Discussion

Age and Vision

Consistent with the findings of previous research (e.g., Kline & Scialfa, 1996; Owsley & Sloane, 1990) and as hypothesized (Hypothesis 1) Spatial vision of older people in this study was generally inferior to that of their younger counterparts. As in prior studies, the acuity of the elderly observers was worse than that of the young observers for both the near and far test distances (e.g., Gittings & Fozard, 1986; Pits 1982) and an age deficit was observed on both near and far measures of contrast sensitivity at intermediate and high spatial frequencies (e.g., Elliott, 1987; Scialfa et al., 1992). The only exception to this general trend was the absence of significance for the large mean age difference on the far contrast sensitivity measure at 24 c/deg. This reflects the high variability in the measure at this spatial frequency, in conjunction with the relatively low number of observers used in this study (24). Composite measures of contrast sensitivity extracted from the CSF also showed an age-related decline. The Mean Contrast Sensitivity and the Mean Peak Contrast Sensitivity were significantly higher for young observers for both the near and far measures. Although not significantly different, the spatial frequency at which peak sensitivity occurred on the near CSF, was somewhat lower for the older participants. These findings are generally consistent with previous studies (e.g., Owsley et al., 1983). On the far measure, however, the spatial frequency at which peak sensitivity occurred was not different for the two age groups.

As might be expected, older participants more frequently reported general visual as well as health symptoms than did young participants. Elderly people were more likely to need corrective lenses (12 vs. 5 participants), particularly of the multifocal type (7 vs. 0 participants). Consistent with this and the normal age-related loss of accommodation (i.e., presbyopia) most of the older participants (10 of 12) used corrective lenses for reading. None of the younger participants reported using a reading correction. No serious chronic visual diseases were reported by the participants in either age group.

As has been observed in prior studies (e.g., Kosnik et al., 1988; Kline et al., 1992), elderly participants appear to have considerable knowledge regarding some of their visual limitations. For example, older people rated both their near vision and the ease of reading near materials more poorly and reported having more problems with their near vision than did young participants (see Table 3), an insight validated by significant relationships with the objective measures of their near peak and mean contrast sensitivity, as well near mean acuity.

Age and Self-Reports on Reading

Although the difference was not significant, the elderly participants reported reading more hours per day than did the younger participants. This was somewhat surprising, considering that the young observers were primarily university students in excellent visual health. In addition the older people, as noted above, were more likely to report difficulty in reading close material, a problem that correlated strongly with their near acuity and contrast sensitivity. The unexpected parity between the age groups on the reading measures may have been attributable to non-visual factors. The older participants were as well educated as the young group, and their measured vocabulary scores were just as high. They were also in good general and visual health, and most of them being retired, had more leisure time available to them. They also were as likely to be motivated as their younger counterparts to read in that they rated the importance of both near and far reading at about the same level as younger readers.

Age and Reading Performance

Legibility: Consistent with Hypothesis 2, the elderly participants required larger letters than the young to achieve legibility. This was not surprising given the clear support as hypothesized (Hypothesis 7), for a robust relationship between Acuity and Legibility. Far Mean Contrast Sensitivity across the two age groups was also related to Mean Legibility Size, although, significantly less so than acuity (see Table 17). Some of the composite measures of contrast sensitivity approached a significant relation with Legibility Size for younger participants, but none did so for older participants. In fact, once Mean Acuity had been considered, these other visual measures added little to the predictability of legibility (see Figure 12). The importance of acuity in affecting legibility, and legibility in turn, determining reading speed was also seen in the description of the findings of this study using structural modeling. Acuity explained 80% of the Legibility variance, which in turn accounted for 22% of the Reading Time variance. The practical implications of this for older readers concerned about their reading speed is to optimize their acuity for the reading distance (e.g., through good optical correction) and to select or adjust font sizes, font type, and reading distance to maximize print legibility.

Readability: No age differences were found in the time taken to read prose passages, nor were any of the visual measures related to reading time either within an age group or across them. The relative independence of this task from acuity may reflect that participants were allowed to hold the material at their most comfortable reading distance. Although longer reading times were expected from the elderly readers (Hypothesis 2), this hypothesis was not confirmed. The experimental conditions used in this study may have minimized the likelihood of such an age difference suggested by the absence of any age differences in overall subjective ratings (i.e., across font) of Reading Ease, Reading Discomfort, Clarity of Type, or Font Preference. These relationships may have been significant if smaller print sizes and/or more adverse viewing conditions had been employed. That at least is implied by the strong performance of older readers despite their visual disadvantages. Other factors may also have minimized age differences in reading. Among them, reading was purposefully self-paced, and participants knowing that their comprehension would be evaluated at the completion of each passage, may have emphasized accuracy rather than reading speed.

As hypothesized (hypothesis 8), the time taken to read a passage was affected positively by high legibility. The two measures were positively correlated overall, and approached significance for the two groups separately, again highlighting the fundamental importance of legibility to reading performance.

Contrary to Hypothesis 3, no age differences were found regarding the distance at which fonts were held for reading, nor were any visual measures related to reading distance overall. Within age groups, non-significant, moderately high correlations (about .5) occurred between Reading Distance and Acuity in both the young and elderly groups. As might be expected, readers with better acuity generally held their material further away from them.

There was little evidence in this study that readers, young or old, discriminated between the various subjective dimensions used to assess fonts in this study. Reading Ease, Reading Discomfort, and Clarity of Type were all strongly associated with each other, both across and within age groups (see Tables 12-15). The only minor exception to this pattern was that the Reading Discomfort/Clarity of Type did not quite reach significance in the young group. The Complexity of fonts was related to Reading Ease, Reading Difficulty and approached a significant relationship with Clarity of Type for elderly readers, but not young ones, or the two groups combined.

Although not significant, Reading Time tended to be inversely related with the subjective rating measures for the two age groups (see Table 13). The younger readers who read passages more quickly tended to rate the fonts as easier to read, causing less discomfort, clearer and simpler. Perhaps attributing speed of reading to subjective ratings of fonts. Conversely, older readers with shorter reading times were more likely to rate the fonts as less easy to read, causing more discomfort, less clear, and more complex. Whatever the cause for the reversal, it appears to have been a purely subjective effect as it did not occur on the objective reading measures (i.e., Legibility, Reading Distance).

Do Fonts Make a Difference?

The findings of this study provide an affirmative answer to this question Specifically, on both objective and subjective measures Times New Roman, Bell Gothic and Sabon fonts consistently facilitated reading relative to other fonts and appeared to be particularly suitable for small print applications. Times New Roman Sans was an effective small-print font and, except for reader preference, belonged among the better fonts.

Contrary to the Hypotheses regarding the superiority of sans serif fonts (i.e., Hypothesis 4, 5 and 6), the presence or absence of serifs did not distinguish the better fonts from the worse ones. Two of the best fonts were serif types and two were sans serif fonts. This was particularly evident when comparing Times New Roman and Times New Roman Sans fonts, which differed only on the presence or absence of serifs, yet were comparable in regards to legibility, readability and reader ratings. These findings suggest, that charasterists other than the presence or absence of serifs are the fundamental determinants of font effectiveness. In accord with Hypothesis 9, two aspects of the fonts in particular, condensation and complexity, appeared to determine font effectiveness in small-print applications. Two condensed fonts (Univers Condensed and Garamond Condensed), and one highly complex font (Century Old Style) were consistently low on the performance measures. In addition to being condensed, Univers Condensed was also rated as an extremely simple font. It may be that extremes of simplicity/complexity also make fonts less usable. The effects of The Sans on reading was highly inconsistent. It was rated as a fairly simple font, high on Legibility, low on Reading Distance, Reading Ease, and Preference and intermediate on Reading Discomfort and Clarity of Print.

To determine what other characteristics of fonts might affect their utility, the relationships between percentage x-height (the percent of a fontís x-height to its cap-height), font contrast (the percentage of a fontís thin stroke width to its thickest stroke width), the length of a standard sentence printed in the same x-height (see Appendix G and H), vulnerability to low-pass spatial-frequency filtering (i.e., the amount of high frequency information that can be removed from a display and still be legible) and reading were examined. The outcome of these analysis are presented in Table 18. For all of the participants, fonts with greater difference between x-height and cap-height tended to be held further away to be read, but took the younger readers longer to read. Neither effect was significant. Percentage x-height was not related to any other measures, nor did the difference between thick and thin strokes (i.e., contrast) predict the performance of the fonts on any of the reading measures evaluated. Fonts that produced long sentences length were positively related to subjective ratings. They were rated as being easier to read, causing less discomfort, having better clarity of letters and were generally preferred. These same fonts also tended to be more legible, could be held farther away to be read, and could be read more quickly. This factor reflected greater "openness" of font (i.e., a combination of greater letter width, and increased spacing between letters and words). The low-pass filter thresholds of fonts related strongly to several of the reading measures. Consistent with previous research (Dewar, Kline, Schieber, & Swanson, 1994), low-pass filter threshold was highly related to Legibility. Fonts with greater resistance to such filtering (i.e., blurring) were rated as causing less discomfort, having clearer letters, and easier to read. They were also preferred, and tended to have a greater comfortable reading distance. Surprisingly, however, while fonts that were resistant to blurring were read faster by elderly readers, they were read more slowly by young readers. Possible explanations for this reversal of results may be attributable to visual limitations of the elderly participants and motivational differences between the age groups and are discussed later in this section. Presumably resistance to low-pass filtering also reflects the openness of a font (i.e., a combination of stroke width, inter letter spacing and openness of the counters (white space within letters) and is reflective of fonts that are not condensed. This again lends support to the recommendation that condensed fonts should be avoided.

Table 18: Overall Person Correlations Between Reading Measures and Objective Font Measures for all Participants  

% x-height

Contrast

Sentence Length

Low-pass Filter

Legibility

0.27

-0.31

-0.40**

.88*

Reading Distance

-0.58**

0.00

0.70**

-.43

Reading Time Young

-0.64**

-0.05

0.56**

-.58

Reading Time Elderly

0.26

0.12

-0.49**

.38

Reading Ease

0.20

-0.02

-.75*

.54

Reading Discomfort

0.28

-0.09

-.79*

.83*

Letter Clarity

0.19

-0.22

-.75*

.75*

Preference

0.35

0.18

-.92*

.65

* = Significant at p<.05 ** = Approaches significance

Contrary to expectation, age differences in reading were little affected by font characteristics. Age interacted with font on only two measures, Simplicity/Complexity and Reading Time. The significant age by font interaction on Simplicity/Complexity reflected the fact that younger readers rated serif fonts as more complex than sans serif fonts, while elderly readers did not distinguish font complexity based on serifs. Shorter reading times for younger people tended to be associated with fonts that yielded longer Reading Times for older readers. This effect was not significant after Geisser-Greenhouse correction. The absence of a font effect on reading rate in this study was in accord with some of the previous literature (e.g., Mansfield, Legge & Bane, 1996; Moriarty and Scheiner, 1984; Paterson & Tinker, 1932; Poulton, 1965; Smither and Braun, 1994). These findings disagree with studies that have found serif fonts to be either superior (Nolan, 1959; Prince, 1966; Vanderplas and Vanderplas, 1980) or inferior (Shaw, 1969; Sorg, 1985) to sans serif fonts. The reasons for such inconsistencies in font effect on reading speed are not clear. Perhaps the serif effects observed in earlier research were produced by other features of the particular fonts used, such as condensation or complexity. Such an interpretation at least is consistent with the impact of these variables in the present study.

Overall, preferred fonts were read from farther away, were rated as easier to read, causing less discomfort and having greater clarity (see Table 14). Fonts that were also more legible, were associated with less rated discomfort while reading. The positive relationship found earlier among subjective reading measures across participants (i.e., Reading Ease, Reading Discomfort and Clarity of Letters) were also seen when examined across fonts, and all three of these variables were inversely related to Reading Distance. The relationship between Reading Time and Legibility that was seen across participants, however, did not occur when analyzed across fonts.

When possible age differences in the relationships between fonts and measures of reading were considered on a font-by-font basis, font characteristics were generally more important for elderly readers than younger ones. While 19 of the correlations, indicative of these relationships (see Table 15), were significant for elderly participants, only 14 were significant for young participants. Inspection of these correlations reveals a reversal in the relation between reading time and other dependent measures between young and elderly participants. Although this apparent reversal is without obvious explanation, possible explanations relate to possible age differences in visual comfort, emphasis on comprehension, and visual problems experienced in reading. As expected, among older readers, fonts that took less time to read, were more legible, read from further away, rated as being easier to read, causing less discomfort, having clearer, simpler letters, and were preferred. This relationship might have been mediated by the greater visual limitations of older readers (e.g. reduced acuity, presbyopia). In other words the visual limitations of older peopleís visual functioning may be important determinants of their reading performance and reading preferences. The stronger acuity/legibility relationship for elderly readers (r=0.83) than younger ones (r=0.59) is at least consistent with this suggestion. Surprisingly, among the younger readers more time was spent reading fonts that they judged as being superior. A possible explanation for this is that being students the young participants are assessed by examinations on a regular basis, and may have been motivated to optimize their comprehension ëscoreí on the passages. While the general comprehension questions were designed only to encourage participants to read the passages carefully, they may have been perceived as an examination deserving of careful preparation. That is, the young participants may have been optimizing comprehension by spending more time on fonts that facilitated such an effort (i.e., the "easier" fonts). This is reflected in the drop out rate (only one young but five elderly participants dropped out). There was little evidence that elderly participants were less motivated overall than their younger counterparts. Previous research (e.g., Furchgott & Busemeyer, 1976; Powell, Eisdorfer & Bogdonoff, 1964) indicated that the elderly were just as well, if not more motivated than younger people based on objective physiological measures. Whatever its basis, when the Ranked Reading Time for the two groups were compared across the eight fonts, they were inversely related (r= -0.57). Certainty this effect cannot be related to the content of the materials since it was randomized with font and matched across each age/sex groups.

Differential effects of font on young and elderly readers were also seen in the relationships between the Simplicity/Complexity rating and the other subjective measures. While Simplicity/ Complexity was robustly and positively related to Reading Ease, Reading Discomfort, and Clarity of Type (correlations >.85) for older participants, none of these relationships were significant among the young. Again, their visual limitations may have made these discriminations more salient for older readers.

Study Limitations and Future Research

Although some legibility differences were found between the fonts tested, they may have been more robust had this study used smaller size steps between lines on the legibility charts. The 0.1 minarc step in critical detail between adjacent lines (i.e., 1/5 x-height) and corresponding 0.5 minarc change in x-height, may have been inadequate to reveal small differences in font legibility. Future research dedicated to evaluating font differences in legibility might be well-advised to use smaller size steps between print lines.

Although an evaluation of the effectiveness of small print size was the purpose of this study, the relative absence of age differences may also indicate that the print tested was not small enough to "push" the readers who participated in this study. Mean Reading Ease was low (3.38/7.0) overall as well as for the young (3.45/7.0) and elderly groups (3.32/7.0) separately. It may be useful in future research to set print size for readers based upon an initial assessment of individual legibility threshold rather than at the same level for all participants, as was done in this study.

Although the effect only approached significance, the fact that fonts that were read quicker by young people required longer times for older readers was nonetheless surprising. So too were the inverse relationships between Reading Time and most other reading measures for the two groups. It seems that the best fonts for the young readers are not always the best for older readers in all regards. A structural modeling study that tests a large number of readers of different age over a wide range of reading measures might help to define the degree to which different factors contribute to effective reading in the young and elderly.

Another limit on the generalization of the results of this study is that it examined the effects of fonts only on the reading of prose passages. Many important information, such as instruction on medicine bottles involve short passages that contain numbers (e.g., "Take 2 tablets 3 times daily"). Future research should examine the effects of fonts on short passages that contain numbers as well as text.

Another limitation of the present study was that reading was evaluated under viewing conditions that were near ideal. In the "real world", small print is often read under sub-optimal conditions such as reduced lighting, low luminance or colour contrast, poor print and relative motion between reader and material. The high luminance, and contrast (black print on a white background), in combination with high resolution print (600 dpi) used in this study may have minimized age differences. Future research in this area should consider possible interaction between viewing conditions, and font characteristics. Given the visual losses normal in old age, this work would be particularly informative if it included participants with widely varying visual abilities.

Finally, the high correlations among Reading Ease, Reading Discomfort, and Clarity of Type may have been artificially high due to method variance. These three measures were always assessed in immediate sequence, and participants may not have differentiated sufficiently between them. If these measures are to be used in the future, it may be best to assess them separately, and also to provide participants with "well-anchored" operational definitions/explanations of each rating dimension prior to their use of it.

Conclusions and Recommendations

In the printing industry, fonts are usually selected for specific applications for the "feeling" they portray, and based on the assumption that commonly-used fonts are more or less equally effective. The results of this study question the validity of this assumption. Fonts vary significantly in their legibility and readability and they appear to do so somewhat differently for young and elderly readers, perhaps partly reflective of the different visual capacities of the two groups. It is important that fonts be selected based on empirical data for specific reading task(s), the content of the written material, the visual characteristics of the "target" audience, the criticality of the message, and the viewing conditions in which the print might be read.

Based on the findings of this study, what general recommendations can be offered regarding font selection? First, small print should not use condensed or complex fonts. It should be printed with the highest possible spatial resolution with high letter-to-background contrast (i.e., black on white). The fonts used in this research that appear to best fit these recommendations for small print are Times New Roman, Bell Gothic and Sabon and avoid the use of condensed or complex fonts. Finally, if the information is truly important, it probably should not be conveyed with small print unless there is no alternative.



This page last updated June, 1998 by: Kevin Connolly - <gkconnol@acs.ucalgary.ca>