Experiments on the Presentation of On-Screen Information
Previous research had indicated that visually impaired people (VIP) can
use systems such as interactive digital television. Their requirements
for certain features and how these features are displayed is likely to
vary because VIP are not a homogeneous population. If the possibility
to adapt the interface to their needs is available, how would VIP like
information displayed and presented to them? This report is an investigation
into their preferences for presenting certain visual features e.g. text
size, contrast, flashing or scrolling and icons size and shape. It will
also look at a few screenshots and how VIP regard them.
5.1 Method
5.1.1 Subjects
Some of the subjects were recruited from a database of visually impaired people willing to help with scientific research. Others were approached in the RNIB resource centre and asked if they would be willing to participate. Only partially sighted subjects were chosen because the experiments focused on the visual presentation of on-screen information.
5.1.2 Procedure
The experiments were conducted in the RNIB research library on a computer with a 15 inch monitor. Before a subject arrived, the tests were set up to facilitate a smooth progression through the session. An experimental script was used to ensure the subjects were told the same information and asked questions in a uniform way [appendix 5.1].
To commence, the subjects were briefed about the purpose of the experiments and their data protection rights and they were asked if they had any questions or if they were ready to start. A few questions were posed to get some background details about the subject. Then the subject's basic level of vision was ascertained. This included an acuity test and questions about their level of vision.
Acuity was measured using different sized characters displayed on the monitor [appendix 3.3]. This was always set to display at 100% magnification so that it was the same size for every subject. Figure 5.1 shows what line the letter was on and the size it was displayed at. The subjects were all positioned at 2.70m from the screen for this test.
Figure 5.1 Table to show the x-height of characters on each line in the acuity test
|
Line number
|
Height on screen (mm)
|
|
1
|
40
|
|
2
|
24
|
|
3
|
15
|
|
4
|
13
|
|
5
|
8
|
|
6
|
6
|
|
7
|
5
|
|
8
|
4
|
|
9
|
3
|
|
10
|
2
|
Subjects were then asked what was the cause of their visual impairments and where in their visual field they could not see when they were looking straight ahead of them e.g. left, right, up, down, patchy across the field and centrally.
The subject was shown a screenshot of a sports match and asked to position their chair where they usually would in relation to a screen as shown. Then their distance from the screen was measured.
A contrast test was then used to decipher their level of contrast sensitivity [appendix 5.2]. Simple four letter words with an x-height of 13mm were displayed in the centre of the screen against a black background. The contrast of the words was altered along the grey scale as a percentage e.g. 100% contrast was white on black, 75% was a light grey on black. The exact values were calculated from the grey scale used in the colour feature in Word. Because of the way this test was created, all the subjects were shown the words in the same order. The percentage difference in the shades of grey provided an indication of the level of contrast they required to distinguish between text and its background.
Different sizes of text as shown in figure 5.2 were presented in white on black, 100% contrast [appendix 5.3]. Subjects were asked their preferences for the smallest, largest and preferred sizes. Two columns of text were presented [appendix 5.4], with small text at the top with an x-height of 5mm and a larger version of the same text below. The left column had enlarged characters with an x-height of 9.5mm and the right column with characters that had increased vertical height which had an x-height of 12mm. Subjects were asked which column they preferred.
Figure 5.2 Table to show the x- height of characters on each line in the text size test
|
Line number
|
Height on screen (mm)
|
|
1
|
4
|
|
2
|
4.5
|
|
3
|
5
|
|
4
|
5.5
|
|
5
|
6
|
|
6
|
6.5
|
|
7
|
7
|
|
8
|
7.5
|
|
9
|
8
|
|
10
|
8.5
|
|
11
|
9
|
|
12
|
9.5
|
|
13
|
10
|
|
14
|
10.5
|
|
15
|
11
|
|
16
|
11.5
|
|
17
|
12
|
|
18
|
12.5
|
|
19
|
13
|
|
20
|
13.5
|
|
21
|
14
|
|
22
|
14.5
|
A sentence of flashing text with an x-height of 8mm was shown on the screen [appendix 5.5]. Subjects were asked if they could read the text and if they liked it flashing. They were then presented with the scenario that the text could flash approximately 3 times and then stop. This would call their attention to it then stabilise so that they could read it. They were then asked whether they would mind this happening.
The next test concentrated on scrolling text created in HTML [appendix 5.6]. Two independent variables, size and speed were altered. The 2 speed levels were equivalent to speed 3 and 5 in HTML. There were 3 levels of size which were equivalent to an x-height of 6mm, 9mm and 12 mm. Each level of size was displayed at both speeds so 6 conditions were created. This was a within-subjects design with the subjects being shown ever condition. The order the sentences were presented in was randomised. The scrolling text was offered and the subject asked to read the sentence when the experimenter said 'Go' as it appeared on the screen. Care was taken to ensure the text remained hidden until the moment the subject was asked to read it. The screen was set up to reveal a fraction of the window (but not the text), then the refresh button was pressed whilst the screen was maximised to reveal the text. The stop clock to record the time was always started after the second word had appeared on the screen (all the sentences started with the same words). The first sentence was a practice one. At the end, the subject was asked if (s)he would prefer for the text to be stationary.
If navigation cues are given on TV screens, they tend to be presented as icons so the shape and size of icons was tested. Three shapes, a square, circle and triangle were presented and the subject asked which they could see the most clearly [appendix 5.7]. This was done at 3 different sizes, 18mm, 23mm and 30mm top to bottom. The order these were presented in was randomised between the subjects. Then the shapes - 32mm top to bottom - were presented with the word OK inside them, the text inside had a x-height of 8mm. Again the subject was asked which they could read most clearly.
Lastly, four screen-shots - taken from the Broadband Bananas video archive - were presented and the subject asked what they thought about the presentation. It was explained that some of the images were not of the best quality in this test. They were prompted to comment on the colour, legibility, layout etc. if they were not forthcoming with remarks.
If they had not completed a questionnaire about features they might require, they were requested to do so [appendix 4.1]. Finally they were debriefed and thanked for their participation.
5.2 Results
A total of 32 subjects took part in these experiments, 22 were males and 10 were females. The minimum age was 16 and the maximum was 82. There was a mean age of 46 with a standard deviation of 20.7. The table below shows the types of television present in the subject's households.
Figure 5.3 Table to show the televisions in subject's households
|
Television Type
|
Frequency of Subjects
|
| No TV |
2
|
| Analogue |
16
|
| Satellite and Analogue |
6
|
| Cable and Analogue |
6
|
| Terrestrial and Analogue |
2
|
The subjects displayed a range of visual impairments. The most common visual impairment was nystagmus which 7 subjects reported, 6 subjects said they had Retinitus Pigmentosa, cataracts or glaucoma, 2 of those with glaucoma also had cataracts. Other subjects reported their VI was due to: albinism and nystagmus, cone rod dystrophy, ushers and an underdeveloped optic nerve. One person had an accident and another a tumour. Eight people in this sample had more than one visual impairment.
Subjects' responses to a question about their level of colour perception gave an indication of their colour perception. Twenty subjects thought they were not colour blind whereas 6 knew they were and 6 thought there colour perception was not perfect.
For the acuity test, 4 subjects could not read any of the letters displayed on the screen. The table below shows how many subjects could read letters displayed at different sizes similar to a Snellen chart.
Figure 5.4 Table to show the results of the acuity test
|
Height on screen (mm)
|
Frequency of Subjects
|
|
40
|
12
|
|
24
|
4
|
|
15
|
-
|
|
13
|
6
|
|
8
|
5
|
|
6
|
2
|
|
5
|
-
|
|
4
|
-
|
|
3
|
-
|
|
2
|
-
|
The distance of the subjects from the monitor was measured whilst they were watching a picture of a sports match. The minimum was 20cm whereas the maximum was 270cm. The average distance was 105.3cm with a standard deviation of 69.7.
For the contrast test, 1 subject could not read the text at the size it was displayed (13mm) so no results were recorded for them. Ten subjects needed at least a contrast of 20% difference between black and white. Six subjects could cope with a 30% difference, 4 with a 40% difference, 5 with a 50% difference and 6 needed a difference of 60% between black and white. This shows a range from 20% to 60% with an average of approximately 40%.
For small amounts of information (not for continuous text), the table below shows the smallest size text that subjects were prepared to read. The smallest text size people were prepared to read was 5mm and this ranged up to 14.5mm. This indicates a wide range of 'small' text sizes that could be applied, some of which are not small at all.
Figure 5.5 Table to show the smallest text size subjects preferred
|
Smallest Size of Text (mm)
|
Frequency of Subjects
|
|
5
|
2
|
|
5.5
|
4
|
|
6
|
2
|
|
6.5
|
4
|
|
7.5
|
2
|
|
9.5
|
4
|
|
10.5
|
4
|
|
11.5
|
2
|
|
12.5
|
2
|
|
14.5
|
6
|
Similarly, the largest text size that people thought they could comfortably read are described in the table below.
Figure 5.6 Table to show the largest text size subjects preferred
|
Largest Size of Text (mm)
|
Frequency of Subjects
|
|
10.5
|
2
|
|
12
|
2
|
|
12.5
|
4
|
|
13
|
2
|
|
14
|
2
|
|
14.5
|
8
|
|
larger than the largest displayed
|
12
|
The largest text size that people were prepared to read started at 10.5mm and went beyond the 14.5mm which was the highest size that was tested. These results indicate that one size does not fit all. The 'smallest' text some people could read was larger than the 'largest' other people were prepared to read.
Figure 5.7 Table to show subjects preferred text size
| Preferred Size of Text (mm) | Frequency of Subjects |
|
7.5
|
2
|
|
8.5
|
2
|
|
9.5
|
4
|
|
10.5
|
4
|
|
11.5
|
2
|
|
13.5
|
4
|
|
14.5
|
14
|
The preferred text size was also across a wide range as shown in the table above. The preferred text size fell between 7.5mm and 14.5mm. This range overlaps with the highest end of the 'small' text size and the whole of the 'large' text size range.
Unsurprisingly Spearman's correlation between the text sizes shows that people that wanted the larger 'small' text sizes also wanted the 'larger' large text sizes. There was a positive substantial correlation, rs = 0.85, n = 32, p<0.01.
When shown different methods of presenting enlarging text, 20 subjects preferred the left hand column (enlarged text) and 12 preferred the right column (vertically expanded text). There was not a significant difference between the number of people who preferred each column, X2 (1) = 2.0, P>0.05.
For the flashing text, half the subjects (n=16) could read the text and half could not. When asked if they liked the text flashing, 28 subjects replied 'No' and only 4 did not mind it. A Chi-Square Test showed this was statistically significant, X2 (1) = 18.0, p<0.01. When asked if they would mind if it flashed about 3 times and then stopped, 26 people said 'No' and only 6 said 'Yes'. Again, this was statistically significant, X2 (1) = 12.5, p<0.01.
Regarding subjects who wanted to stop the flashing text, 18 wanted to also stop scrolling text but 10 did not. Of those who did not mind flashing text, 2 wanted to stop scrolling text and 2 did not. So these functions cannot be grouped together because those who did not want one function did not necessarily not want the other function.
The next experiment tested 2 variables, the 'speed' and 'size' of scrolling text. The 'reading speed' was calculated by subtracting the time taken for the whole sentence to appear on the screen as this differed slightly depending on the size. The Mauchly's Test of Sphericity was significant for the interaction of 'speed and size' so these significance levels can not be accepted. The 'speed' effect only had 2 levels and so was not applied to this test and the 'size' effect was not significant so its results can be accepted.
The between subject effects and main effect of 'speed' was not significant. Despite this some subjects (n=8) thought that even the fastest speed was too slow. Another 7 subjects thought that the slower speed was a nice pace. Only the factor 'size' had a significant effect on the reading speed of the scrolling text, F (2) = 50.8, p<0.01. From the means shown in figure 5.8 the reading speed was faster as the text size increased so the largest sized text was read the quickest. The subjects were observed to be most comfortable reading the largest size scrolling text. A couple of subjects mentioned that the scrolling text appeared to flicker to them.
Figure 5.8 Table to show the average reading speed by size
| Text | Mean Reading Speed (s) |
| Small | 5.8 |
| Medium | 3.4 |
| Large | 2.2 |
When asked if they would like the text to stop scrolling completely, 20 subjects replied 'Yes' and the other 12 said 'No'. There was not a statistically significant difference between these responses, X2 (1) = 2, p>0.05.
For the smallest sized shapes, when asked which they could see the clearest, the majority replied the square [n = 24]. Four said the circle, another 4 said 'any' and no one thought the triangle was the clearest. A Chi-Square test showed these results were significant, X2 (2) = 25.0, p<0.01.
For the medium sized shapes, when asked which they could see the clearest, the majority replied the square, n = 22. Four said the circle, another 4 said 'any' and only 2 thought the triangle was the clearest. A Chi-Square test showed these results were significant, X2 (3) = 33.0, p<0.01.
For the largest sized shapes, similar results occurred. 18 subjects thought the square was clearest, 8 said the circle, 4 said 'any' and only 2 thought the triangle was the clearest. A Chi-Square test showed these results were significant, X2 (3) = 19.0, p<0.01.
When asked which shape had the clearest writing within it, the majority replied the circle, n = 14. Eight said the square, another 8 said none and only 2 thought any of them were clear. A Chi-Square test showed these results were significant, X2 (3) = 9.0, p<0.05.
For the final part of the session, subjects were asked to comment on the presentation of various screen-shots from iTV. These images were displayed full screen and were presented in the same order from the simplest to the most complex.
Screen-shot 1:
The most iterated statement from the visually impaired subjects about this image was that they could not read the information in the box. Six subjects said the information would be in a better position at the bottom of the screen because it was important to be consistent. Four subjects actually missed the box completely and had to be prompted to look for it. One subject had their nose on the screen to decipher the information box. The red icon especially was not obvious to six subjects and another four subjects found the red icon annoying. The colour contrast within the box was thought to be poor. Another subject commented that "I would have to know how the remote operates too."
Screen-shot 2:
"Ah, crumbs" Twenty-six subjects mentioned that the print was too small on this screen. The next main complaint which was from twenty-two subjects was that the colour contrast was very poor especially the text in black on red. The overall contrast was though to need improving. This led to the comments about speech being needed. It was mentioned that it was not intuitive what should be pressed. A couple of subjects mentioned they did not know where to look. The uniform position of information was highlighted as an important consideration again. One subject said they did not like the mixture of text and images because it made it hard to pick out the main features of either. The concluding remark of one subject was "I'm not going to use this."
Screen-shot 3:
The main comment about this screen was that the background was patterned - which was regarded negatively - this was repeated by 19 subjects. It made it more difficult to read the text positioned on it. "What is the red writing on the left?" The text was thought to be too small by six of the subjects. Overall these issues gave the screen-shot the impression to 10 subjects that it was too busy. Ten subjects commented that the coloured button icons on this screen were OK, some also said that the writing underneath them was not, it should not have been coloured. One subject would have preferred the icons to be larger. Another mentioned that they found it very difficult to see the button icons especially the red one. Having the programme maintained on the screen was seen as a positive thing by 6 subjects. This screen received a mixed response. Some subjects thought it was "Awful" whereas other thought it was "Pretty, but the background is no good."
Screen-shot 4:
Twelve subjects appreciated the contrast of the background with the foreground. Although another six thought there was too much white, which made it too bright. The text was too small and needed more contrast. This was especially in relation to the numbers in the list at the bottom left. "This would be difficult to use at a distance." Overall six subjects thought there was too much information for them "Oh my God, this is the web not TV". Whereas other subjects thought that overall the screen was understandable "Oh, this is the best one!". Another four commented that if the information was kept in the same position and context that they would try to get used to it. The red and green icons meaning was lost on some subjects because the colours red and green where used and they were unsure what they related to. Not one subject was expecting a panel requesting their vote on a specific topic so some people found this confusing. The highlighting had a good contrast to the rest of the screen so was commented on. On this screen subject's responses were mixed, which partially depended on their expectations. This resulted in some subjects saying they would not use such a TV screen. Some features were liked whereas others were regarded as needing improvements.
5.3 Discussion
From the results described in this report, contrast necessities differ from person to person. Generally the higher the contrast the more people can read the information. Depending on the text size, a contrast of 60% was adequate for this sample of VIP.
In terms of text size, the situation becomes more complicated. From the results:
The results were skewed towards the largest text sizes in each case so triangles have been used to represent this. It will not be optimum to determine one size text for use of interactive digital television. Instead a range of available sizes would be more appropriate for the users. Within this experiment and from the acuity test results, it is thought that a range from 8mm to 14.5mm will suit most subjects. This range will need to be tested again with subjects in context. With other information on the screen, this may not be suitable.
The results of the test to determine the way to enlarge text was inconclusive.
If text was flashing at all, it is concluded that it should flash a few times and then stop. This will call the person's attention to the information and then stabilise the text so the person can read it. Some subjects in this experiment would prefer for flashing text to never be seen so the option to disable it should be allowed. This was reinforced by a comment from a subject who had epilepsy.
Just because people did not want one function did not necessarily mean that they did not want another function. This was shown by an option to stop scrolling and / or flashing text.
In terms of scrolling text, speed did not have a significant effect, but size did. The speed people preferred differed. The larger the text, the quicker the subjects could read it and the more comfortable they were doing so. The results were inconclusive about whether scrolling text was acceptable. A couple of subjects mentioned the scrolling text looked like it was flickering. This leads to the conclusion that the option to disable this feature should be available so that the user can decide for themselves.
If icons are represented by shapes, the clearest shape reported was the square. This occurred regardless of the size of the square. However when there was text inside the shape, the circle was more appropriate. Text within the shape seemed to be less legible so an alternative way of coding the information is important.
It is important for information to be presented in the same position on different screens. This consistency enables VIP to check to see if there are further options available easily otherwise they can miss them completely.
The size of the text was mentioned frequently and is therefore an important consideration. Contrast is a very important variable too. Patterned backgrounds reduce the legibility of the whole screen significantly. Red is a colour to be avoided to display or code information and should never be used in relation to green. If only visual, the coding of information is lost on people who can not see the visual information so redundant coding is necessary.
High information density is not favourable. Information needs to be clearly and concisely presented. Any new features need to be intuitive and easy to use.
5.4 Conclusion
The visually impaired people that participated in these experiments were enthusiastic and eager to have features that would increase their access to interactive digital television. By providing the option to set their preferences, this could become a possibility.
From the research described in this report and previous research, it is necessary for certain features to have a range available e.g. Text size.
Some features should have the option to be disabled e.g. flashing text and scrolling text.
Information that is represented in a purely visual medium also needs another form of coding.
5.4.1 Limitations
After transference to a computer system, the screen shots were not as clear as they would have been if displayed on a television monitor.
5.4.2 Further work
Further work is thought to be necessary into the range of text sizes that people could use. This will eventually need to be applicable in context of a dynamic screen. The methods to enlarge this text will also need to be researched.
The contrast of colours needs to be explored.
The density of information is an area that has been neglected for too long.
| Previous Page | Contents Page | Next Page |
Last updated: 20.11.2009 © Copyright reserved Website design: Digital Accessibility Team