But Will They Get It?

Most of us are familiar with readability statics, at least if you routinely spell check documents created in Microsoft Word. You know . . . that Readability Statics dialog box that pops up after your spell check is complete, like the one shown below.

The two bottom lines show readability metrics for two popular algorithms, Flesch Reading Ease and the Flesch-Kinkaid Grade Level. Ok . . . admit it . . . you have been interested in seeing how high of a grade level has been assigned to your prose, at least once.

So we see these statics routinely, but what do they really mean, and how can they used to craft A Better Message?

THREE INDICES

There are several common metrics for readability. They attempt to quantify how easily a message can be read and understood by evaluating patterns in text. For three common metrics, the parameters evaluated are sentence length and the amount of multisyllabic words used in those sentences. The metrics I have encountered most often are Flesch Reading Ease, Flesch-Kinkaid Grade Level, and the Gunning-Fog Index. These are calculated as follows:

Gunning-Fog Index:

·       Grade Level = [(Average Sentence Length) + (percentage of multisyllabic words)] x 0.4

·       The exact method for accumulating data and calculating the result can be seen here.

·       The result suggests the grade level for which a text is most appropriate.

Flesch Reading Ease:

·       Flesch Reading Ease = 206.835 - (1.015 x Average Sentence Length) - (84.6 x Average Syllables per Word)

·       The method for counting words for use in the formula is given here.

·       The result occurs within a range of 0-100, with 100 being the most easily read. A table that further defines the score is given here.

Flesch-Kinkaid Grade Level:

·       Grade Level = (0.39 x Average Sentence Length) + (11.8 x Average Syllables per Word) – 15.59

·       The formula and counting method are expressed here.

·       Again, the result suggests the grade level for which a text is most appropriate.

As you can see, the results for the first and last indices are expressed as grade level. I did not find a descriptor on this, but I assume these reference common grade levels in the US.  I am not sure that these indices translate directly to grade levels in the educational systems used in other countries. If you know, please post a comment . . .

THEIR APPLICATION

In order to see how these indices rated text, I ran several passages through a readability calculator that provides readability statics according to multiple parameters. Here is what I found:

From my October 29, 2009 blog post, “Web 2.0 for Gearheads”:

Flesch reportedly identified 65 as a plain language index, so my text is just about there, but the grade levels assigned by Gunning-Fog and Flesch-Kinkaid differ by 4 years.

I also ran some text from the Starbucks website – here’s what came up:

This passage showed similar, slightly lower trends, and again showed a significant difference between Gunning-Fog and Flesch-Kinkaid.

Given contemporary concerns, I tried a passage from flu.gov, which provided this result regarding a passage addressing the flu concerns for pregnant women:

These results showed the site requires the reading skills of a 6^th to 8^th grader, which is appropriate for general audiences.

A review of a short text from my local motor vehicle regulator provided these results:

Hmmm . . . grade 14 to 18 grade for a site almost every adult in the State of New Jersey might have to reference . . . I have my doubts.

The indices above are valuable in several ways. They can help authors tailor content to an intended audience. In the case of young students, these indices can ensure that readability is maintained by keeping adult-written language simple to understand. For a public-facing communication, these indices can be used to ensure that the language remains sufficiently plain to be clear to a very broad audience, or, in the case of the NJDMV, flag a comprehensibility problem.

Applying these indices effectively suggests that the intended audience is known by the author. If it isn’t, keeping the text to grade 7 or 8 (or in the case of Flesch Reading Ease, above 60-70) is a good guideline.

READABILITY OF VISUAL DESIGN

Someone asked me if these indices might be used to assess visual design of documents. I suppose an algorithm might be worked out given an industry or academia-based effort to build consensus on how to measure visual "readability". In the case of text, the indices I selected use syllables and sentence length, two easily-defined qualities. But defining common visual literacy for the subjective perception of qualities such as color, unity, rhythm, contrast, typographic effects and the host of other design parameters seems daunting.

I suppose one could establish “visual readability” by analyzing performance-based measurements derived by experimentation. But that sounds a lot like usability testing. Perhaps when some critical mass of usability research has been completed, the complex patterns involved in visual comprehension will become more reproducible, allowing for development of complex algorithms. Until then, I see that usability testing will be required to establish visual design functionality.

What are your thoughts? Are there visual readability hypotheses that I am not aware of? Or is the subject just too gnarly to tackle in a consistent manner?

By the way, MS Word is telling me this piece has a Flesch-Kinkaid Grade Level of 10.8 . . .

The Presidents and Their Approval Ratings

On October 29, 2009, I posted a piece on imagery, which included several data visualizations. I am not interested in making this a political blog, but  I recently stumbled across an interesting interactive data visualization that I thought I'd post here.

This one shows the presidential approval ratings data for each US president since Harry Truman. There is a static screenshot below - go to USA Today's site to interact with the visualization, which you can adjust to compare specific presidents.

Interestingly, it shows repeated cycles of high approval at the outset of each Presidency followed by lower ratings at term's end, except for Presidents Clinton and Reagan. I wonder how the movement of the 9/11 terror trial to New York City, announced yesterday, will ultimately play out for Mr. Obama's approval ratings at the end of his term?

Nice job by William Couch, Kristen Novak, Michelle Price, and Joshua Hatch at USA Today.

A Website for Reflective and Visual Learners

I recently built a website on GoogleSite. It is designed for an audience with a particular learning style - Mine. The site, Torque Wrench Basics, trains viewers to use a torque wrench. To identify my learning style, I completed the Index of Learning Styles Questionnaire, sponsored by North Carolina State University. There are four continuums on which learning styles are evaluated. Learners are:

1. Active to Reflective
2. Sensing to Intuitive
3. Visual to Verbal
4. Sequential to Global

A discussion of the various learning styles can be read here. My profile was identified as follows:

My profile indicates that I am a moderately-to-strongly reflective and visual learner, with a tendency towards sensory rather than intuitive learning. To accommodate an audience with my learning tendencies, I used design elements that would help learners like me understand and retain the training content. The elements that made it into my final design are listed below by each learning style component.

Active vs. Reflective: Moderately-To-Strongly Reflective 
Active Learners like to learn by doing. Reflective Learners tend to learn by thinking about the matter for a while. I am a little of both, but more of the latter. To satisfy this description, I included two elements in my site:

• Review questions, where appropriate, to allow for reflection during the training.
• Links to sources outside of the training tool for additional information, study and reflection.

The review questions were included to force the viewer to do what they do best, reflect on the material in a particular section of the site. I used a two-page format for my reviews - the first was a questions page:

Following the "Answers"  link brought a copy of the page with the corresponding responses:

Sequential viewing of these pages immediately following the related training content allows, even forces, time for reflection by target audience to improve understanding and retention.

Links to outside sources of information, such as dictionary definitions and Wikipedia articles, were included to promote further reflection.

Sequential vs. Global: Neutral

Sequential Learners thrive when there is structure and order, while Global Learners take a more eclectic approach, jumping around until the piece s fall into place and they suddenly "get it". With neutral results on this scale, I thought it best cater to both traits. To address both ends of the spectrum, I incorporated the following features:

• Sequentially-numbered major pages for sensing learners. These presented the content in 5 modules.
• Color-coded icons at the top of content pages that show global learners the relationship of the page materials to the remaining site content.
• Titles in the content area colored to match the training module in which the page is located, again for global learners.

All of these features are shown below. The red title nearest the content indicates that this page is located within Module 1: Principles.

I also used sequential links at the bottom of each page for sensing learners, and site map that could be expanded to show the all of tha pages across the site, visible on every page for global learners. These features are show in the two screenshots below.

Sensing vs. Intuitive: Slightly-to-Moderately Sensing

Intuitive Learners thrive on learning the relationships within the subject of study. Sensing Learners are better with facts, but need to know the practical applications for the information. I used the following featuers to accommodate these traits.:

• Inter-page navigation links labeled as to the content, for example “Calibration” rather than “Next”.
• Content that favors actual use over theoretical constructs, related to field or job applications.
• Graphics composed of actual photographs or photo-realistic illustrations rather than abstract line diagrams.
• Supplemental fact-based content that reflects “real world” applications, such as lists of, and links to, calibration labs.
• A question-based navigation feature, picture in the next two screenshots, that:
• Allows answers to be found quickly.
• Responds to Sensing Learners’ affinity to application-infomation. 

The links in the screenshot above bring the user to the page that responds to the listed questions, an application-oriented result that caters to our Sensing Learner's need for practical information.

Visual vs. Verbal: Moderately-To-Strongly Visual

The learning style for this audience is strongly visual. To met the needs of Visual Learner, textural information was minimized. Again, photographs and photo-realistic sketches were used to illustrate key topics, and in two instances vidoes were used. In the first instance, a short video was used to allow Visual Learner's to see and hear how a torque wrench works. This video was very, very short:

The next video also meets the needs of Visual Learners, but was also used to underscore the real world consequences of the training content, something that could prove important to our Sensing Learners.
   


Closing Thoughts
Developing the actual content for the site proved to be fun. Developing the site itself proved to be a bit more challenging. Google Sites didn't have the feature sets I had originally planned on, such as buttons that functioned as links, and this caused me to implement solutions that were more crude than my intentions.  All the same, I think the site hits the design objectives, and the process I used to adjust this site is one that can be used in other web-based training applications. After all, adjusting the message based on audience analysis is an important key in creating A Better Message.