Seminar Shatters Ancient Myth

by Susan E. Smith, Quill Contributor

Don't believe everything you read. And definitely don't believe everything you hear. That was the first lesson we learned from Jean-luc Dumont during his April 20th telephone seminar, "Magic Numbers", held at CheckFree in Waterloo.

Most of us are familiar with the "7 + or – 2 rule". This rule says that for any list, slide, or menu, you can include five, seven, or nine items, and the user will be just fine with it.

Well, turns out it's not quite that simple.

The Origin of the Magic Numbers

So where did this rule come from, anyway? According to Dumont, it was derived from Dr. George Miller's 1956 Psychological Review article, "The magical number seven, plus or minus two: some limits on our capacity for processing information". After reading this article, Dumont concluded that it has been misinterpreted all along. In his phone seminar, Dumont gave us his revised interpretation of Dr. Miller's research, along with some practical suggestions for presenting information that users can manage.

The 7 + or – 2 Myth: What George Miller Actually Meant

Miller's paper describes the results of a series of experiments involving ring tones (as in, one tone tells you your car door is open, a different one tells you your keys are in the ignition, another tells you your lights are on, and so forth). He tested to see how many different tones a person could remember and identify.

Subject testing produced the following conclusions:

• 3 tones – remembered easily (no errors)
• 4 tones – a few errors
• 5 tones or more – frequent errors

If you plot the results on a graph, you get a line that is straight at first, then curves towards a horizontal line (an asymptotical limit).

So where does 7 + or – 2 come from? It's a math thing. You take the mean (2.6) and the standard deviation (0.6) from Miller's data, and you plug them into some equations, and you come up with 7 + or – 2. Take my word for it. That's as technical as I want to get.

The number of items that can be remembered without error is not all that's important. The category of information should also be considered when determining a numerical limit.

Getting Around the Limit: Miller's More Useful Lessons

The true limit, it seems, is lower than we thought. Fortunately, there are ways around it. Miller proposed these strategies:

• Use relative judgments, rather than absolute ones. For example, use a graph to show size relative to a scale and to other measurements.
• Group related items together wherever possible (as is often done with menus, or links on a Web site). This gives the information a visual structure so that users can process it more easily. Another example is taking a large number like 6000000 and adding spaces to it (6 000 000). This lets you immediately see the structure, and you know instantly what the number is, without counting the zeros.

In addition, Dumont recommends that you:

• Use a book's table of contents as a way to quickly convey the structure of the document, rather than providing an exhaustive list of the topics within. The index does that, and does it very well. The contents should serve a different purpose. Keep it to two pages if possible by leaving out the lower-level entries so that it reveals the book's structure at a glance.
• Vary several stimuli concurrently as a way to encode the same information. For example, at an intersection, you see both a stop sign and a white line painted on the road. This is called effective redundancy. Giving the same message twice, in two different ways, helps to reinforce it. It also means that if one of the stimuli is not available (e.g., snow covers the line on the road), the message is still there.
• Use a sequence of absolute judgments instead of a single, more complex judgment. More simply put, this means chunking. For example, if you have nine items in a list, present them as three groups of three items.

Upper Limits Revisited: There Is Magic Everywhere

Three is the most magical number of all, because humans can look at a grouping of three things and instantly take in the fact that there are three, without having to stop and count. Three opens up a new dimension, and introduces a middle ground.

Five is also a magical number because it is the span of attention up to which humans can process globally. Five is a useful upper limit on the number of items in any list.

To ensure global processing, Dumont recommends these limits:

• Limit the number of items to five per group (upper limit).
• Limit the number of levels to three (a nice number for any grouping).
• Include no more than two levels in an overview; also use two for effective redundancy.

Beware of extrapolation: don't carry research over to where it does not apply. And, above all, don't believe everything you read.

About Susan E. Smith

Susan is a former editor of this newsletter who enjoys hiking, writing, photography, gardening, pie-baking, reading, movies, cats, tea, chocolate, and spending time with her friends. If you see her, ask her how that novel is coming along. If enough people ask, it just might make her feel guilty enough to actually start writing it. At last count, her combined lifetime income from non-technical writing and photography was $240.25. For now, at least, she plans to keep her day job at NCR.