Introduction

A few years ago, two studies were published that showed clearly how much trouble designers had when trying to predict human performance in their systems (Bailey, 1993; Tullis, 1993). More recently, Miika Silfverberg, Scott MacKenzie and Panu Korhonen at the Nokia Research Center in Finland conducted an interesting study comparing two devices. They had users enter text information into cell phones. One group entered the information using "one-handed thumb input," where subjects held the phone in one hand and pressed the keys with the thumb of the same hand. The other group used "two-handed index finger input," where participants held the phone in their preferred hand and pressed the keys with the index finger on their other hand.

Which way do you think is fastest? A group of 101 website designers and usability professionals were asked to indicate which would elicit the fastest performance (Bailey, 2001). Twenty-three percent voted for the "thumb input" and 77% thought the "two-handed input" would be faster. The Nokia study showed that the two-handed approach was 12% faster than one-handed. In this case, over 3 out of 4 participants guessed correctly.

A second study compared entry speed using the two-handed approach with the "T9" approach. T9 is produced by Tegic Communications in Seattle. T9 uses a dictionary to disambiguate the letters when entering text into a cell phone. For example, to key "t-h-e" users pressed 8-4-3-0 (the 0-key is used as a SPACE). T9 then compares the word possibilities to its linguistic database to "guess" the intended word. It shows the most common word as a default, and if users want to select another word, they press the star key for NEXT.

The same group of website designers and usability professionals were asked to indicate which would elicit the fastest performance – the normal (traditional) way of entering letters into a cell phone or the T9 approach. Fifty-eight percent voted for the "normal" method and only 42% thought the T9 method would be faster. The Nokia study showed that T9 was 35% faster than the "normal" method. T9 was amazingly fast, allowing entry speeds of 33 words-per-minute, compared with only 21 wpm for the "normal" approach. In this case, over half of the participants missed the essential cues for making the correct judgment.

Having designers guess the best way of achieving optimal user performance is very difficult. Their design decisions can be improved by ensuring that designers are familiar with the research literature, and by effectively using performance-based usability testing.