TABLE OF CONTENTS

Part 1 : Design criteria for the visual presentation of information

Introduction

1 The organization of information content

1.1 Defining the information content

1.2 Highlighting information content

1.3 Number of units of information

1.4 Hierarchies: degrees, locations, size, and tone

1.5 Editorial consistency

2 Typography

2.1 Choosing typographic fonts

2.2 Typographic styles

2.3 Types of text

2.4 Legibility of letters / The resolving power of the eye

2.5 Legibility of letters / comfortable reading

2.6 Upper case versus lower case letters

3 Layout and readability

3.1 Type size, column width, column style and leading

3.2 The use of space

3.3 The spatial arrangement of text

3.4 Setting titles

4 Use of color and tone

5 Evaluation: specificity of each situation of implementation

References

Glossary

Part 2 : Example: Preliminary study for a redesign of the COPD Academic Detailing sheet

Part 3 : Design Guidelines for the visual presentation of information

1 Information content

1.1 Defining the information content

1.2 Highlighting information content

1.3 Number of units of information

1.4 Hierarchies: degrees, positions, size, and tone

1.5 Editorial consistency

2 Typography

2.1 Choosing typographic fonts

2.2 Typographic styles

2.3 Types of text

2.4 Legibility of letters / continuous text / comfortable reading

3 Layout

3.1 Type size, column width, column style and leading

3.2 The use of space

3.3 The spatial arrangement of text

3.4 Setting titles

4 Use of color and tone

4.1 Color contrast

4.2 Color coding

4.3 Background tone

5 Evaluation: specificity of each situation of implementation

6 Summary: steps of the design process /check list

PART 1 : Design criteria for the visual presentation of information

Introduction

Basic assumptions for this report are:

1. that the visual presentation of information affects legibility, comprehension, and memorization of messages;

2. that facilitation of the reading task increases the likelihood that the intended public will read the texts and follow their recommendations;

3. that the design of the visual presentation also affects the sense of trust of the readers on the messages.

The evaluative reactions of the public support some of our assumptions. In a study of type form variables, and of the perceived readability and the perceived hazard of messages, it was found that there is a correlation between the two perceptions. Good readability might help create a better awareness of the hazard described by a message (Silver, Kline, and Braun, 1993, p. 824).

The set of criteria discussed in this report could be applicable to many instances of visual communications, fundamentally printed communications. These criteria are broad, and their implementation is exemplified by our recently completed project on the evaluation and re-design of the COPD sheet (see Part 2 of this report).

This document is aimed at helping Academic Detailing Program leaders direct the production of documents, but this will always require the participation of an intelligent designer. An intelligent designer can be defined as someone who can make good decisions at all levels of the conception and production of a document, and who can articulate verbally the reasons for those decisions, in a clear and convincing way, to non-designers (i.e. the project leaders). Design decisions should be based as much as possible on empirical evidence. The designer is expected to have worked in the field of information design (not only in advertising and consumer products and services’ communications). The term “information design” defines a field within visual communication design. This field is specifically centered on the organization and communication of information, as opposed to advertising design that is oriented to the production of persuasive communications, or educational design that is directed at engaging learners in the development of life skills.

The criteria proposed in this document derive from principles related to four aspects of the visual presentation of information:

1. the organization of informational content;

2. typographic selection;

3. layout and readability; and

4. use of color

1 The organization of information content

1.1 Defining the information content

Content experts will select the information to be included in an Academic Detailing sheet at the optimum level of detail on the basis of assessing the needs of users, and in the context of other available sources of information, such as the AMA Guidelines. Reference to more extended sources will be useful, as indicated in our own research with users early in 2005 (Frascara and Ruecker).

Information designers in dialogue with content experts can determine where and how textual information can be presented diagrammatically, thus affecting the amount of space required for the presentation of the information. This must be done bearing in mind not only the possibility for compacting information but also, and fundamentally, the way in which diagrams can help comprehension and memorization.

Recommendation 1

Content experts should begin to work with information designers once the optimum level of information extension and detail are arrived at. This should be done bearing in mind the size of the final product (in our case, both sides of a letter-size sheet), the operational objectives of the piece, and other sources that complement the information. This will define the amount of information to be included in the academic detailing sheet, and the references to other sources.

1.2 Highlighting information content

There is a personal cost associated with reading a document – a time cost. One has to convince the user that the time necessary to read it is well used. This can be done by highlighting some key words that may show at a glance the relevance of the material to the reader (Desaulniers, 1987, p. 56).

The recommendation to highlight some key words in the text is based on Appleton’s theory of prospect and refuge and on Gibson’s theory of affordances. According to Appleton (1975), the enjoyment of representations of landscape in paintings is connected to landscapes that show wide panoramas. He argues that primitive human groups sought locations to settle where they could see for a distance while having the backs covered. Therefore, up on the hillside and in caves. This experience carries into daily behavior, making people prefer situations where a perception of the surroundings (in terms of space, time, and information) allows one to foresee what is coming. In this sense, it would be appropriate to highlight key words that could immediately show the relevance of the document to the people exposed to a hazard.

At a finer level of analysis, according to Gibson (1979), human groups have wandered through spaces for thousands of years assessing what the environments could offer them. We have in this process developed a built-in attitude to see what is there for us, in any situation. Gibson discusses needs for shelter, terrain, water, fire, objects, tools, and other animals. We can be dealing here with different subjects, but, as much as our ancestors were assessing what the environments could afford them, we constantly assess whether moves and efforts are worth our while or not. Providing a reader with a fast possibility for assessing the personal relevance of a given text and the specific nature of the information faced, is pivotal in connection with the creation of the disposition to read the text and to follow its contents.

Dixon (cited by Wright) has shown that instructions are followed more accurately when they start by giving the reader a high level description of the task. That is, subjects perform better regarding comprehension of instructions when a global description of the task is provided at the beginning rather than at the end of an explanation (Wright, 1994, p. 1.9). Having a general idea of what one will confront, and having it fast, improves our ability to process information. Along the same line, in a more recent paper, Stan Ruecker argues that rich prospect interfaces, that is, interfaces that show a large number of meaningful indicators of a database content, facilitate navigation and use (Ruecker).

Recommendation 2

Since quick understanding of what exactly a message is about fosters reading and compliance, keywords should be emphasized – in both subtitles and within continuous text – to stress the relevance of the document to the users.

1.3 Number of units of information

The totality of the content should be divided into a maximum of five sections. Karl Jung discussed pre-perceptual structures of our mind that filter the continuing flow of information. He suggested that above-below and left-right are non culturally dependent human ways of organizing the visual field, but are generally human; these four locations imply the notion of “center” as a relative location. Lakoff and Johnson have discussed the importance of metaphor for the development of abstract thinking. These two sources complement each other in support of the notion that five is normally the number of things people can remember relatively easily. Other studies indicate that this number generally varies plus two or minus two depending on the individual in question. It is of course easier to remember one thing than two, and in some cases five can be too much to distinguish, recognize and memorize, such as in the case of degrees of severity. The “magic five” should be considered a maximum number, but regarding hierarchies, it is easier to remember three rather than five levels (as it happens in “above, middle and below” or “left, right and center”, given that hierarchies move along one dimension).

George Miller (1956), however, discussed the “magic seven” (plus two or minus two) as the key memory capacity number. He discusses several studies proposing that seven might be the usual limit for our capacity to remember locations, hierarchies or differences in a variety of scales with precision. While he proposes seven, I am more in favor of sticking to five, on the same grounds that I will later use to distinguish threshold reading from comfortable reading. If some individuals are indeed able to remember nine items with ease, many are not, and I prefer to recommend a number that all the users of the Academic Detailing sheets will be able to remember or distinguish with ease.

Several studies indicate that we have a very high ability to estimate numbers of items in groups of five or fewer elements. In a study exposing random groups of dots for less than 1/10 of a second, and asking subjects to estimate their number, it was found that at this speed it is not possible to count. Subjects could only guess. When dots were six or more, the accuracy of the estimations fell dramatically (Oberly, 1924, p. 332).

Recommendation 3

The information should be divided into a maximum of five sections and each section should be divided into a maximum of five parts. Numbering the parts helps memorization of both the number of units and their sequence in either time or importance, particularly when they are three or more.

Author’s note: all examples and their captions in this document are set in blue to distinguish them from the main text.

Example: BC’s newsletter “the review” has 10 subtitles in bold, three underlined, four tables and one summary. It would be useful to study the possibility of dividing the ten subtitles into at least two groups. It would be useful to number the subtitles to help memorization of content and sequence.

1.4 Hierarchies: degrees, locations, size, and tone

Tone, size and indentations serve to denote importance in texts and must be used consistently and sparingly in a document. A reader can easily distinguish two levels of importance. Three is also possible. More have to be carefully implemented.

Recommendation 4

• A maximum of three different levels of importance should be used (Extendable to five if indispensable).
• Levels of hierarchy in tables should be consistently located: unless otherwise established by specific standards, if worst category is placed above and best below for one table, they should be in the same position for all tables.
• Size and tone should clearly indicate importance: working on white ground the most important item should be larger, darker, or larger and darker than the following one. Avoid contradicting indicators such as a title in grey larger but lighter than a subtitle in black.

Example taken from “Management of Dyslipidemia and the Prevention of Cardiovascular Disease/Monitoring supplement” showing a lack of clarity of levels of importance, since the larger subtitle is lighter in shade than the smaller sub-subtitle:

Other examples: The title of “the review” (BC’s newsletter), is too large in comparison to the information “Statin update” and “March 2005.” The name of the newsletter is quite predictable, and therefore carries little information. It is used only to file the publication in the right folder. When searching in the collection of the newsletters for the most recent issue on a particular topic, the date and the topic are quite important and should be emphasized. This emphasis requires a larger size, and also better spacing from the main title and the text. In the same publication there are subtitles that are bold and others that are underlined: it is not clear which is more important.

Example from: “Model for Estimating the 10-Year Risk of Coronary Artery Disease

(death or non-fatal MI)”

1. Original version. The use of all capitals reduces legibility:

Risk calculator should NOT be used for patients with DIABETES MELLITUS, ANY ATHEROSCLEROTIC DISEASE, PATIENTS WITH CHRONIC KIDNEY DISEASE OR UNDERGOING LONG TERM DIALYSIS. These patients are automatically in the HIGH RISK category.

2. Modifications to improve legibility:

2.1 Use of bold style instead of all caps increases legibility and conspicuity. When more emphasis is required for some words, the text can be set in all caps, 1 or 2 points smaller, or in small caps (if they exist in the font used):

Risk calculator should NOT be used for patients with Diabetes Mellitus, any atherosclerotic disease, patients with chronic kidney disease or undergoing long term dialysis. These patients are automatically in the HIGH RISK category.

2.2 Use of bold style, 2 points extra leading, and list formatting instead of prose improves performance:

Risk calculator should NOT be used for patients with:

• Diabetes Mellitus,

• any atherosclerotic disease,

• chronic kidney disease, or

• patients undergoing long term dialysis.

These patients are automatically in the HIGH RISK category.

1.5 Editorial consistency

Readers of documents get tuned in with the way information is presented as soon as they see the beginning of a document (Frascara, 2006, in press). It is important that all sections of a document be organized in a similar way, i.e. from the general to the particular, avoiding alternating different approaches to the organization of the content’s sequence.

Recommendation 5

To facilitate comprehension, wherever possible, the organization of the information should be consistent across the sections of a document and across documents in a set.

Example: In “Managing of Dyslipidemia and the prevention of Cardiovascular Disease/Monitoring supplement,” the layout of the titles for the tables is different in Table 1 from tables 2 and 3.

Table 1

Major Causes of Secondary Dyslipidemia include:

 

Table 2: Predisposing Factors for Myopathy/Rhabdomyolysis

2 Typography

Author’s note: Since readers of this report may come from fields other than the practice of

typographic design, a glossary is included at the end of Part 1 of this report. Whenever current research has not contradicted the results of original studies, original studies are cited for reference. While we have refined our ability to perceive factors that were overlooked in the past, and are better at contextualizing research findings, much of the research done on legibility since 1925 still holds.

2.1 Choosing typographic fonts

Font selection involves choices of typeface, size, and style. From a human factors perspective there are two main areas to consider: legibility, the facilitation of letter and word recognition studied as a perception problem; and readability, the facilitation of the comprehension of texts, studied as a cognition problem. This section and makes general and specific recommendations for typographic choices.

The alphabet is a system of combinable components. These are basically 26 capitals, 26 lower cases, 10 numerals and some additional 20 signs. These 82 components should be compatible in any combination possible, so that reading does not become hampered by uneven weights, random heights, and variable tonal densities, inclinations, spaces, or shapes. The basic 52 letters imply 2704 possible cases of sequences of two letters. They should be designed to relate well to each other in any sequence. Even though the shapes of the roman alphabet letters are well established, variations can be dramatic and the choice of a typeface deserves scrutiny of its quality. (Figure 1)

Figure 1: Examples from Bodoni Book, Courier and Chicago typefaces, all 24 pt.

It is clear that it is not possible to design a typeface by trial and error, and it is also easy to believe that many poor designs of typefaces crowd the market. Choosing a typeface requires attention to the system’s ability to perform well for the kind of reading required. Attention grabbing and continuous reading are quite different functions. Ease in continuous reading is based on a good balance between consistency of style and tonal weight across an alphabet and, at the same time, sufficient distinction between 8 letters. Height, tonal weight, and style should maintain coherence in a given font. The tonal weight is the ratio of black to white in the area of a letter, as created by the strokes and the spaces between them.

One of the basic elements for ease of reading is word recognition. For this the fundamental cues are the initial letter, the length of the word, and its profile – mainly ascenders and descenders and the internal tonal structure of the word (Schiepers, 1976, p. 51). The fixation points separated by the saccadic movements of the eye are guided by the blank space between important words (not articles or prepositions). One reads words by fixating foveal vision around the second letter of a word, letting peripheral vision make up whatever is to the right. According to Spencer, only four letters of normal size fall within foveal vision at a usual reading distance. The other 12 to 15 letters that we normally recognize with accuracy fall away from foveal vision, and therefore away

from our maximum ability to resolve their details (Spencer, 1968, p. 19). Reading speed relates to using peripheral vision efficiently, and taking a number of guesses. In order for this to happen, words must maintain tonal homogeneity. The tonal density of each letter, despite variations of shape, should be similar to every other one, so that the ability of the visual system to spot its next useful fixation point could be enhanced. In support of this, Spencer cites several researchers, including Pillsbury, who showed that a typing error in the first part of a word is more easily recognizable than in the second part. This happens because the second part of a word is normally guessed, and only seen in peripheral vision (Spencer, 1968, pp. 17-18).

Simple and geometric letterforms, such as Paul Renner’s Futura (1920s) (Figure 2) or Herb Lubalin’s Avant Garde (1960s), are not good for running prose. They ignore the need for a good balance between similarity of style and distinctiveness of each letter. Variations of tonal density between letters are extreme in these typefaces, creating random points of visual attraction in words, and negatively affecting legibility. These variations fool the eye in its task of searching for the most efficient places to focus on during its brief fixations.

Figure 2: Paul Renner: Futura typeface, 1920s, 16 pt.

Figure 3: Adrian Frutiger: Univers typeface, 1950s, 16 pt.

Figure 4: Max Miedinger: Helvetica typeface, 1950s, 16 pt.

The sans serif alphabets created in the 1950s, such as Univers (Adrian Frutiger) (Figure 3) and Helvetica (Max Miedinger) (Figure 4) recovered the traditional interest in optical corrections. Their legibility, however, has been surpassed by later alphabets whose letters have wider apertures and are more distinct from one another, such as Frutiger, designed in the 1970s and others designed later (Figures 5 to 8). This improvement in legibility is based on increasing the opening left by inward turning strokes – as in the a or e – and is supported by research related to the Landholt Ring (Sanders and McCormick, 1993, p. 94). The Landholt Ring measures visual resolution by allowing a small opening in the ring, one that at certain distances becomes invisible. Increasing the gap will increase its detectability. This same phenomenon applies to the resolution – and therefore the distinction – of several small letters such as a, e, c, and o. Serif faces have the advantage that they are massively used for continuous text, and therefore facilitate legibility (habit is a factor for legibility). Sans serif faces are useful when clear distinctions of weight or small sizes are required.

Figure 5: Adrian Frutiger: Frutiger typeface, 1970s, 16 pt.

Figure 6: Studio Meta: Offi cina Sans typeface, 1990s 16 pt.

Figure 7: Lukas de Groot: The Sans, 1990s, 16 pt.

Figure 8: Details of letter design endings showing their different resolution: Helvetica and The Sans, 28 pt.

An alphabet is a system of elements that should show a good balance between similarities and differences. Similarities, so that a reader can subordinate letters to words and words to sentences. Differences, because the reader must be able to distinguish letters from one another in order to easily distinguish words from one another. All letters in a given weight of a typeface should have a similar tonal density. Apertures of letters should be generous and visible.

Recommendation 6

When selecting a typeface for running prose, recently designed or re-designed faces such as Times New Roman, Officina (Sans or Serif), or The Sans, are recommendable. They provide homogeneous tonal density in the composition of words, they facilitate letter identification, they relate to the most usual fonts, and have sufficient distinction for ascenders and descenders, thus facilitating the detection of word profile.

Type examples

(Note: To assess the features of these fonts one should print the examples. Screen rendering is not always faithful to the styles and fonts chosen)

Example: “the review” (BC’s newsletter) uses a typeface similar to Helvetica that has closing endings for some letters, i.e.: a, e, s. This creates difficulties for the legibility of texts in small sizes, such as those that appear in some of the tables.

2.2 Typographic styles

The intention to improve legibility in small sizes of type led designers of the 1950s to abandon the 5 to 8 proportion between lower case and upper case, and change it to a 6 to 8 proportion (Figure 9). This, however, generated the need to increase line separations, to avoid the visual jamming of lines. This is often ignored in the design of scientific or technical texts where space is at a premium, in an attempt to comply with requirements related to information content. It could be argued that the height of the characters of these texts falls easily within the range of legible sizes.

They become, however, discouraging, because setting a text in all caps, or setting without leading a contemporary typeface that maximizes the x-height of the lower cases, will result in an overwhelming appearance. There is always a need to add about 20% leading (line separation) up to 12-point type, and at least 10% for 14-point and up.

Figure 9: Relation 5:8 and 6:8 between lowercase and caps in different typefaces. Bembo and Helvetica, 36 pt.

Poulton (1972, p. 161) found that to attain a similar x-height of about 1.2 mm, Times Roman and Univers required 6.5 pt. type, while old style Perpetua required 8.5 pt. type. The legibility of text is supported by the size of the face of the letters, and not by their body size. Tinker has been criticized for reporting his studies referring to point size and not to face size. When he began his studies all alphabets were designed in a 5 to 8 proportion. The 6 to 8 relation of lower to upper cases, has been compounded by reducing the length of the descenders (so as to harmonize with shorter ascenders), resulting in a 20% increase of the height of a lowercase letter, plus about 5% more due to the increased size of the capitals. Considering a 127-point size font (44.6 mm), the

X-height of a Helvetica uppercase is 28 mm, while that of an old style Times is 25.5. The descender of a lowercase q in Helvetica will be 8 mm for that point size, and 9 for Times. Measuring from the bottom edge of the descender to the top of the uppercase X, the Helvetica will measure 36 mm and the Times 34.5. The height of their lowercase x shows the most signifi cant difference: 21 mm for Helvetica and 15.9 mm for Times. The basic dimensions to control in order to assess the quality of a typeface are height, width, and weight.

Height: There should be an apparent constant x-height of all lower-case letters; an apparent constant height of capitals; and an apparent constant height of all ascenders and descenders in lower case letters. Movements away from the norm are possible, but ascenders higher than capitals are not practical because they might invade the space of text lines above. However, fonts with the letter t shorter than the letter l are no exceptions, since the t has more distinctive features and therefore it does not need a long ascender above its horizontal stroke to be recognizable. The use of alphabets with short ascenders for d or b is not advised, since these extensions are essential to distinguish some letters from others.

Width: This will vary, according to the complexity of the letterform. The guiding criteria to ensure that no letter stands out in an alphabet is to maintain tonal density constant, that is, to maintain a similar ratio of black to white in all letters. To achieve this, the letters a-b-c-d-e-g-o-p-q may have equal width. Other groups with similar width are f-r-t; h-n-s-u; i-j-l; k-v-x-y-z; and m-w.

Weight (tonal weight): Consistency in weight means consistency in tonal weight. It is necessary to provide an even weight throughout a word. This is necessary because reading is led by tonal contrast. Strokes should have an apparently consistent thickness across letters, and from horizontals to verticals to diagonals and curves. Changes in their thickness, from thick to thin, should be based on the same rationale for every letter, and this is based on the influence of the calligraphic tradition on the design of the roman alphabet. Ascending diagonal strokes (from left to right) are normally thinner than descending diagonal strokes; vertical strokes are thicker than horizontal strokes (this possibly derives from architecture, where the column that supports should be thicker than the supported roof); curved strokes change smoothly their thickness according to

the evolution of their direction, following the logic of a calligraphic pen.

Type style: While the use of italics can be accepted as a way of calling attention in a subtle way to a few words within continuous prose, long texts in italics significantly reduce reading speed. Tinker found the slow down to be about 5%. He also found that when texts in italics combine with other negative conditions, such as low light level or small type size, the effect is markedly strong (Tinker, 1963, p. 55). In some conditions, Tinker found that texts in bold face type have proven to be as readable as texts in a regular weight (Tinker, 1963, p. 62). In general, however, people tend to prefer texts in a regular weight. Semantic or visibility reasons may combine sometimes in favor of boldface texts, but some bold faces reduce the legibility of counters, and in some cases this can result in a loss of legibility for a few characters. A text might need emphasis in a word or two, to attract the reader to the main points of a paragraph. Bold text is a preferred option for this purpose, but the difference should be easily visible, particularly when bold text is used to facilitate location of specific items in a long text. Recommendation 7 Text should be always set in medium weight roman style (also called regular, plain or normal).

Recommendation 8

Italics should be used within text where distinction is necessary.

Recommendation 9

Bold type can be used within the text where there is a need for emphasis, or for titles that require distinction of importance.

2.3 Types of text

Different types of text require different typefaces. Preferences will depend on length of text, use of text, and purpose of the message.

Length of text: a) A few words (as in titles); b) A few sentences (as in summaries); c) Long text (as in continuous prose).

Use of text: a) Continuous reading; or b) Information scanning/searching.

Purpose of message: a) To inform; b) To direct, persuade, or warn.

Recommendation 10

Typographic detection should be guided by the length, use, and purpose of the text to be set. When reading a few words, skillful readers will do well across a wide variety of choices. Differences in performance will become more critical in long texts, where all the recommendations of this report should be considered. Information scanning will require a typeface that allows a wide range of weights for emphasis.

Recommendation 11

No more than three levels of importance through differentiation of style should be attempted for information scanning within texts.

Recommendation 12

The typographic selection for directing, persuading or warning may use more texts in capitals, bold style, and sans serif faces.

2.4 Legibility of letters / The resolving power of the eye

One way of establishing reliable criteria for the size of type to ensure legibility, is to resort to optometry. Visual acuity can be measured as the visual angle subtended by the eye when perceiving a shape, measured in minutes of arc. The viewing angle is formed by the relation between stimulus size and viewing distance. In the case of what is considered normal vision, the height of the letter E should be 5 minutes of arc, or .0015 radians – the required ratio of letter height to viewing distance (Smith, 1984, p. 176). This means that at a one meter viewing distance the letter should measure 1.5 millimeters (Grether and Baker, 1972, p. 58). This is the standard used today for the definition of normal vision.

Given that the letter E is formed by five horizontal elements (three strokes and two spaces), someone with normal vision should be able to resolve details that measure one third of a millimeter per meter of viewing distance. A person is easily able to detect the presence of such element in the visual field, and something even thinner could be spotted. This is true, but regarding legibility, we must address the capacity of the eye to resolve details, and not merely to detect their presence. If we look at several parallel lines, we will be able to count them if we have normal vision, and as long as their thickness and the separations between them are at least one third of a millimeter per meter of viewing distance.

The Snellen chart prepared for semi-literate people works with square forms similar to the upper case E in different orientations. Subjects are supposed to clearly indicate the direction of the elements. In this system, the smallest readable E, that establishes whether one does or does not have normal vision, is fi ve times the size of the resolution threshold (three horizontal strokes and two spaces). Sanders and McCormick argue that, when the overall size of the letter is constant, people can equally distinguish the letter E when the strokes are thinner. They recommend a stroke-width to letter-height ratio of 1:6 to 1:8 for black type on white ground, and 1:8 to 1:10 for the opposite. They go on to recommend that when illumination is reduced, thick letters become more readable, and a ratio of 1:5 is more efficient (Sanders and McCormick, 1993, p. 103).

Recommendation 13

The ratio of stroke thickness to height for letters should be between 1:5 to 1:8

Recommendation 14

The absolute minimum size for letters to be read at close range (35 cm distance) should be 0.5 millimeters (see below for comfortable reading)

2.5 Legibility of letters / comfortable reading

It is different to define a “threshold value” and a “comfortable reading” value. If one observes a person approaching a written notice, this person will get much closer to the text than what the threshold value requires. This will normally result in a comfortable reading value of two to three times the threshold value, increasing the size of the letters to 3 or even 4.5 millimeters per meter of viewing distance.

Duncan and Konz report that when subjects were asked to read numerals they approached them until subtending and angle of 23 minutes of arc, instead of the 5 required for error-free reading. Numerals require larger visual angles because they do not form units such as words that can be recognized in a holistic manner. The reading of letters involves guessing and probability of occurrence. In numbers, instead, this is not the case, and therefore a reader will require a greater size for reading strings of numbers. Among the military, the recommendation for the size of critical data is three to five times the threshold value (Smith, p. 173). This can be considered in the context of warnings and main section headings for Academic Detailing sheets. Smith notes that the subtended angle range recommended in a Military Standard (U.S. Military Standard 1472B-1974) varies from 6 to 37 minutes, that is, from barely above the normal acuity threshold of 5 minutes to a generous 7.4 times higher. (Figure 10)

Figure 10: Comparison of recommendations for letter sizes (Smith, 1984, 176)

In the history of requirements for minimum sizes in all kinds of warnings or mandatory notices, it has been common among manufacturers to address minimum height requirements by producing ultra-narrow letters with hairline strokes. This is not a good interpretation of the intentions of such regulations, which are meant to help users and consumers read sizes, contents, warnings, expiry dates, and the like. As discussed above, for a letter to maintain a good level of legibility, the thickness of its elements must be between one fifth to one eighth of the height of the character, and the character’s width should be no less than three fifths of its height (Grether and Baker, p. 107). This derives from the fact that the letters E, B and S, for instance, have three horizontal strokes and two spaces separating them, that is, five elements to be discriminated. They have three vertical elements. The same maximum number of elements applies to almost all lowercase letters, with the exception of m and w, which should therefore be wider than the rest. In old typewriters, given their fixed space allowance for all letters, it was quite possible to spot these letters in a typescript – their strokes jammed together and their tone darker than the rest. No letter has more than five horizontal elements and normally no more that three vertical ones.

Extensive research on the relative legibility of typefaces with different widths has been reported by Young, Laughery and Bell at the 1992 Human Factors Society meeting in Santa Monica (Cited by Wogalter and Vigilante, 2003, p. 328). They found that any reduction of the width of a given font affected its legibility in a significant way. According to Tinker (1963, pp. 70, 71, 95), the ideal typesize range for continuous reading at close range – such as in the case of a book or a pamphlet – seems to be between 9 and 11 points, set solid in a 19 pica column. When compared with 10-point type, 8-point type appeared to slow down reading speed by 3.8%. Interestingly, 11-point type showed an increase of reading speed of about 5% when compared to 10, but 12-point type took 2.1% longer time to read, according to some studies reported. As already mentioned above, Tinker used old style typefaces. The face size of the 10-point type he was measuring would be equivalent to 8-point type today. The size of the face of a Univers lower case letter in a given point size is at least 20% larger, and sometimes more, than a Times Roman of the 1940s.

The size of type is one of the most important determinants of readability. Studying the readability of different sizes of print and its infl uence on quick perception of words, Tinker reports that words set in 12-point type were perceived about 47% better than words set in 6-point type (1963, p.223).

Recommendation 15

The ratio width to height of a letter is critical for legibility, and condensed letters should not be used in search for increasing the amount of text in a given space. The width-to-height proportion of a letter should be around 3 to 5.

Recommendation 16

Since a comfortable reading size for the x-height of letters is 4.5 mm per meter of viewing distance, a type size of about 1.5 mm is recommended for close range reading, as in the case of the Academic Detailing sheets. This results in 8 to 12 pt size text, depending on the typographic font. The required stroke width to letter height ratio should be 1:5 to 1:8.

2.6 Upper case versus lower case letters

Much has been discussed about the relative legibility of words in upper case as opposed to lower case. In signage or titles, the usual reading of one to three words does not show significant differences. For long texts the situation is different, because texts in all caps remove differences in word profile – one of the strategies of the visual system to recognize words. In an experiment comparing the legibility performance of texts in Univers all caps with other texts in lowercase letters, working in small sizes, Poulton found that the caps had to be 1.44 mm high to be as legible as the lowercase Univers with an x-height of 1.19 mm (Poulton, 1972, p. 160). Poulton refers readers to Tinker and Paterson (1928), when he states that in 9-pt and 10-pt body sizes, capitals are not read as quickly as mixed capitals and lowercase letters (Poulton, 1972, p. 158). Tinker, in an experiment reported in 1940, indicated that according to reading speed tests (and readers opinions) lowercase type is much more legible than texts set all in capitals (Tinker, 1963, pp. 57, 60). The same is supported by Hartley (2004, p. 921).

Confusions concerning the superiority of either capitals or lowercase letters probably originated with early legibility studies that compared the legibility of individual letters, upper and lowercase.

These studies were often flawed on two counts: 1) they measured exclusively the legibility of individual letters, and 2) they often worked within a given point size, disregarding the different actual size of capitals and lowercase letters.

Habit has much to do with readability. We are used to reading texts in upper and lower case. To us, initial letters in upper case indicate proper names or beginning of sentences. By the same token, it could be that habit supports serif against sans serif faces when reading long texts, such as we are used to seeing in books, newspapers, and magazines.

In his famous article on the legibility of small typefaces, Poulton found that Times Roman and Perpetua were more legible than Univers; even when the Univers texts were enlarged to a 1.48 mm x-height, well above the 1.24 and 1.28 mm used for Times Roman and Perpetua respectively (Poulton, 1972, p. 161).

Recommendation 17

Since word recognition is based on the initial letter, the length of the word, and its profile (ascenders and descenders and the word as a pattern), a mix of upper and lowercase letters is more recommendable than uppercase letters alone for setting continuous texts.

Recommendation 18

Uppercase letters are more legible when used individually. They could be used when identifying sections of a text or other units by one letter. They can also be used for short strings of words, such as in signs, warnings, or titles.

3 Layout and readability

Different from legibility that concerns itself with the recognition of letters, readability concerns itself with the comprehension of texts. The concept was coined in the 1940s, and was later disseminated in the typographic design community by Herbert Spencer, a significant contributor to studies in the field (founder of the Legibility Research Unit at the Royal College of Art in London). He was possibly the first designer to develop systematic research on legibility, a field investigated initially by psychologists like Miles Tinker since the 1920s. Incidentally, Tinker was against the use of the term “Readability” arguing that it was confusing due to the different uses given to it (Tinker, 1963, p. 4). Concurring with George Klare on the usefulness of the concept and its relation with comprehension (Klare, 1984, p. 479), for the purpose of this report, however, I prefer to distinguish between legibility and readability. James Felici, in his thorough and recent book on typography (Felici, 2003, p. 103) uses this distinction.

One way of studying legibility is through research on the relative legibility of individual letters, that is, on the different recognition performance they show due to distance or other conditions. It is different to study the ease and speed with which a text is read. The first is relevant to type designers and to those choosing typefaces for a given project (Tinker, 1963, p. 33). If one designs alphabets, it helps to know which individual letters require modifications to improve their legibility. For the layout of text, however, this is largely irrelevant. In this report, both the legibility of alphabets – that is based on the design of its individual components –, and the readability of texts – that is based on the manipulation of layout – are of central importance.

Readability, more than being associated with perception (as is legibility) is associated with cognition. Variations of reading performance between different sectors of the public are quite dramatic – depending on educational level and lifestyle –, but one can expect that users of Academic Detailing sheets will all possess high reading skills.

One should also recognize that the comprehension of texts (a cognitive act) is tinted by emotional tones (Frascara, 1999; Sinclair, 2002). Emotional tones are influenced by aesthetic judgements. Interestingly enough, Tinker found that subjective judgements about the legibility of texts kept a high correlation with opinions about their aesthetic quality (Tinker, 1963, p. 50). He also found that, in many cases, “judged legibility agrees closely with speed-of-reading scores” (Tinker, 1963, p. 51).

3.1 Type size, column width and leading

Readability of continuous text – the most common fi nal aim of text composition – is affected by a large number of factors. Three of them are very closely related: typesize, column width and leading (Felici, 2003, p. 119). Type size has already been discussed in this report. It should be added, however, that type size not only contributes to ease of reading, but also to memorization of texts (Shaver and Wogalter, 2003, p. 830).

In regards to the size relation between signal word and text in warning signs, ANSI recommends that the signal word should be “at least 50% greater than the height of a capital H in the majority of the message panel wording” (ANSI, p. 7). However, Silver and Brown reported that people were better disposed to read a sign where the signal word was 2 points larger than the text than where it was 4 points larger. This was surprising for the researchers, but it was thought that possibly the similarity of the sizes provides more importance to the text (1993, p. 623). The same was found in another study of perceived readability: subjects judged that when there was no significant difference in size between the signal word and the text, readability was enhanced (Silver, Kline and Braun, 1994, p. 824).

Columns of text must be wide enough to avoid the fatigue created by excessive fracture of the text fl ow; but they should be narrow enough to permit an easy return to the next line. Leading helps balance things out: excessive column width may be compensated by generous leading. Columns that are too narrow, however, offer no chance for compensating the reading jerkiness. These columns will most likely create a very dramatic rag if the text is set flush left, or will show intense variations in tonal density and frequent hyphenations when columns are set justified. In the practice of design, the rule of thumb for appropriate column width is 60 to 70 characters per line (Spencer, 1968, p. 35). In Tinker’s field research it was found that 10- point type with 2 points leading was optimally readable between 14 and 30 picas. The original text, however, is not clear in the specification of the upper end, since it shows 31 picas as slowing down reading by 2.4% (Tinker, 1963, p. 79).

Tinker found that columns that were 9 picas and 43 picas wide reduced the reading speed with 10-point typesize. When set solid (no extra line space), good readability dropped around the 19 picas measure. For 12-point type with 2 points leading, he found that the ideal column measures were between 17 and 33 picas. For 8-point type set solid the recommended column widths ranging between 13 and 25 picas. When setting the 8-point type with 2 points leading, the upper end of the range was extended to 36 picas (Tinker, 1963, p. 81). For 6-point type the recommendable column widths fall between 9 and 25 picas. When 2 points leading is applied to 6-point type, the useful column size ranges from 14 to 28 picas.

The increase of reading time for texts set in columns that are too wide, has appeared to be due to a high number of regressions in the eye movements, possibly because of difficulties in locating the beginning of the next line in a text. This is particularly noticeable when a text is longer than five lines. It is easy to find the second line in a text two lines long. The same is true for three and four lines. But when a text extends to seven or more lines it becomes less easy for the visual system to immediately locate the next fixation, forcing the eyes of the reader to move back and forth several times.

In a study crossing the variables of type size, leading, and column width, Tinker found that for 10-point type the best situation was to use a column 19 picas wide and 2 points leading. The worst situation measured was type set solid in a 9 picas column, and type set solid in a 43 picas column. Two points leading in a 14 picas column was only marginally inferior to the best situation, and so was the one point leading sample in a 14 picas column. Surprisingly, the results were not smooth. In solid set and one point leading, reading performance was better at 31 picas than at both 19 and 43. This lack of smooth worsening of the reading performance when departing from the best situation suggests the presence of processes in the reading task that we do not quite understand (Tinker, 1963, p. 95).

Tinker developed a guideline comparing six type sizes, identifying their best column width and leading for continuous reading. He first determined the best line length and leading for each size, and then compared them to one another in relation to reading speed. All the test material was set in Granjon with 2 point leading. He identified the best situations as follows: 6-point/14 picas, 8-point/16 picas, 9-point/18 picas, 10-point/ 20 picas, 11-point/22 picas and 12-point/24 picas. In these, the best performing one was 12-point, and the worst, 6-point, showing a 5% slower reading time (Tinker, 1963, p. 103). In the practice of visual communication design, and particularly in the design of short texts, this would not appear as a major difference, but, when designing long texts, a fatigue element might be associated to the reading task.

In an approach that omits minor differences in performance, Tinker proposes what he calls “safety zones” for commonly used typesizes for continuous text, ranging from 6 to 12-point type. Omitting small details, his recommendations are as follows:

• For 6-point type, line length ranges from 14 to 28 pica lines, with 2 to 4 points leading.
• For 8-point type, the measure can extend to 36 picas.
• For 9-point type, the leading may extend from 1 to 4 points, and the line length from 14 to 30 picas.
• For 10-point type, the line length should be 14 to 18 picas with 1 to 4 points leading; and 31 picas with 2 points leading.
• For 11-point type, the range is defined as 16 to 34 picas line with 1 to 2 points leading.
• For 12-point type, 17 to 33 pica line with 1 to 4 points leading (Tinker, 1963, pp. 106-107).

The size of letters is fundamental for readability, but this can be seriously compromised by departures from conventional arrangements, such as setting a word over a vertical line, one letter above the other.

Recommendation 19

Leading improves readability for long texts. Fonts designed on a 6-to-8 ratio of lowercase to upper case should be set with at least 2 points of extra leading.

Recommendation 20

Typesize, column width, and leading are important factors for readability. The recommended type sizes and column widths for optimum readability – using 2 point leading – are: 6-point: 14 picas / 8-point: 16 picas / 9-point: 18 picas / 10-point: 20 picas / 11-point: 22 picas / 12-point: 24 picas.

3.2 The use of space

Letter size, letterform and stroke thickness alone do not guarantee the best possible performance in legibility. Letter separation, word separation, line separation, and margins also affect legibility.

Letterspacing: The letterspacing should be designed so that letters forming a word can be seen as a unit. The spacing program should however maintain easy identification of each letter. A typeface like Souvenir, for instance, has too tight a spacing program for the size of its counters; this renders the counters more visible than the spaces between the letters, hampering reading efficiency in continuous texts. Tight spacing in an expanded letter will frequently shift tonal cues from helping guide the fixations between saccadic movements efficiently, to attracting the eyes to meaningless configurations created by the meeting of two contiguous letters or the blank areas inside one. The word spacing program of a type font should be in harmony with the letterspacing program. In addition to the relative differences between letters, type width as a variable relates to

the possibility of condensing the form of type. A usual trend in informational sheets is to cram as much information as possible, resulting in very discouraging texts. Type density can be very negative in relation not only to the ability of people to read the texts, but also to their willingness to read at all (Young, Laughery and Bell, 1992, p. 504).

Leading by itself seems to have little bearing on legibility. Tinker has reported that with 12-point type very similar performance was found in a 25 picas column from solid setting to 8 points leading. In the case of 8-point type, in a 17 picas column, a similar phenomenon was found, although in this case the solid setting showed poorer results (Tinker, 1963, pp. 92-93).

Anderton and Cole (1982) reported findings about how letter separation affects legibility. As common knowledge in typography indicates, when characters are set too tight the legibility performance of any font deteriorates.

Recommendation 21

Texts should be properly spaced between letters, between words, and between lines. Texts set in all caps might require additional letter separation.

Example taken from “Management of Dyslipidemia and the Prevention of Cardiovascular Disease/Monitoring supplement”

Setting the column style as flush left improves legibility:

It is better to use a column width where each line fits about seven to twelve words or

45 characters (23 picas for 12-point type):

For 10-point type, use 19 picas:

Example taken from Dyslipidemia Newsletter, March 2005. (Helvetica regular 10pt set solid, justified)

The same text set with 3 point extra leading and flush left. The 3 extra points between lines facilitate reading, as does the column style, that avoids erratic word spacing.

The same text set with 3 pt extra leading and flush left; set in The Sans, using bold instead of underlining to avoid invading the type area.

3.3 The spatial arrangement of text

The spatial arrangement of type contributes to the readability of texts. The ratio between printed area and total surface area is an important factor, given that it determines the size of margins, which act as boundaries defining the inside and the outside of a document.

Tinker reports that readers appreciate the presence of generous margins and believe they contribute to the beauty and the legibility of texts (Tinker, 1963, p. 126). Within the continuous text of Academic Detailing sheets, the design of warnings should have a good margin surrounding the message, so as to ensure its separation from the prose, and its conspicuousness.

Paragraph style, punctuation, emphases, line length, leading, column format, indents, information chunking, numbered or bulleted entries, and line breaks, all contribute to facilitating comprehension, memorization, and compliance.

In the terrain of written instructions and warnings, it is widely agreed that a text set as continuous prose is less effective in terms of comprehension than a text separated in chunks that make sense. It has also been found that there is a correlation between ease of comprehension and compliance. This has been supported by recommendations appearing in the 1999 Federal register of the Federal Drug Administration of the USA. Before this, Wogalter and Post (1989), cited by Wogalter and Vigilante (2003, p. 329), coincided with Morrow, Leirer and Sheikh (1988), in finding that instructions set in a list format instead of prose format generated better performance in test subjects.

Other researchers recommend bulleted lists as superior to prose (Hartley, 2004, p. 926; Wogalter, Shaver, and Chan). The ANSI concurs in recommending “outline format” and “outline with bullet format” as superior to “continuous format” to enhance readability (ANSI, p. 17). Desaulniers brings out an interesting observation: he suggests that while subjects in an experiment spent similar amount of time reading information presented either in list or paragraph form, there might be a higher tendency to skim the text in a paragraph format (Desaulniers, 1987, p. 58). In another study it was found that the presentation of information as a flow chart led to easier learning and more compliance in the explanation of procedures (Desaulniers, Gillan and Rudisill, 1988, p. 291).

In a study discussing legal documents, it is recommended that a less official-looking document will improve the willingness of people to read it (Wogalter, Howe, Sifuentes and Luginbuhl, 1999, pp. 593, 609). This can also have a bearing on the layout for the design of Academic Detailing sheets. Formatting, in general – that is, moving away from continuous prose and engaging in giving the text a form that responds to its content – seems to improve text comprehension performance (Shaver and Wogalter, 2003, p. 829), and it also cuts down reading time.

The form of the column does not seem to affect readability in a significant manner. Justified texts and flush left texts seem to perform equally well, as reported by Spencer on the basis of tests performed by Zachrisson at the Graphic Institute in Stockholm in 1965 (Spencer, 1968, p. 37). This is also supported by Hartley, who, however, argues that unjustified text might be more helpful for less able readers (Hartley, 2004, p. 923). While some difference was found among low reading skill people, favoring unjustified texts, subjects tended to respond equally well to both formats. It is a common practice amongst communication designers to avoid justified texts when columns are narrow.

It is believed that word spacing should be even throughout a text. It is visually disruptive to have a noticeable tonal difference from line to line when words are forced to fill a given column width, particularly if this column is narrow. Hyphenation is an option in unjustified texts, in order to avoid dramatically different line lengths (Crosby, Fletcher and Forbes, 1970, p. 55). It is also recommended that hyphenation should be used sparingly, and that no more than two consecutive lines should be hyphenated.

Recommendation 22

Chunks of texts are more readable than continuous prose.

Recommendation 23

For the purpose of summaries, lists are more effective than continuous of prose.

Recommendation 24

While empirical evidence does not establish significant differences in readability performance, unjustified texts (flush left) are to be favored, given that they provide even tone in the lines of text and that the line length variation helps locate the next line when reading.

Recommendation 25

Narrow columns should be avoided, and, if used, however sparingly, hyphenation must be used.

Example: In “the review” (BC’s newsletter), the titles of tables are too close to the tables, to the point that descenders contact the frames. Even though reading is possible, more space would facilitate it.

In “Model for Estimating the 10-Year Risk of Coronary Artery Disease (death or nonfatal MI)” the table has excessive space between factors to be compared. The structure of the table should facilitate comparison.

3.4 Setting titles

When setting a title in capitals, it is useful to have access to a letterspacing option. To facilitate legibility, titles in all caps should normally use extra letterspace and word space. This is necessary because capitals are not designed to compose full words, but to serve as initials. This extra space is measured in tenths of a point, and it will take about 0.7 of a point for a 20 point title to be properly spaced if set in all caps. If some letter pairs are too tight – like MI – or too open – like VA –, they can be treated individually.

This is a difficult task that requires some printing tests, because screen rendering of letterspacing is not very good, particularly in non-professional typesetting programs such as Microsoft Word.

Recommendation 26

Titles set in all caps should be specially spaced, at about 0.7 pt of extra letter space for 20 pt type. If some letter pairs are too tight – like MI – or too open – like VA – they can be treated individually. For titles set in all caps 20 points type, 4 points extra leading are required.

Recommendation 27

Long titles are better if set in upper and lower case: this uses less space and increases legibility. For upper and lowercase titles 2 extra points of leading is sufficient up to 20 points.

Recommendation 28

When setting titles all in caps, it is better to set them using a font that has small caps, that are designed to combine with themselves instead of with lower case letters.

Example taken from “Management of Dyslipidemia and the Prevention of Cardiovascular Disease/Monitoring supplement”

The title as it appears in the document (Times New Roman 16 pt. All caps set solid):

1. Line space can help legibility (4 extra points):

2. When setting a title in capitals, it is useful to have access to a letterspacing program. To facilitate legibility, texts in all caps should normally use extra letterspace and word space, in this case 7/10 of a point (If some letter pairs are too tight – like MI – or too open – like VA –, they can be treated individually):

3 Long titles are better if set in upper and lower case: this uses less space and increases legibility (here the same title in 18 instead of 16 points. For upper and lowercase 2 extra points of leading is sufficient at this type size, instead of four):

4 Line breaks can respond to units of meaning:

5 Important words can be distinguished through size:

6 Where possible, titles can be reduced to the essential keywords:

7 When setting titles all in caps, it is better to set them using a font that has small caps. These are designed to combine with each other instead of with lower case letters (The Sans)

4 Use of color and tone

There are two main concerns when using color, one is perceptual, and is related to the need to control contrast between letters and grounds (or figures and grounds in general), and the other is cultural, and relates to the symbolic or codified use of color. There are several national and international standards concerning the codified use of color. One must look into local uses in every case. The sign for exit, for instance, is red in North America and green in Europe. It has been internationally established that traffic lights and signs are red for stop, yellow for caution, and green for proceeding. Brown and blue are used for parks and information respectively. The enclosing shape of traffic signs (Octagon and triangle) and the position of lights in traffic lights (red at the top) help color blind people interpret the system. This is possible because in the process of obtaining a driver’s license one has to learn the code.

About 6% of healthy (some authors say 8%) adult males, have some form of color blindness (Grether & Baker, 1972, p. 71). These people tend to confuse red with yellow and green with white, hence in many situations they confuse red and green. The standard colors for danger, warning, and caution are red, orange and yellow respectively.

The poor perception of red suggests not to rely on color alone. When designing a color coding system, care should be taken so that the shades (degree of darkness) are sufficiently distinct for people with severely reduced sensitivity for color. If one analyses the US Army-Navy Aeronautical Specification AN-C-56; the scale of ten spectral colors developed by Chapanis and Halsey (1956); and the US Government Federal Specifications II-C-595 one finds that they are examples of sets built in search for a reliable distinction level by all sighted people.

The colors of backgrounds affect legibility. Luckiest and Moss measured the visibility of type on 10 colors of paper, and found that only the “Fairly saturated yellowish red or reddish orange” showed significant reduction in visibility, and also in reading speed. It seems that accommodation to conditions allows for a rather broad range of tonal contrast between type and ground, although black on white or on light shades of sepia, cream and buff, tend to be preferred (Luckiest and Moss, 1933, p. 25).

The visibility of black type on color ground, or color type on black ground, depends to a great extent on tonal contrast. Miyake, Dunlap and Cureton studied the ability of subjects to correctly respond to a series of brief exposures of black numerals on color paper (1930, p. 340). They found that black type read very well on white, yellow and green (I suppose, a light green), but red and green type on black ground resulted in low legibility scores (Unfortunately the colors are not reported with precision and it is therefore difficult to generalize results). Tests of reading speed show that red print on a dark green background can extend reading time by about 40%. Tonal similarity and hue conflicts are key situations to avoid when planning colored type on colored ground.

Color combinations that result in poor legibility cause an increase in regressions, an extension of fixations’ duration, and therefore an increase in perception time; they increase fatigue and overall slow down the reading speed. Color does not appear to affect legibility in test situations in which size and tonal contrast appear as leading variables, but Nilsson (1999, p. 2) argues that this is due to using speed of perception as an indicator of legibility. When legibility measures are based on distance and not on speed, color appears as a relevant dimension. Nilsson reports on several figure-ground color relations that were significantly more effective than black on white: “green/white, black/yellow, and green/yellow, with white/green, blue/yellow, black/red and blue/white being at least equally effective” (p. 8). Optimum readability favors black type on off-white backgrounds, but as long as sufficient tonal contrast is maintained, a wide variety of color combinations can be used.

Color also has a bearing on meaning and attention: in certain contexts, red seems to be internationally associated with danger. Adams and Edworthy (1995, pp. 2232-3) found that in relation to a warning sign “the signal word had to be twice as big in black as in red to give the same perceived urgency.” Perceived tonal difference does not equal actual tonal difference. For a shade of black to appear as mid-gray, it has to have 20% black, as it is the case for the standard gray card used in photography to base light meter measures. On a 20% black ground, black type and white type appear with equal intensity.

Recommendation 29

Color type on color ground should maintain significant light intensity difference to guarantee legibility (approximately 50% perceived reflectance difference).

Recommendation 30

Poor situations such as medium to dark type on black ground can significantly reduce reading speed and should be avoided.

Recommendation 31

Long texts should normally be printed black on white or off-white ground.

Recommendation 32

White type can be used on color grounds as soon as the ground is perceptually 50% as dark as black, that means, equivalent to a 20% black screen.

Examples: In “Management of Dyslipidemia and the Prevention of Cardiovascular Disease/Monitoring supplement,” the title “MONITORING SUPPLEMENT” is placed on top of a band that is too close in tone to the type.

In “Model for Estimating the 10-Year Risk of Coronary Artery Disease (death or non-fatal MI)” the use of alternating horizontal light-gray and white bands for each line would facilitate eye-tracking the 10 different lines, instead of using different gray backgrounds for the three vertical columns that are easier to follow.

In “the review” (BC’s newsletter), the gray used for background of titles in boxes is too dark, creating difficulties particularly for the legibility of small texts. Those grays should be reduced to a 20% of black.

Recommendation 33

Codified use of color should be consistent with national standards, and be maintained consistently in a collection of documents.

5 Evaluation: specificity of each situation of implementation

This report is written from the perspective of the practice of design. In the practice of design scientific information is used as much as possible, with the understanding that real situations are complex and present specific conditions, and that the existing knowledge always requires intelligent implementation, sensitive to the different conditions confronted in each case. While typographic quality contributes to the efficiency and effectiveness of scientific information in regards to both comprehension and compliance, there are many other factors at play as well.

This report is written with complete awareness that the central problem is not to design a text, but to increase people’s safety through the implementation of the best medical practices. This requires a good understanding of the public being addressed and the situations of implementation (Frascara, 1997, p. 8).

Recommendation 34

The effectiveness of new documents should be monitored once they are implemented to assess the degree to which the objectives have been met.

REFERENCES

Adams, A. S. and Edworthy, J. (1995). Quantifying and predicting the effects of basic text display variables on the perceived urgency of warning labels: Tradeoffs involving font size, border weight and color, Ergonomics, 38/11, 2221-2237.

Anderton, P. J. and Cole, B. L. (1982). Contour separation and sign legibility. Australian Road Research, 12, 103-109.

ANSI Z535.4-2002. American National Standard for Product and Safety Signs and Labels.

Appleton, J. (1975). The experience of landscape. London: John Wiley & Sons.

Chapanis, A. and Halsey, R. (1956). Absolute judgements of spectrum colors. Journal of Psychology, 42, 99.

Craig, J. (1999). Designing with type (4th ed.). New York: Watson-Guptill.

Crosby, T., Fletcher, A. and Forbes, C. (1970). A sign systems manual. New York: Praeger Publishers.

Desaulniers, D. R. (1987). Layout, organization and the effectiveness of consumer product warnings. In Proceedings of the Human Factors and Ergonomics Society 31st Annual Meeting (pp. 56-60). Santa Monica, CA: Human Factors and Ergonomics Society.

Desaulniers, D. R., Gillan, D. J. and Rudisill, M. (1988). The effects of format in computerbased procedure displays. In Proceedings of the Human Factors and Ergonomics Society 32nd Annual Meeting (pp. 291-295). Santa Monica, CA: Human Factors and Ergonomics Society.

Dixon, P. (1987). The processing of organizational and component step information in written directions (pp. 24-35). Journal of Memory and Language, 26.

Duncan, J. and Konz, S. (1976). Legibility of LED and liquid crystal displays (pp. 180-186). In Proceedings of the SID, 17(4).

Felici, J. (2003). The complete manual of typography. Berkeley, CA: Peachpit Press.

Frascara, J (Ed., in press), Designing effective communications. New York, NY: Allworth Press.

Frascara, J. and Ruecker, S. (2005), Final Report, submitted to the Alberta Drug Utilization Program, April 18.

Frascara, J. (1999). Cognition, emotion and other inescapable dimensions of human experience. Visible Language, 33.1, 74-89.

Frascara, J. (1997). User-centred graphic design. London: Taylor & Francis.

Grether, W. F. and Baker, C. A. (1972). Visual presentation of information. In H. P. Van Cott and R. G. Kinkade (Eds.). Human engineering guide to equipment design. Washington: US Government Printing Office.

Hartley, J. (2004). Designing instructional and informational text. In D. H. Jonassen (Ed.), Handbook of research on educational communications and technology (2nd. ed., pp. 917-947). Mahwah, N. J.: Erlbaum.

Klare, G. R. (1984). Readability and comprehension. In R. Easterby and H. Zwaga (Eds.), Information design (pp. 479-495). London: Wiley & Sons.

Lakoff, G. and Johnson, M. (1980). Metaphors we Live by. Chicago: University of Chicago Press.

Luckiest, M. and Moss, F. K. (1933). The dependency of visual acuity on stimulus distance. Journal of the Optical Society of America, 23(1), 25-29.

Miller, G. A. (1956). The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information, The Psychological Review, vol. 63, pp. 81-97.

Miyake, R., Dunlap, J. W., and Cureton, E. E. (1930). The comparable legibility of black and colored numbers on colored and black backgrounds, Journal of General Psychology, 3, 340-343.

Morrow, D. G., Leirer, V. O. and Sheikh, J. (1988). Adherence and medication instructions: Review and recommendations, Journal of the American Geriatric Society, 36, 1147-1160.

Nilsson, T. (1999). Legibility and visual effectiveness of some proposed and current health warnings on cigarette packages. Prince Edward Island: Health Canada-UPEI.

Oberly, H. S. (1924). The range of visual attention, cognition and aprehension. American Journal of Psychology, 35, 332.

Pillsbury, W. B. (1897). A study in apperception. American Journal of Psychology, 8, 315-396.

Poulton, E. C. (1972). Size, style and vertical spacing in the legibility of small typefaces. Journal of Applied Psychology, 56, 2, 156-161.

Ruecker, S. (2003). The affordances of prospect. Unpublished doctoral dissertation, University of Alberta, Edmonton.

Sanders, M. S. and McCormick, E. J. (1993). Human factors in engineering and design (7^th ed.). New York: McGraw-Hill.

Schiepers, C. (1976). Global attributes in visual word recognition. Nijmegen: Centrale Reprografie.

Shaver, E. F. and Wogalter, M. S. (2003). A comparison of older vs. newer over-the-counter (OTC) nonprescription drug labels on search time accuracy. Proceedings of the Human Factors and Ergonomics Society 47th Annual Meeting (pp. 826-830). Santa Monica, CA: Human Factors and Ergonomics Society.

Silver, N. C. and Braun, C. C. (1993). Perceived readability of warning labels with varied font sizes and styles, Safety Science, 16, 615-625.

Silver, N. C., Kline, P. B. and Braun, Curt C. (1994). Type form variables: differences in perceived readability and perceived hazardousness. Proceedings of the Human Factors and Ergonomics Society 38th Annual Meeting (pp. 821-825). Santa Monica, CA: Human Factors and Ergonomics Society.

Sinclair, S., Moore, S., Lavis, C., and Soldat, A. (2002). The influence of affect on cognitive processes. In J. Frascara (Ed.), Design and the social sciences (pp. 178-193). London: Taylor & Francis.

Spencer, H. (1968). The Visible Word. London: Hastings House.

Smith, S. L. (1984). Letter size and legibility. In R. Easterby and H. Zwaga (Eds.), Information design (171-186). London: Wiley & Sons.

Tinker, M. A. (1963). Legibility of print. Aimes: Iowa University Press.

Tinker, M. A. and Paterson, D. G. (1928). Influence of type form on speed of reading. Journal of Applied Psychology, 12, 359-368.

Wogalter, M. S., Howe, J. E., Sifuentes, A. H., and Luginuhl, J. (1999). On the adequacy of legal documents: factors that infl uence informed consent. Ergonomics, 42, 593-613.

Wogalter, M. S. and Post, M. (1989). Printed tutorial instructions: Effects of text format and screen pictographs on human-computer task performance, Proceedings of Interface 89 (pp. 133-138). Santa Monica, CA: Human Factors and Ergonomics Society.

Wogalter, M. S., Shaver, E. F. and Chan, L. S. (2002). List vs. paragraph formats on time to compare nutrition labels. In P. T. McCabe (Ed.), Advances in ergonomics 2002 (pp. 458- 462). London: Taylor & Francis.

Wogalter, M. S. and Vigilante, W. J. Jr., (2003). Effects of label format on knowledge acquisition and perceived readability by younger and older adults. Ergonomics, 46, 327-344.

Wright, P. (1994). Enhancing the usability of written instructions. In Proceedings of Public Graphics(pp. 26-30). Lunteren, The Netherlands: Delft University and Utrecht University.

Young, S. L., Laughery, K. R., and Bell, M. (1992). Effects of two type density characteristics on the legibility of print. Proceedings of the Human Factors and Ergonomics Society 36th Annual Meeting (pp. 504-508). Santa Monica, CA: Human Factors and Ergonomics Society.

Useful additional references

U.S. Army-Navy Aeronautical Specification AN-C-56.

U.S. Federal Register of the Federal Drug Administration/1999.

U.S. Government Federal Specifications II-C-595.

U.S. Military Standard 1472B (1974).

GLOSSARY

(For an extended glossary on typography see James Craig, 163-171; and James Felici, 293)

Ascender: the upward vertical stroke of a letter above the height of the lowercase x, i.e. as in b, d and f, for instance.

Counter: space inside a letter, derived from the term “counterform.”

Descender: the downward vertical stroke of a letter below the base of the lowercase x, i.e. as in g, j, and q, for instance.

Flush left: text set with optically constant space allowed between words, resulting in longer and shorter lines that create a ragged right profile for a column. Also range left/ragged right. Opposite: Flush right

.

Font: a given proportion and weight of a typeface, i.e. roman bold.

Justified: a text set to fi t the width of a column through the manipulation of the space between letters.

Leading: space between two lines of text, derived from the time when this space was created by using thin or thick pieces of lead between lines.

Pica: a measure equivalent to 12 points, used to indicate the width of columns.

Point: a measure equivalent to 1/72 of an inch.

Solid setting: a text with no extra space between lines, i.e. with no leading. A 10-point type text set so that there are 10 points of distance between the base of every line of text.

Stem: a vertical or near vertical stroke.

Stroke: the linear mark that draws the form of the letter.

Typeface: a particular form of representing the alphabet, in all its variations of proportions and weights.

Unjustified: a text set normally against a vertical left margin but allowing each line to have different lengths depending on the number of characters. Also ranged left/ragged right. It could also be set against a right margin.

X-height: the height of the face of a font measured on the basis of the letter X, that has a fl at top and bottom, and has no ascenders or descenders in lowercase.

Weight (also tonal weight): the relative ratio stroke thickness to white space in a letterform, defined also as the stroke width to height ratio.

PART 2 : Example: Preliminary study for a redesign of the COPD Academic Detailing sheet

Note: The text that follows is based on a section of the previous report on the COPD Academic Detailing sheet, and has been extended to describe typographic decisions in greater detail (See item 5 below)

During our first round of study developed around the evaluation and re-design of the COPD Academic Detailing sheet, it was found that these instruments are used in two basic situations: 1) general review and updating when the user sees the sheet for the first time; and 2) quick reference, usually when looking after a specific patient. For this purpose, the sheets should facilitate finding, comprehending, and remembering information. They should also serve to refer general practitioners to sources where more detailed information can be found.

The performance specification list can be outlined as follows:

1. It should be easy to find a sheet in a group of similar material.

2. It should be easy to store the sheet.

3. It should be easy to identify the topic of the sheet.

4. It should be easy to follow the intended sequence of the text.

5. It should be easy to read all texts.

6. It should be easy to find a specific section in the text.

7. It should be easy to identify different degrees of severity where they exist.

8. It should be easy to find references to drugs.

9. It should be easy to remember the contents of the sheet.

Design responses to these requirements

1. It should be easy to find a sheet in a group of similar material

Each topic for an academic detailing sheet will have a distinct color. Wherever possible, this color may be based on the color of the corresponding disease in the AMA Guidelines. The main titles For the whole sheet and for its main sections are white, reversed out from an 80% tint of the main color. All colors for the system of sheets are dark, so that white type will read well even on an 80% tint. The backgrounds for the inserts are a 15% tint of the same color, so that black type can be read well. A color bar at the top and at the outer edge of the sheet facilitates finding the right sheet in a binder.

Colors selected should be based on existing standards, but a high perceptual performance expectation might require to review existing standards.

2. It should be easy to store the sheet

The sheet is lettersize, laminated, and is prepared with three holes for optional storage in 3-hole binders.

3. It should be easy to identify the topic of the sheet

The main topic of the sheet is conspicuously set at the top and on the edge of the front and back of the sheet, set in white against the color used for the particular sheet. The color areas will also help, and, where appropriate, they will be in harmony with the color system used in the AMA Guidelines.

4. It should be easy to follow the intended sequence of the text

The reading sequence has been consistently arranged in two columns. Section titles have been numbered and made conspicuous through the use of white lettering on color ground, avoiding the tonal difficulties of the previous version, and facilitating recall of the number of sections in one sheet. This also intends to facilitate reference. A vertical line has been added to separate columns from one another.

5. It should be easy to read all texts

The font chosen is “The Sans” by Lukas de Groot (figure 7). This type is characterized by consistency of stroke thickness, that facilitates use in small sizes without affecting the perception of individual elements of a letter. It also has well resolved endings, preventing confusion between similar letters (figure 8). The font has a full set of variables, which include the “extra-bold,” used in this case to highlight keywords within the body of the text.

The selection and use of fonts and setting sizes, styles, and values, respond to needs for hierarchies and groupings based on principles of integration and segregation, and involve minor optical adjustments.

Font: The Sans (exception: numerals in Trebuchet for inserts)

Main title top of page: 24 pt, The Sans Bold, bold caps, reversed out, with +10 letterspacing (Reversed out titles are semi-bold so that they appear similar in weight to the bold variety used for titles. The extra letterspacing is advisable for texts in all caps and for large typesizes)

Pre-title top of page: 8 pt The Sans, Sans Caps (the pre-title and the copyright line in smaller size helps concentrate the attention on the substance of the document). The copyright line for top title: 8 pt The Sans Italics

Main title on side of page: 18 pt The Sans Bold, bold caps, reversed out, with +10 letterspacing

Signature on side of page: 8 pt The Sans italics

Section titles: 16 pt The Sans Semi-bold, bold caps, reversed out

Numerals in section titles: 18 pt The Sans Extra-bold, bold caps (The numerals 2 pts larger than the titles do not stand out excessively but reinforce their presence and help memorization)

Text: 9 pt regular on 12 pt

Subtitles in text: 9 pt The Sans Extra-bold, bold italics

Highlighted words within the text: 9 pt extra-bold upper and lower case (the extra-bold possibility offered by the typeface chosen allows easy location of highlighted words).

Stronger highlighting within text: 9 pt extra-bold, bold caps

Titles in inserts: 12 pt The Sans Bold

Sub-titles in inserts: 10 pt The Sans Extra-bold

Numerals in inserts: Trebuchet 10 pt bold (the old style numbers for The Sans could be confusing for this purpose because of their variable baseline and x height).

Column width: 19 picas, allowing an average of 9 words per line, well within optimum readability standards.

Text editing: the layout attempts to keep units of meaning together, avoiding awkward breaks of sentences and words, as well as the leaving of a fi rst or last line of a paragraph in a different page or column. Hyphenation is used at a minimum.

Line lengths of a capacity for approximately ten words avoid the excessive line length of the previous version. Type shows consistent high contrast with the ground, avoiding the printing of color type on a color ground with similar tone, which hampers reading.

White spaces between typographic elements and in margins have been used to facilitate integration and segregation of units of information.

A white hairline separating different areas of color maintains the rectangular simplicity of each area and helps separate each unit of information.

6. It should be easy to find a specific section in the text

Subtitles have been emphasized through size and typographic weight, facilitating “sign posting” within the text.

7. It should be easy to identify different degrees of severity where they exist

A scale “caution/warning/danger” (yellow/orange/red) associated with the notions of severity of the disease, and based on ISO standards on color coding for safety signs, has been used to activate the visual appearance of the sheets and to call attention to levels of severity.

8. It should be easy to find references to drugs

Important content in the text has been highlighted in extra-bold face, to facilitate finding specific references to drugs or conditions.

9. It should be easy to remember the contents of the sheet.

It is hoped that the clear identification and numbering of sections, the use of a branding color for each topic, the consistent standardization of type sizes and weights, and the formal consistency of the layout will help recalling the structure of the sheet, and therefore its content.

PART 3 : Design guidelines for the visual presentation of information

The points presented in this document are supported by Part 1, “Design criteria for the visual presentation of information. ” For easy reference, the sections’ numbers here correspond to it.

1 Information content

1.1 Defining the information content

Define information content bearing in mind:

• the size of the final product (in our case, both sides of a letter-size sheet);
• the objectives of the piece (information update and quick reference); and
• other sources that complement the information and could be referred to.

Once the optimum level of information extension and detail are arrived at, content experts should begin to work with information designers, to determine:

• information segmentation,
• information sequence, and
• ways of presenting the information (i.e. prose, list, diagram, chart, table, etc).

1.2 Highlighting information content

Since quick understanding of what exactly a message is about fosters reading and compliance, keywords should be emphasized – both through subtitles and emphases within continuous text – to stress the relevance of the document to the users.

1.3 Number of units of information

The information should be divided into a maximum of five sections, and each section should be divided into a maximum of five parts. Numbering the parts helps memorization of both the number of units and their sequence in either time or importance, particularly when they are three or more.

1.4 Hierarchies: degrees, positions, size, and tone

• A maximum of three different levels of importance should be used (Extendable to five if indispensable).
• Size and tone should clearly indicate importance: working on white ground the most important item should be larger, darker, or larger and darker than the following one. Avoid contradicting indicators such as a title in grey larger but lighter than a subtitle in black.
• Levels of hierarchy in tables should be consistently located: unless otherwise established by specific standards, if worst category is placed above and best below for one table, they should be in the same position for all tables.

1.5 Editorial consistency

The organization of the information should be consistent across all sections of a document and across all documents in a set. Titles, subtitles, texts, notes, tables, diagrams, etc, should always be organized in the same way throughout a document or a collection.

2 Typography

2.1 Choosing typographic fonts

For running prose select a typeface recently designed or re-designed. Faces such as Times New Roman, Officina (Sans or Serif), and The Sans, are recommendable.

• they provide homogeneous tonal density in the composition of words (no individual letter stands out in a text);
• they facilitate letter identification (all letters are easy to distinguish from one another);
• they relate to the most usual fonts (familiarity with a font facilitates legibility); and
• they maintain sufficient distinction for ascenders and descenders (this facilitates word recognition).

The Sans has a wide range of weights (from extra bold to extra light), and therefore is useful when there is a need to distinguish bold from regular. It is also very good when using small sizes.

Times New Roman is very good because people are very used to it for continuing texts but it does not offer a wide range of styles and tonal weights.

Officina offers good legibility and is a narrow face, good when long texts put pressure on space available. It, however, does not offer a wide range of styles and tonal weights.

Georgia and Arial are good for electronic documents because they were designed for Microsoft Word.

2.2 Typographic styles

Roman regular: Continuing texts should be always set in medium weight roman style, also defined as regular, or plain.

Italics: Italics should be used within the text where distinction is necessary and no difference in hierarchy is intended.

Bold or extrabold: Bold type can be used within the text where there is a need for emphasis, or for titles that require distinction of importance. Use of a font that clearly distinguishes regular from bold is recommended. Some fonts include extrabold.

Light or extralight: this can be used when printing quality is optimal and there is a need for a secondary text. It can be used instead of reducing the size of type to indicate less importance, or for titles that are expected to be subtle.

2.3 Types of text

Continuous reading of long prose requires generous leading and use of upper and lower case.

Short titles and lists allow use of texts set in all caps.

Capital letters are more legible when used individually, and therefore could be used when identifying sections of a text or other units identifiable by one letter.

Information scanning requires a typeface with a wide range of weights for emphasis (from extra light to extra bold). No more than three levels of importance through differentiation of style should be attempted for information scanning within texts.

For directing, persuading or warning texts, capitals, bold style, and sans serif faces are recommended.

2.4 Legibility of letters / continuous text / comfortable reading

For continuing texts, sizes should be between 9 and 12 points in regular weight.

Condensed or expanded letters should be avoided.

3 Layout

3.1 Type size, column width, column style and leading

The fonts recommended above should be set with a line separation of at least 2 extra points. For narrow columns it is indispensable to format them as flush left.

Justified narrow columns generate awkward blanks between words that hamper reading. Spare use of hyphenation also helps.

The recommended column widths for different type sizes to secure optimum legibility using 2-point leading are:

• 6 to 8 point: 15 picas
• 9 to 10 point: 19 picas
• 11 to 12 point: 23 picas

Columns that are too wide or too narrow should be avoided, since they create reading problems. However sparingly, hyphenation use might be necessary when setting texts in narrow columns.

3.2 The use of space

Texts should be properly spaced between letters, between words, and between lines. Type should not be crammed or randomly spaced to fit column widths.

3.3 Spatial arrangement of text

Blank space is useful to facilitate integration and segregation of sections in a document. Words should be properly separated from one another and no attempt should be made to cram words in a line. Unjustified texts (flush left) are therefore to be favored. Whenever possible, texts should be broken in chunks to facilitate comprehension.

3.4 Setting titles

Titles set in all caps should be specially spaced, at about 0.7 pt of extra letter space for 20 pt type. If some letter pairs are too tight – like MI – or too open – like VA – they can be treated individually.

Long titles are better if set in upper and lower case: this uses less space and increases legibility. For upper and lowercase titles 2 extra points of leading is sufficient up to 20 points. For titles set in all caps 4 points are required. When setting titles all in caps, it is better to set them using a font that has small caps, that are designed to combine with themselves instead of with lower case letters. Line breaks in titles can respond to units of meaning: this facilitates comprehension and reading speed. Important words can be distinguished through size in order to focus the attention of the reader. Where possible, titles can be short, reduced to the essential keywords.

4 Use of color and tone

4.1 Color contrast

Color type on color ground should maintain significant tonal contrast to guarantee legibility (only light color type on dark color ground or vice-versa should be used). Medium to dark type on black ground should be avoided.

Long texts should normally be printed black on white or off-white ground. White type can be used on color ground if the background is equivalent to at least a 20% black screen.

4.2 Color coding

Codified use of color should be consistent with existing standards, and maintained consistently in a collection of documents.

4.3 Background tone

The use of light shades of gray or color can help distinguish units of text that need separation, such as in inserts or tables This can also serve to highlight chunks, or to relate elements with one another.

5 Evaluation: specificity of each situation of implementation

The effectiveness of new documents should be monitored once they are implemented to assess the degree to which the objectives have been met.

6 Summary: Steps of the design process /check list

1. Define information content (identify related information sources)

2. Define information segmentation (no more than five sections)

3. Define information sequence.

4. Define ways of presenting the information (i.e. prose, list, diagram, chart, table, diagram, etc).

5. Select a typeface that is readable in small sizes, has a good range of styles, and does not take too much space (For print: The Sans; for electronic documents: Arial or Georgia, that were specially designed for Microsoft Word. Also Times New Roman renders well on screen).

6. Define column width, considering between 7 and 12 words per line or 40 to 80 characters.

7. Use flush left composition (not justified), particularly for narrow columns.

8. Define type styles and sizes to distinguish hierarchies: titles, subtitles, sub-subtitles, text and emphases in text. Use size, tone and style consistently to distinguish levels of importance.

9. Define range of type sizes, between 6 to 12 points for text, and set it with two or three points of extra leading (line space).

10. Maintain editorial consistency so that the same kind of problem is solved in the same way across the document or the collection of documents.

11. Use blank spaces to facilitate reading and recognition of segments of information.

12. Use color and tone making sure that color or black texts maintain good contrast against their backgrounds, and that when color is used as a code it is in agreement with existing codes and it is used consistently.