Web Accessibility for Older Readers: Effects of Font Type and Font Size on Skim Reading Webpages in Thai

Transcription

1 Web Accessibility for Older Readers: Effects of Font Type and Font Size on Skim Reading Webpages in Thai Sorachai Kamollimsakul 1,2, Helen Petrie 1, and Christopher Power 1 1 Human Computer Interaction Research Group, Department of Computer Science, University of York, United Kingdom 2 School of Information Technology, Suranaree University of Technology, Thailand {sk750,helen.petrie,christopher.power}@york.ac.uk Abstract. Most guidelines for making websites accessible for older people have been developed for the Latin alphabet. Currently, there are no web design guidelines for the Thai language or for Thai older people. Our research investigated the effect of font type and size in Thai on skim reading for Thai younger (21-39 years) and older (59-72 years) adults. There were two levels of font types (Conservative and Modern, which correlate to serif and sans serif types, respectively) and three levels of font sizes (12, 14, 16 point). There was a significant effect of font type on reading time per web page, but not for font size or age group. There was also a significant main effect of font type and font size on reader preferences, but no effect of age group. These findings form the basis of recommendations for evidence-based web design guidelines for the Thai language. Keywords: Web Accessibility, Older Adults, Web Design Guidelines, Font Type, Font Size, Thai Language. 1 Introduction There are many web design guidelines for making websites more accessible to older people [1]. However, most of these guidelines lack evidence-based research to sup-port their recommendations. Furthermore, there is some confusion when web designers want to apply these recommendations, as the guidelines often provide different and even contradictory recommendations. Two examples of such confusion are font type and font size of text on webpages. Seven sets of guidelines make recommendations about font type: Zhao [2] recommends choosing font type based on legibility without explaining what that means; Agelight [3] similarly suggests choosing font type based on familiarity and legibility ; the SPRY Foundation [4], Holt [5], SilverWeb [6], the National Institute on Aging/National Library of Medicine (NIA/NLM) [7], and the Nielsen Norman Group [8] all recommend using a sans serif font type; SilverWeb [6] warns against the use of fancy font types; and NIA/NLM [7] warns against the use of serif, novelty, and display font types. K. Miesenberger et al. (Eds.): ICCHP 2014, Part I, LNCS 8547, pp , Springer International Publishing Switzerland 2014

2 Web Accessibility for Older Readers: Effects of Font Type and Font Size 333 The recommendations about font size are also varied. WebCredible [9] suggests that less than 12 point is too small to read but do not recommend a specific font size; the Nielsen Norman Group [8] and Zhao [2] recommend at least 12 point; Agelight [3], the SPRY Foundation [4] and SilverWeb [6] recommend 12 to 14 point; while NIA/NLM [7] recommends 12 or 14 point. Petrie, Kamollimsakul and Power [1] found that most web design guidelines are in English and relate to text presented in the Latin alphabet. Some research is beginning to investigate text presentation on the web in other languages such as Chinese [10], and Hebrew [11], but theree are no web design guidelines which have been developed for the Thai language or for Thai disabled and older people. This paper investigated the effect of font type and font size on skim reading tasks for Thai younger and older adults both using both objective and subjective measures with a view to developing evidence-based recommendations for websites in Thai. 2 Method 2.1 Design A three way mixed design was used, with age group as the between participants inde- vs pendent variable (younger vs older adults), and font type (two levels: conservativee modern font types) and font size (12, 14, and 16 point) as the two within participants variables. Two dependent variables are discussed here: time spent reading each webtype page, and participants ratings of their preferences, for each combination of font and font size (measured on 5 point Likert items). Fig. 1. An example of Thai conservative font type (left) and modern font type (right) The Thai Conservative font type includes an extra circle at the beginning of most of consonants and vowels while the Modern font type does not, as shown in Fig 1. Thus, the Thai Conservative font type corresponds most closely to a serif font in the Latin alphabet and the Thai Modern font type correspond most closely to a sans serif font. Each participant undertook six skim reading tasks on a specially created website about the Olympic Games, one task with each of the six combinations of font type and font size. The task was to skim read the page as quickly and accurately as possi- was chosen as this is a very common reading task on websites. Order of presentation of the combinations was counterbalanced between ble and to then answer four multiple choices questions about the page. Skim reading participants

3 334 S. Kamollimsakul, H. Petrie, and C. Power 2.2 Participants There were 42 participants, 21 older and 21 younger adults. All participants were native speakers of Thai. The mean age of the older adults was years (standard deviation 3.32, range years), 11 were male and 10 were female. The mean age of the younger adults was years (standard deviation 4.92, range years), 8 were male and 13 were female. 2.3 Equipment and Materials Participants completed the experiment on a personal computer running Windows XP and Internet Explorer 9 with 21.5 inch LED Monitor, a standard keyboard, and 2-button mouse with a scroll-wheel. A website about the Olympic Games with six different pages of text was created for the study. Each page had 354 words of text, separated into four paragraphs. There were four multiple choice questions for each page. 2.4 Procedure Participants were briefed about the study and completed an informed consent form and a short demographic questionnaire. Participant were asked to sit approximate 57 cm from the monitor. They were then asked to make themselves familiar with the computer, monitor, mouse, and web browser. The participants were given a practice task to familiarize themselves with the task required in the study. For each experimental trial, participants skim read the page with the appropriate combination of font type and size, pressed the spacebar when they had finished reading, and answered the questions. This process was repeated for each of the six tasks. After completing all the tasks, participants were asked to rate their preference for each combination of font type and font size on a 5 point Likert scale (1 = not at all preferred, 5 = very preferred). As a reminder, examples of the combinations of text font and size were provided. 3 Results An Analysis of Variance (ANOVA) on the time spent per webpage found that there was a significant main effect for font type, (F (1,40) = 11.05, p < 0.005), while the main effects for font size and age group were not significant. There were no significant interaction effects between any of the variables. Fig. 2 illustrates the mean for conservative and modern font types and shows that the conservative type produced significantly shorter reading times than the modern type.

4 Web Accessibility for Older Readers: Effects of Font Type and Font Size 335 Fig. 2. Mean time spent per webpage for Conservative and Modern fonts On the preferences ratings, an ANOVA found that there was a significant main efp < fect for font type, (F (1,40) = , p < 0.001), and font size, (F (2,80) = 81.09, 0.001), but no main effect for age group. All participants (both young and older adults) significantly prefer the conservative font type (mean preference rating: 3..75; standard deviation: 1.08) over the modern font type (mean preference rating: 2..20; standard deviation: 1.07). In investigating the main effect for font size, an LSD post hoc analysis found that 12 point was rated significantly lower in preference (p < 0.001) than 14 point and 16 point. While 14 point was rated significantly lower in preference (p < 0.05) than 16 point (mean preference rating 12 point: 2.12 standard deviation 1.21; 14 point: 3.21 standard deviation 1.18; and 16 point: 3.58 standard deviation 1.12). In addition, t-tests were conducted to investigate whether ratings of preference were significantly above or below the mid-point of the rating scale, to investigate whether participants were positive or negative about each font type or font size. One sample t- tests against the neutral mid-point rating of 3 showed that the ratings for the conserva- the tive font type were significant higher than neutral (t (125) = 7.75, p <.01), but ratings for the modern type were significant lower than neutral (t (125) = 8.44, p <.001). In addition, the ratings for 12 point were significant lower than neutral (t (83) = 6.69, p <.001) while 16 point was rated significant higher than neutral (t (83) = 4.77, p <.001) but 14 point was not significant different from neutral (t (83) = 1.66, n.s.) There was significant interaction effect between font size and age group, (F (2,80) = 5.13, p < 0.01). Fig. 3 shows the means for the interaction. A Scheffé post hoc anal- (p ysis showed that younger adults significantly preferred 14 or 16 point over 12 point < and p < 0.05, respectively). There was no significant difference between 14 point and 16 point. While older adults show a clear significant preference for 16 point over 14 point and 12 point (p < 0.05 and p < 0.001, respectively) and they significant- t for ly preferred 14 point over 12 point (p < 0.001). In addition, one sample t-tests against the neutral mid-point rating showed that both younger and older adults 12 point was significant lower than neutral (younger adults: t (41) = 3.03, p <.01; older adults: t (41) = 7.49, p <.01) while 16 point was rated significant higher than neutral (younger adults: t (41) = 2.76, p <.01; older adults: t (41) = 3.94, p <.01) but 14 point was not significantly different from neutral (younger adults: t (41) = 1.86, n.s.; older adults: t (41) = 0.426, n.s.).

5 336 S. Kamollimsakul, H. Petrie, and C. Power Fig. 3. Mean preference ratings for younger and older adults for 12 14, and 16 point textt There is also a significant interaction between font type, font size, and age group, (F (2,80) = 7.67, p < 0.001) ). Fig. 4 illustrates the mean for the interaction for younger adults for conservative and modern fonts and 12, 14 and 16 point text. A Scheffé post hoc analysis showed that, for younger adults, 12, 14, and 16 point conservative font type was not different from each other in preference rating. There was a significant difference in rating in modern font type: 12 point was rated significantly lower than 16 point (p < 0.005) but there was no significant difference between 12 point and 14 point, or between 14 point and 16 point. In addition, 12 point Modern font type was significant lower in rating than every size of Conservative font type (12 point (p < 0.005); 14 point (p < 0.001) ); 16 point (p < 0.001)). One sample t-tests against the neutral mid-point of 3 showed that for younger adults, the mean rating for all point size in Conservative font type were significant higher than neutral (12 point: t = 2.22, df = 20, p <.05; 14 point: t = 6.14, df = 20, p <.001; 16 point: t = 3.44, df = 20, p <.005), but the mean rating for Modern font type Fig. 4. Mean preference ratings for younger adults for Conservative and Modern fonts and 12, 14 and 16 point text

6 Web Accessibility for Older Readers: Effects of Font Type and Font Size point was significant lower than neutral (t = 14.98, df = 20, p <.001) while 14 point and 16 point were not significant difference from neutral (14 point: t = 1.82, df = 20, p >.05; 16 point: t = 0.46, df = 20, p >.05). Figure 5 illustrates the mean for the interaction for older adults for Conservative and Modern Fonts and 12, 14 and 16 point text. A Scheffé post hoc analysis showed that, for older adults, Conservative font type, 12 point was rated significantly lower in preference than 14 point and 16 point (p < and p < 0.01, respectively). While 14 point and 16 point were not significant difference (p > 0.05). For Modern font type, 12 point was rated significantly lower than 14 point and 16 point (p < 0.05 and p < 0.001, respectively) but there was no significant difference between 14 point and 16 point (p > 0.05). In addition, 12 point Modern font type was significant lower in rat- ing than every size of Conservative font type (12 point (p < 0.05); 14 point (p < 0.001); 16 point (p < 0.001)). 14 point Modern font type was significant lower in rating than 14 and 16 point Conservative font type (p < 0.005). 16 point Modern font type was significant lower in rating than 16 point Conservative font type (p < 0.05) ). One sample t-tests against the neutral mid-point of 3 showed that for older adults, the mean rating for Conservative font type 12 point was significant lower than neutral (t = 2.26, df = 20, p <.05) while 14 point and 16 point were significant higher than neutral (14 point: t = 4.95, df = 20, p <.001; 16 point: t = 10.03, df = 20, p <.001). For Modern font type, 12 point and 14 point were significant lower than neutral (12 point: t = 20.61, df = 20, p <.001; 14 point: t = 4.18, df = 20, p <.001) while 16 point were not significantly difference from neutral (t = 1.16, df = 20, n.s.). Fig. 5. Mean preference ratings for older adults for Conservative and Modern fonts and 12, 14 and 16 point text 4 Discussion and Conclusions This study found that Thai font type had a significant effect on skim reading time, but font size did not and perhaps surprisingly, there were no differences in reading times between the two age groups. The conservative font yielded significantly quicker skim

7 338 S. Kamollimsakul, H. Petrie, and C. Power reading times across all other variables, with a 6.39 second or 7.3% increase in reading efficiency compared to the modern font. On participants preferences for font type, both younger and older adults preferred the conservative font type over the modern font type. For font size, the results are more complex. In general, younger adults significantly preferred 14 and 16 point over 12 point but expressed no difference in preferences between 14 and 16 point. Older adults preferred 16 point over 12 and 14 point and preferred 14 point over 12 point. In addition, there was an interaction between font type, font size, and age group. For the modern font, the pattern of preferences is similar for both younger and older adults, with 12 point rated very low by both groups. Even 16 point only reached the midpoint of the scale for older adults whereas both 14 and 16 point reached the neutral point for younger adults. However, the conservative font is generally acceptable to the younger adults, it was rated higher than the neutral midpoint for all point sizes. But for the older adults, there was a clear preference for the conservative font at 14 and 16 point over 12 point with 14 and 16 point rated higher than neutral, and 12 point rated lower than neutral. Based on the results from the present study, proposing the conservative font type is good for both younger and older adults. While the recommendation about font size is related to font type. If web developers use the conservative type, 12 point or larger is appropriate for younger adults while 14 point or larger is appropriate for older adults. When web developers use modern font type, 14 point or larger is acceptable for younger adults and 16 point is acceptable for older adults. These results are very unexpected if one is making predictions from web design guidelines for the Latin alphabet. Thus they provide the first evidence for web design guidelines for Thai websites, with useful evidence for developing websites for both younger and older readers. Acknowledgements. We thank all our participants in the study for their time and effort. We also thank Institute of Social Technology, Suranaree University of Technology, Nakhon Ratchasima, Thailand for providing the room in which to conduct the study. Finally, we thank the National Science and Technology Development Agency (NSTDA), the Royal Thai Government for the scholarship, which supports the first author in his Ph.D. programme at the University of York, UK. References 1. Petrie, H., Kamollimsakul, S., Power, C.: Web accessibility for older adults: effects of line spacing and text justification on reading web pages. In: Proceedings 15th International ACMSIGACCESS Conference on Computers and Accessibility (ASSETS). ACM (2013) 2. Zhao, H.: Universal Usability Web Design Guidelines for the Elderly (Age 65 and Older), 3. AgeLight: Interface design guidelines for users of all ages,

8 Web Accessibility for Older Readers: Effects of Font Type and Font Size Holt, B.J., Komlos-Weimer, M.: Older Adults and the World Wide Web: a Guide for Web Site Creators. SPRY Foundation, 5. Holt, B.J.: Creating senior-friendly Web sites. Issue Brief, Centre for Medicare Education, 1(4) (2000) 6. Kurniawan, S., Zaphiris, P.: Research-Derived Web Design Guidelines for Older People. In: Proceeding 7th International ACMSIGACCESS Conference on Computers and Accessibility (ASSETS), Maryland, pp (2005) 7. National Institute on Aging/National Library of Medicine (NIA/NLM): Making your website senior friendly: A check list, 8. Pernice, K., Estes, J., Nielsen, J.: Senior Citizens (Ages 65 and older) on the Web. Nielsen Norman Group (2013) 9. Fidgeon, T.: Usability for Older Web Users, user-friendly-resources/web-usability/older-users.shtml 10. Chan, A.H.S., Lee, P.S.K.: Effect of display factors on Chinese reading times, comprehension scores and preferences. Behaviour & Information Technology 24(2) (2005) 11. Parush, A., Shwarts, Y., Shtub, A., Chandra, M.J.: The impact of visual layout factors on performance in Web pages: A cross-language study. Human Factors 47(1) (2005)