant two-way interactions between Verticality and Horizontality (F<1) or a two-way interactions between Verticality and Size (F(2,220)=1.05, p=0.35). The analysis did reveal a two-way interaction between Horizontality and Size (F(2,220)=6.81, p<.01, η2= .058). A separate ANOVA splitted by Size shows that this interaction with Horizontality is caused by the big boxes (F(1,71)=12.97, p<.01, η2=.162). The big boxes are recalled faster when they are positioned at the right side of the diagram (M=9.44, SD=2.36), than on the left side of the diagram (M=6.97, SD=2.91). So, if the same content in the big boxes on the left appears in a big box on the right, their recall orders score increases (+2.47), see figure 12. There was no three-way interaction between Size, Verticality and Horizontality on recall order (F(2,220)=2.64, p=0.07). Recall order score 18 | Having a Top Spot 9 8 7 6 Left Right Verticality Fig.12. The effect of Horizontality on the recall order of the big boxes. The recall order score has a range from 12 (remembered first) to 1 (remembered last). If we compare the mean recall order score, see table 4, of the variable Horizontality we notice that information shown on the right side of the chart is recalled earlier (M=8.33, SD=2.64), than information displayed at the left side (M=8.07, SD=2.94). When looking at the variable Verticality, the scores show that information that appeared in the top of the chart was recalled earlier (M=8.70, SD=2.98) then information that was shown in the bottom section of the chart (M=7.51, SD=2.37). The recall order score of the variable Size indicate that the information in the small boxes is recalled firstly (M=8.27 SD=2.75), followed by the information in the big boxes (M=8.23, SD=2.90) and the medium boxes (M=8.10, SD=2.77). Tab.4. Means (and SDs) for the Recall Order of Size (Big, Medium, Small), Verticality (Up, Down) and Horizontality (Left, Right) Left (58) Big Medium Small Total Right (66) Total (124) M SD M SD M SD Up (40) 6.41 3.73 10.04 2.08 8.50 3.22 Down (31) 7.50 2.36 8.38 2.53 7.87 2.43 Total (71) 6.97 2.91 9.44 2.36 8.23 2.90 Up (47) 9.14 3.09 8.00 2.62 8.68 2.93 Down (36) 7.44 2.71 7.25 2.12 7.33 2.37 Total (83) 8.52 3.04 7.62 2.38 8.10 2.77 Up (46) 9.30 2.90 8.48 2.79 8.89 2.85 Down (32) 7.47 1.91 7.27 2.87 7.38 2.87 Total (78) 8.53 2.66 8.00 2.85 8.27 2.75 Up (133) 8.52 3.29 8.89 2.62 8.69 2.98 Down (99) 7.47 2.29 7.56 2.48 7.52 2.37 Total(232) 8.07 2.94 8.33 2.64 8.19 2.79 Note. The recall order score has a range from 12 (remembered first) to 1 (remembered last). H.M. (Heleen) van de Lustgraaf | 19 Conclusion Experiment 2 showed that information that appeared in boxes in the upper half of the web chart were recalled sooner, as compared to when information was presented in the lower half. That means that boxes in the upper half of the web chart are perceived as more relevant. This confirms hypothesis 3. Size did not have an effect on the recall order score. This means that hypothesis 1 is rejected. Horizontality also did not have an effect on the recall score. This means that hypothesis 2 is rejected as well. This shows that being high has a bigger influence on the recall then being big. As such, hypothesis 4 is not confirmed. The combination of a big box (Size) and Horizontality influenced the perceived relevance of information, and therefore the recall order score. When a big box was on the right, its content was significantly recalled sooner then when it was on the left. The top spot for information to get a higher relevance score is in a medium box in the upper left corner (M=9.75, SD=2.52). Discussion Figure 13 shows the boxes that had benefit from their position, when looking at the recall order score. The information in these boxes was remember sooner and therefore was perceived as more relevant, than when the same information was presented in another box. Fig.13. Boxes from which the perceived recall order score of the content was influenced positively by position Whether participants were left or right-handed did not influence the recall score, when the analysis was divided by horizontality. That this effect is not found might be caused by the few number of the left-handed participants; 2 of the 42 participants were left handed. Due to this RELEVANT IS DOMINANT HAND is changed into RELEVANT IS RIGHT. RELEVANT IS RIGHT is found with in this experiment. A ANOVA showed an effect of Horizontality, but only in combination with Size. The present study found evidence for the role of RELEVANT IS UP in perceived relevance of information. No evidence was found for the conceptual metaphor of RELEVANT IS BIG. This is in contradiction with the results of Schilperoord (to be published), who found that the size of an item is more prominent in deciding what has a powerful role, than verticality. Research (Pichert & Anderson, 1977; Steffensen, JoagDev, & Anderson, 1979) has shown that a reader's background knowledge serves to highlight particular elements and increase their memorability. The amount of background knowledge that the participants indicated to have of the general topic did have no influence on the mean recall order of items 20 | Having a Top Spot (F<1). This means the difference in recall is caused by other variables, such as horizontality and verticality. General Discussion The purpose of the present study was to investigate the role of the conceptual metaphors, RELEVANT IS UP, RELEVANT IS RIGHT and RELEVANT IS BIG in the processing of information in web charts. The predictions were that these three metaphors would influence the perceived relevance of the information. Two experiments were conducted, in which the perceived relevance was measured by two means; relevance score and recall score. The results of the present study suggest that only the conceptual metaphor RELEVANT IS UP affects the perception of relevance of particular information. Two experiments showed that the concept of relevance is related to spatial positions and sizes. Experiment 1 showed a three-way interaction between Verticality, Horizontality and Size. The main results of this experiment can be seen in figure 14 on the left. Experiment 2 showed that there is a main effect of Verticality on the recall score, and that there is a two-way interaction between Size and Horizontality. The main results of this experiment can be found in figure 14 on the right. The results of Experiment 1 and Experiment 2 are presented in the middle figure in figure 14. This figure shows an area for which the recall was best; the upper half.. Boxes Relevance Increased Area of Benefit Boxes recall Increased Fig.14. Boxes from which the perceived relevance of the content was influenced by position (left) and the boxes from which the perceived recall order score of the content was influenced by position (right) are combined to form the figure in the middle which indicates the area (grey) in which content gets an increased relevance/recall. The stronger the color, the stronger the evidence is that this area benefits from its location However, some restrictions to these results are necessary. The division of the web chart into a left and right side, and in an upper and lower side, has played a main role in this experiment. For example, in figure 13 the small upper left box is indicated as a position that creates bigger relevance. Nevertheless, when comparing the two upper boxes on the right and on the left side, the righter box is placed lower than the small box on the left. It could be that the division of left and right has played an influence on the results about Horizontality. It could be that the results are a combination of the variable Horizontality and Verticality. The differences in vertical position are small, but they are there. A better design of the web chart would have all the upper and lower boxes on the exact same height and all the left and right boxes on the exact same width, see figure 15 for an example. This design should also be presented in mirror image, to even out differences. An H.M. (Heleen) van de Lustgraaf | 21 example of such a difference that should be evened out is the upper big box. It is presented relatively more to the left then the medium and small box. When the mirror images are also evaluated, this would not cause any problems. Up Left Right Down Fig.15. A web charts with clearer upper, bottom, right and left division. I n Experiment 1 and 2, no main effects of horizontality and size are found, only two-way or three-way interactions. In the following section, the absence of these effects will be discussed. Size The size of the boxes increased from 100% (small), to 125% (medium), to 150% (big). When increasing the boxes, the font size of the information in the boxes increased with it, see figure 16. This was done in order to prevent the white framework surrounding the text being a different width, which would create another variable. Sm a l l box Font size: 7.5 Box size: 100% Medi um box Font size: 9.5 Box size: 125% Bi g box Font size: 11.5 Box size: 150% Fig.16. Example that difference in size also influences difference in font size. Font: Adobe Fangsong Std R Bold. Translation text; “In some countries they eat licorice in combination with chocolate and vanilla ice-cream” In the two experiments, no results were found for the dependent variable of size. This could be caused by the decision to make the font size bigger when the box was made bigger. Research on tag clouds (Rivadeneira, Gruen, Muller & Millen, 2007) shows that recall for words with a larger font size is significantly higher than for words with a smaller font size. Strangely, no significant results were found, even though that there were actually two variables, box size and font size. This should make the same information better to remember when it is presented large, then when it is presented small. The absence of the effect of size can be caused by the minimal difference between the boxes. An increase of 25% per box might not be enough to make a real difference. In future experiments it would be wise to create greater difference in box sizes. 22 | Having a Top Spot Horizontality A key factor in the absence of an effect might be that the group of left-handed participants were considerably smaller then a right-handed group. In experiment 1 and 2 only 4 left-handed people in total participated, compared to 80 right-handed participants. This is a flaw in this master thesis and future experiments should aim for a better contribution of dominant hands on participants. Schubert’s (2005) experiment was used in the introduction for the discussing the metaphor of POWER IS LEFT or RIGHT. Question marks were presented next to these results by Schubert, because he blamed it to causality relations among the sentences. But what if these results are actually the results of the horizontality preferences of power visualizations by left handed people? Figure 17 shows again the figures that were used in this experiment. As you might remember, the task was to find a figure that best visualized the notion of the sentences. These sentences contain a powerful, powerless or horizontal relation among agent () and patient (○). When focusing on the horizontal figures, the results indicate that these figures were mainly chosen for horizontal power relation sentences. Figure 18 is taken from Schubert’s results section (2005) and it gives a closer examination of these results. Fig.17. Figures Schubert (2005) These figures make very clear that horizontality is seen for horizontal power relations, because the majority chooses this when presented with an equivalent sentence. When being presented with a powerful sentence, only 5.6% chose the figure where the agent is on the left and only 0.6% the figure were the agent is on the right. This indicates a slight favor to the left position for power and a favor for the right position as powerless. This is incongruent with the findings of this research. It could be to caused by the dominant-hands of the participants of Schubert. Unfortunately he does not rapport on this. Holder (1997) claims that 70-90% of the world population is right-handed. So if a researcher does not take measures to make sure distribution is equal, the majority will be right-handed. Powerfull “● has influence on ○” Powerless “● is weaker than ○.” Equivalent “● pulls ○.” Fig.18. Percentages chosen for propositions with powerful and powerless agents, and horizontal relations The end of the line indicates the agent’s () position. Reprinted from “Your highness: vertical positions as perceptual symbols of power” by T.W. Schubert, 2005, Journal of Personality and Social Psychology, 89, p. 1–21 H.M. (Heleen) van de Lustgraaf | 23 To summarize, perceived relevance of information can be influenced by verticality. This influence is strongest in the medium and small boxes. The experiments indicate that the top spot to make your information more relevant is in a medium box in the upper half. Further research should take the following three points in consideration. First, they should aim for an equal contribution of dominant hands on participants. This should be done in order to be able to discover whether the dominant hand of participants is an indication on which he perceives information to be more relevant. Secondly, future research should aim for a greater difference in box sizes. In the present research effects of size were not found. Another feature to considered while increasing the size of the boxes is whether the font size should increase with it. Research of Rivadeneira, Gruen, Muller and Millen (2007) indicated that recall for words with a larger font size is significantly higher than for words with a smaller font size. Increasing the font size would increase the possibility of finding an effect, but this would be due to the font size instead of the box size. Thirdly, researchers should use a web chart with a clean cut between left/right and up/down and present is to one of the two groups in mirror image. On top of this all, before we are able to implement the results in the design world, an experiment on a chart that is found in the ‘real world’ is recommended. Literature Benoit, W.L. (1998) Forewarning and persuasion. In M. Allen & R.W. Preiss (eds.), Persuasion; advances through meta-analysis (pp. 139-154). Cresskill, NJ: Hampton Press Britton, B.K., Meyer. B.F., Simpson, R., Holdrcdgc, T.S. & Curry, C. (1979). Effect of the organization of text on memory: Tests of two implications of a selective attention hypothesis. Journal of experimental Psychology: Human Learning and Memory, 5, 496506. Casasanto, D., & Chrysikou, E. (2011). When Left is 'Right': Motor fluency shapes abstract concepts. Psychological Science. 22(4) 419–422 Cuadra, C. A, E. H. Holmes, R. V. Katter & E. M. Wallace (1967) Experimental studies of relevance judgments: Third Progress Report. TM-3347, System Development Corporation, Santa Monica, January Freebody, P. & Anderson, R.C. (1986). Serial position and rated importance in the recall of text. Discourse Processes, 9,31-36. Freedman, E.G., & Shah, P. (2002) Toward a model of knowledge-based graph comprehension. In M.Hegarty, B.Meyer, & N.H. Naryanan (eds.), Diagrammatic representation and inference (pp. 8-31). Berlin: Springer-Verlag Goatly, A. (2007) Washing the brain; Metaphor and Hidden ideology, John Benjamins Publishing company. Chapter 2; Power metaphors. Gomulicki, B. R. (1956). Recall as an abstractive process. Acta Psychologica, 12, 77-94. Grady, J. (1997). Foundations of meaning: Primary metaphors and primary scenes. (Doctoral dissertation, University of California, Berkeley, 1997). 24 | Having a Top Spot Grady, J. (1999). A typology of motivation for conceptual metaphor: Correlation vs. resemblance. In R. W. Gibbs & G. Steen (Eds.), Metaphor in cognitive linguistics (pp. 79– 100). Amsterdam, The Netherlands: Benjamins. Hegarty, M. (2011) The cognitive science of visual-spatial displays: implications for design. Topics in cognitieve science, 3, 446-474. Holder, M.K. (1997) Why are more people right-handed? Scientifuc American Inc. http://www.scientificamerican.com/article.cfm?id=why-are-more-people-right Retrieved 2013-06-04. Johnson, R. E. (1970). Recall of prose as a function of the structural importance of the linguistic unit. Journal of Verbal Learning and Verbal Behavior, 9, 12-20. Katter, R.V. (1968). The influence of scale form on relevance judgments, Information Storage and Retrieval, 4(1), pp. 1-11. Lakoff, G. & Johnson, M. (1980) Metaphors We Live By, University of Chicago Press Maas, A. & Russo, A, (2003). Directional bias in the mental representation of spatial events. Psychological Science, 14, 296-301. Ortiz, M.J. (2010) Visual Rhetoric: Primary Metaphors and Symmetric Object Alignment . Metaphor and Symbol, 25: 162–180 Pichert, J.W. and Anderson, R.C. (1977). Taking different perspectives on a story. Journal of Educational Psychology, 69, 309-315 Pinker, S. (1990) A theory of graph comprehension. In R. Freedle (Ed.) Artificial intelligence and the future of testing (pp. 73-126) Hillsdale, NJ: Erlbaum. Rivadeneira, A.W., Gruen, D.M., Muller, M.J., Millen. (2007). Getting our head in the clouds: Toward evaluation studies of tagclouds. In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 995-998. ACM, New York Schilperoord, J. (2012, October). Metaphors of Power. Powerpoint presented at the master course Multimodality and Communication, Tilburg University, Tilburg Schubert, T.W. (2005). Your highness: vertical positions as perceptual symbols of power. Journal of Personality and Social Psychology, 89, 1–21 Sperber, D. & Wilson, D. (1995) Relevance; communication and cognition. Oxford: Blackwell Steffensen, M. S., Joag-Dev, C., & Anderson, R. (1979). A cross-cultural perspective on reading comprehension. Reading Research Quarterly, 15, 10–29. Wood, W. & Quinn, J.M. (2003) Forewarned and forearmed? Two meta-analytic syntheses of forewarnings of influence appeals. Psychological Bulletin, 129 (1), 119-138 H.M. (Heleen) van de Lustgraaf | 25 Appendix es Content Appendix A – Content boxes – text, count and recall words. Appendix B – List Variable – content distribution Appendix C – Example forms Experiment 1 – relevance score Appendix D – Example forms Experiment 2 – recall order score 26 | Having a Top Spot Appendix A – Content boxes – text, count an d recall words In table A1 and A2 the information of the items that were used in experiment 1 and 2 are presented. Table A1 presents the sentences of the fire tornado topic and table A2 presents the sentences of the licorice topic. In these tables the token count (T), work count (W) and code per sentence are specified. To see how much a sentence differed from the mean, the token and word count are subtracted from the mean word(W-M) and token count (T-M). For experiment 2 (recall) an evaluation form was created in order to approve or disapprove a recall. In order to get a correct recall, the participants had to recall the concepts visualised in bold in the tables below. Tab. A1 Content of Boxes of the Fire tornado Topic Items with Additional Information Code T T- M W WM In 1923 ontstond in Japan een vuurtornado met de grootte van een stad (A fire tornado emerged in Japan with a size of a town in 1923) Japan 57 -7.8 13 1.1 Er zijn drie types vuurtornado’s, die variëren in hoe gevaarlijk ze zijn (There are three types of fire tornado’s, which vary in how dangerous they are) Types 61 -3.8 12 0.1 Height 64 -0.8 13 1.1 Distance 63 -1.8 12 0.1 Een vuurtornado wordt ook wel eens een vuurduivel of een vuurwervel genoemd (A fire tornado also goes by the name fire devil or fire whirl ) Names 64 -0.8 12 0.1 Een vuurtornado is een fenomeen waarbij vuur een verticale wervelwind vormt (A fire tornado is a phenomenon where a fire forms vertical whirlwind) Phenomenon 65 0.3 11 -0.9 Vuurtornado’s zijn een gevreesd fenomeen, omdat ze moeilijk in bedwang zijn te houden (Fire tornado’s are a feared phenomenon, because they are hard to control) Feared 67 2.3 14 2.1 Speed 70 5.3 11 -0.9 Vuurtornado’s zijn meestal 10 tot 15 meter hoog, maar kunnen groeien tot 1 kilometer (Fire tornado’s are 10 till 15 meters high, but can grow till 1 kilometer) De langste afstand die een vuurtornado heeft afgelegd is bijna 5 kilometer (the longest distance a fire tornado travelled is almost 5 kilometre) Vuurtornado’s verspreiden het vuur waaruit ze ontstaan zijn, met grote snelheden (Fire tornado’s spread the fire from which they originated, with large velocities) H.M. (Heleen) van de Lustgraaf | 27 Vuurtornado’s gaan vaak gepaard met wild vuur, onweer, storm en windtornado’s (Fire tornado’s are often accompanied by wildfire, thunder, storm and wind tornadoes) Co-occur 67 2.3 11 -0.9 Vuurtornado’s ontstaan uit een combinatie van vuur en stijgende warme lucht (Fire tornado’ arise from a combination of fire and rising warm air) Combined 65 0.3 11 -0.9 Trees 64 -0.8 11 -0.9 Duration 70 5.3 12 0.1 Mean Range 64.8 61-74 Mean Range 11.9 11-14 Bomen tot 15 meter worden ontworteld door vuurtornado’s, die daarna in vlammen opgaan (Fire tornado’s can uproot trees of 15 meters, which go up in flames) Een gemiddelde vuurtornado duurt enkele minuten, maar kan oplopen tot 20 minuten. (An average fire tornado takes a few minutes but this can go up to 20 minutes.) Note. Theme ‘Fire tornado’ content, with tokens (T) and words (W) per sentence, giving mean tokens 64.8, with a range of 61 till 74 tokens, and a mean word count of 11.9 words, with a range of 11 till 14 words. Further the difference in word count compared with mean word count (W-M) and difference in token count compared with mean token count (T-M) Tab. A2 Content of Boxes of the Licorice Items with Additional Information Code T T-M W WM Giorgio 65 -0.1 11 -0.8 Root 64 -1.1 11 -0.8 Het eten van grote hoeveelheden drop veroorzaakt een te hoge bloeddruk (Eating large amounts of licorice causes high blood pressure) Pressure 60 -5.1 11 -0.8 De zoute smaak van drop wordt veroorzaakt door ammonium chloride, oftewel salmiak (The salty taste of licorice is caused by ammonium chloride, also known as salmiak) Salt 63 -2.1 11 -0.8 Kilo 63 -2.1 14 2.2 In 1731 ontdekte Giorgio Amarelli hoe drop industrieel gemaakt moest worden (Giorgio Amerelli discovered how licorice industrial should be made in 1731) Drop wordt gemaakt van de zoethoutstruik, die in de zon worden gedroogd (Licorice is made of concentrated licorice root, which is sundried) Een Nederlander eet gemiddeld 32 kilo snoep per jaar, daarvan is 2 kilo drop (A Dutchmen eats an average of 32 kilo candy per year, 2 kilo of this is licorice) 28 | Having a Top Spot Nederland is de grootste producent van drop, daarna komen Spanje en Duitsland (The Netherlands are the biggest producer of licorice, followed by Spain and Germany) Producer 66 0.9 12 0.2 Nederland produceerde dit jaar voor maar liefst 90 miljoen euro aan drop (The Netherlands produced an amount of licorice, worth 90 million Euros) Produced 61 -4.1 12 0.2 Hardness 74 8.9 12 0.2 Originate 68 2.9 12 0.2 Popular 68 2.9 11 -0.8 20 % van het verkochte snoep in Nederland is drop, wat gelijk staat aan 168 miljoen euro (20 % of all the sold candy in the Netherlands is licorice, this equals to 168 million euro’s. ) Spend 62 -3.1 11 -0.8 In sommige landen eten ze drop in combinatie met chocolade en vanille-ijs. (In some countries they eat licorice in combination with chocolate and vanilla-icecream) Ice 67 1.9 14 2.2 65.1 60-74 Mean Range 11.8 11-14 Drop varieert in hardheid door de hoeveelheid ingedikt zoethoutwortel dat wordt toegevoegd (Licorice varies in hardness, soft or hard, by the amount of concentrated licorice root) Niemand weet precies wanneer drop ontstaan is, vermoedelijk in de Romeinse tijd (Nobody exactly knows when licorice originated, probably in Roman times) Drop is populair in Nederland, maar ook in Engeland, Scandinavië, en Duitsland (Licorice is popular in the Netherlands, but also in England, Scandinavia and Germany) Mean Range Note. Theme ‘Licorice’ content, with tokens (T) and words (W) per sentence, giving mean tokens 65.1, with a range of 60 till 74 tokens, and a mean word count of 11.8 words, with a range of 11 till 14 words. Further the difference in word count compared with mean word count (W-M) and difference in token count compared with mean token count (T-M) H.M. (Heleen) van de Lustgraaf | 29 Appendix B – List variable – content distribution In tables B1 and B2 the distribution of the information of the items is presented; the list variable. Table B1 presents the sentences of the fire tornado topic and table B2 presents the sentences of the licorice topic. Tab.B1 Distribution of Items among List Variable for the Fire Tornado Topic. Fire to rnado Box Content code List 1 List 2 List 3 Box Content code List 1 List 2 List 3 1 Type Sul Mur Bdl 7 Trees Sdl Mdr Bul 2 Japan Mul Bur Sdl 8 Combination Mdl Bdr Sul 3 Time Bul Sur Mdl 9 Co-occur Bdl Sdr Mul 4 Names Sur Mul Bdr 10 Speed Sdr Mdl Bur 5 Distance Mur Bul Sdr 11 Feared Mdr Bdl Sur 6 Height Bur Sul Mdr 12 Phenomenon Bdr Sdl Mur Note. Circulation of content, themed ‘fire tornado’. First letter: small box, big box, second letter: up vertical position, down vertical position, third letter: left horizontal position, right horizontal position. Tab.B2 Distribution of Items among List Variable for the Licorice Topic. Lico rice Box Content code List 1 List 2 List 3 Box Content code List 1 List 2 List 3 1 Root Sul Mur Bdl 7 Spend Sdl Mdr Bul 2 Giorgio Mul Bur Sdl 8 Popular Mdl Bdr Sul 3 Percentage Bul Sur Mdl 9 Originate Bdl Sdr Mul 4 Kilo Sur Mul Bdr 10 Hardness Sdr Mdl Bur 5 Salt Mur Bul Sdr 11 Produced Mdr Bdl Sur 6 Pressure Bur Sul Mdr 12 Producer Bdr Sdl Mur Note. Circulation of content, themed ‘licorice’. First letter: small box, big box, second letter: up vertical position, down vertical position, third letter: left horizontal position, right horizontal position. 30 | Having a Top Spot Appendix C – Example forms Exp erim ent 1 – releva nce scor e This appendix presents an example of the forms that are used in experiment 1. The forms are in Dutch, as the experiment was executed in Dutch. Throughout this appendix translations will be given of the forms when needed. Fig.C1. Example Web Chart Figure C1 contains the sentences used for the fire tornado, for the translation see Appendix A, table A1. This figure was presented with the following accompanying text; “Randomly assign the twelve squares above a number from 1 to 12. Make sure the number is clearly written in the square. You can skip the square with “Fire tornado”. After this, give a number that states how relevant you find this information for the topic ‘fire tornado’ on the scale with the corresponding number.” The scale was a 7-point liker-scale that varied form 1 (minimal relevance) to 7 (maximal relevance). See Table C2 for the original text in Dutch. H.M. (Heleen) van de Lustgraaf | 31 Tab.C2. The Original Accompanying Text of Figure C1 in Dutch. Geef de twaalf vierkanten hierboven willekeurig nummers van 1 t/m 12. Zorg ervoor dat de nummers duidelijk in het vierkant staan. Daarbij kunt u het vierkant met “Vuurtornado” overslaan. Geef daarna bij hetzelfde nummer hieronder op de schaal aan, hoe relevant u deze informatie vindt voor het onderwerp “Vuurtornado”. 1. minimale relevantie 0 0 0 0 0 0 0 maximale relevantie 2. minimale relevantie 0 0 0 0 0 0 0 maximale relevantie 3. minimale relevantie 0 0 0 0 0 0 0 maximale relevantie 4. 5. 6. 7. minimale relevantie minimale relevantie minimale relevantie minimale relevantie 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 maximale relevantie maximale relevantie maximale relevantie maximale relevantie 8. 9. 10. 11. minimale relevantie minimale relevantie minimale relevantie minimale relevantie 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 maximale relevantie maximale relevantie maximale relevantie maximale relevantie 12. minimale relevantie 0 0 0 0 0 0 0 maximale relevantie Note. The relevance score scale varied from 1 (minimal relevance) to 7 (maximal relevance). After figure C1 and table C2, a form with demographic information was given. This form asked the participants about their sex, their age, their dominant hand, their faculty and study year. They were also given 3 statements about figure C1, which they had to respond on with a 7-point likert-scale, see table C3. The last question was a multiple answer that asked how the read the web chart. The original questions in Dutch are specified in table C4. Tab.C3. The Statements about the Information in Figure C1. Before this research, I already knew a lot about the topic 0 5 0 6 0 7 The information in this research was new for me 0 0 0 0 0 0 0 Not new Nothing 0 1 1 0 2 2 0 3 0 4 3 4 5 6 7 The information was interesting 0 0 0 0 0 0 0 Not interesting 1 4 5 6 7 2 3 Note. For the original text in Dutch, see figure C4. Everything New Interesting 32 | Having a Top Spot Fig.C4. The Original Question about Demographical Information and the Web Chart in Dutch –Vragenlijst – Ik ben een 0 man 0 vrouw Ik ben 0 linkshandig 0 rechtshandig 0 Zowel links- als rechtshandig Mijn leeftijd is Mijn faculteit 0 Humanities 0 Economics and Management 0 Law School 0 Social and Behavioural 0 Theology Ik ben een 0 eerstejaars 0 tweedejaars 0 derdejaars 0 vierdejaars of ouder 0 pre-master 0 master Stellingen Ik wist voorafgaand aan het onderzoek al veel over het onderwerp 0 0 0 0 0 0 0 Niets 1 2 3 4 5 6 7 Alles De informatie in dit onderzoek was nieuw voor mij 0 0 0 0 0 Niet nieuw 1 2 3 4 5 0 6 0 7 Nieuw Ik vond de informatie interessant 0 0 0 Oninteressant 1 2 3 0 6 0 7 Interessant 0 4 0 5 Ik heb het figuur gelezen (meerdere antwoorden mogelijk) 0 van rechts naar links 0 van links naar rechts 0 van grote vierkanten naar kleine vierkanten 0 van onder naar boven 0 kriskas 0 van boven naar onder 0 van kleine vierkanten naar grote vierkanten 0 anders, namelijk … 0 weet ik niet H.M. (Heleen) van de Lustgraaf | 33 Appendix D – Exampl e forms Exp eriment 2 – recall order scor e This appendix gives an example of the forms that are used in experiment 2. The forms are in Dutch, as the experiment was executed in Dutch. Throughout this appendix translations will be given of the forms when needed. The form about demographical information in Appendix C is also used in experiment 2, but will not be shown here. Experiment 2 worked with a PowerPoint and forms. The first slide of the PowerPoint asked the participant to follow the instruction on the slides and not to touch the keyboard to prevent complications. Between slides participants were explained what would follow, what was expected from them and how much time they would be given. An important slide showed for example figure C1 of appendix C, with the question to study the web chart. They had several minutes to observe the web chart, before the brainteaser in figure D1 emerged. This translation of the accompanying text; “Find the six-numbered code with help from the clues below. De numbers can vary from 0 to 6. The numbers 7, 8 and 9 are not used in the code. Scratch off the numbers on the buttons A-F until the correct answer remains. “ Fig.D1 Brainteaser After the brainteaser the participants received the form presented in table D2. The translation of this form; “Write down everything you know about the topic. There were 12 items in total; try to write them all down. Keep an eye on the time!” Afterwards the form about demographical information was given, see appendix C; figure C4. Tab.D2. Recall Form –Geheugentest – Schrijf alles op wat je nog weet over het onderwerp. Er waren 12 feiten in totaal, probeer ze allemaal op te schrijven. Houd de tijd in de gaten! 1. 2. Etc. Note. The original text had 12 lines to write down recall, to remember participants of the number of items.
© Copyright 2026 Paperzz