CHI+MED logo banner

Familiarity of numbers

Key Points

  • There are patterns in the numbers that medical workers have to type on hospital wards. This suggests that some numbers could be more familiar than others.
  • Often number entry systems are tested by asking participants to enter sets of random numbers, but this doesn't accurately reflect the real tasks performed in hospitals
  • The familiarity of a number has a significant effect upon how a user types it: familiar numbers are faster to type than non-familiar numbers.
  • We have shown that future experiments on number entry need to be designed in a way that takes account of whether numbers used are familiar to the participants.

Testing text entry interfaces
The text keyboard on your smart phone works well most of the time, even if occasionally it suggests an obviously ridiculous word it thinks you are typing. Mostly, you aren't even aware of all the subtle changes it makes to help, whilst you are typing. For example, it might correct spelling errors from your finger slipping, or give some keys a larger target area so they are easy to hit in the first place. It can do this because there is a wealth of knowledge both about the words you are most likely to type, and about how the average person types them. Designers tap into this knowledge to create the best possible experience for people typing text.

When typing numbers, though, similar levels of information have not been available until now. This means that interfaces used for entering numbers have not been optimised in the way text entry interfaces have. By collecting more information about the number entry process, the CHI+MED team aim to make recommendations and inform the future design of number entry interfaces.

Translating this to number entry interfaces
How can similar information be gathered for the task of entering numbers? With text for instance, it's known that typists copy non-words significantly slower than words. Can there be such a thing as a non-number? Decimal points aside, any digits put together can form a number as there are very few rules on how a number can be created. However, previous research has suggested that familiarity may affect the way we read numbers, much as the way we read alphabetic text is affected by its "word or non-word" status.

We wanted to know whether or not this phenomenon is important in healthcare. Doctors, nurses, pharmacists and many others have to enter numbers all the time: in to prescription systems, patient records, monitors, medical devices giving treatments, and so on. The first step was to investigate whether there are patterns in the numbers used in that setting. Through multiple studies, we have showed that some numbers occur far more frequently when programming some devices. When setting up infusions of drugs in one hospital, the digit 0 is used far more frequently than any other, for example, and the numbers 1,000, 100, and 50 were used in nearly half of all infusions we studied. This suggests that there are some numbers that medical workers will be much more familiar with than others.

The next step was to establish whether familiarity has an effect upon the way that numbers are transcribed. To find this out, we ran a series of controlled typing experiments, asking people to copy both words and non-words, and familiar and non-familiar numbers, in different situations. Familiar numbers included those that referred to key historical dates, such as 1066 or 1945, or numbers that related to cultural references such as 999 or 1984.

Effects of familiarity on number transcription
These experiments suggested that familiarity does indeed have an effect upon the way a number is transcribed. We replicated previous results from text entry studies, in that words were typed significantly faster than non-words. We also found evidence that familiar numbers were copied significantly faster than non-familiar numbers.

This effect had not been investigated previously, and as such many studies have not paid attention to whether or not the numbers they are using are familiar. This finding suggests that more care needs to be taken when studying number entry interfaces, as familiarity matters. This is particularly important for the medical domain, given our research has shown that certain numbers do arise more often than others. This may change the way that number entry is studied in future.

This work has ultimately shown that not all numbers can be treated the same in number entry research. Familiarity of a number matters: not all numbers are equal.

Wiseman, S., Cox, A. L., & Brumby, D. P. (2013). Designing devices with the task in mind: Which numbers are really used in hospitals? Human Factors, 55, 61–74.

Cauchi, A., Thimbleby, H., Oladimeji, P., & Harrison, M., (2013). Using medical device logs for improving medical device design.Proceedings of IEEE International Conference on Healthcare Informatics (ICHI 2013), 56–65.

Wiseman, S. E. M. (2014). Designing for numerical transcription typing: Frequent numbers matter. PhD thesis, University College London.