[Sigia-l] length of nav labels

[Jared M. Spool]

At 12:58 PM 8/8/2005, Marcel van Mackelenbergh wrote:

>One remark though on your 7 - 12 guideline. This guideline (and more
>often the 5 - 9 guideline) just keeps on popping up too many times. It
>comes from a research conducted by Endel Tulving in 1964. Tulving had
>his subjects of experiment recall unrelated words and found pretty
>constant numbers. What is important here is that these words are
>unrelated. Once the reader sees that words are related, the reader will
>create larger and therefore fewer chunks of information.

Actually, our 7-12 guideline *didn't* come from either Tulving or Miller.

It came from *our* analysis of more than 3,000 clickstreams, where we 
compared the success of the clickstreams to the number of words in each 
link that appeared in the clickstreams. Clickstreams that contained links 
with 7 to 12 words were far more likely to end in success.

(We looked at dozens of other variables, like the types of words, the 
placement of the underline portion of the link, the color of the link, and 
many, many others. Link length was the strongest predictor of success.)

Turns out that the number seven turns up in all sorts of results, not just 
memory chunking experiments.

>You would make me happy if you replaced "words" by "chunks".

Except that wouldn't describe the results of our research. Sorry I can't 
make you happy this time. Maybe next time. ;-)

Jared


Jared M. Spool
User Interface Engineering
http://www.uie.com    jspool at uie.com

UI10 Spotlight Presenter: Flow author Mihaly Csikszentmihalyi
See details at http://www.uiconf.com