Appropriate Units - Revised
About a week ago, I was mulling over the concept of appropriate units, that is which unit of a system should one use to represent a measurement. If you want to read about it, have a look at this post.
At the end of that post I detail a possible solution, but I wasn’t sold on my solution and wanted a bit of feedback from a wider audience. I devised this questionnaire to provide an interesting and abstracted way of getting the data. Writing the form was interesting, and while I think most people got what I was getting at, there were a number of influencing factors that I should have taken into account. It has been way to long since I last did anything of the type. People were influenced by the choice of unit names and listed colours. I’m not particularly concerned by the unit names as this was fairly unavoidable if I didn’t want to use real units. I also tried to be consistent within a question to avoid favouritism due to it being a interesting word. My rationale for not using real units was that I wanted to avoid the associations between using a cup or using a quart.
Most of my answers come from a group of people primarily using the metric system, and as such a cup would have been favoured to a high degree because of its name. Colours probably had some influence, but I suspect that it was minimal.
An informal analysis of the first set of results I obtained proved interesting, and was corroborated by the comments I received.
Most people don’t unconsciously make a distinction between using a measure and making a measurement. But when forced people do make it. This turned out to be a vital distinction as measuring 160 mL is ok, but 160 teaspoons is definitely not.
In the case of using a measure the main factors were that it had to be as close to one as possible, numbers above should be exponentially weighted as they get higher, numbers below (i.e. between 1 and 0) should be inverted and then weighted in a similar fashion. In addition measures should be favoured over measurements, as:
- More people have basic measurement equipment like cups in the kitchen.
- Measures, while on the whole less accurate, are accurate enough for most forms of cooking.
- People seem to like using measures more. This may just be a habit.
In the case of measurements, the following factors seem important:
- A small starting digit.
- The last significant figure should be weighted in favour of 5 or 0, followed by 2
- Numbers that benefit from being in scientific form of negative should be penalised
- Long numbers should be penalised
- Interestingly, whole numbers are not as important as I thought, however, limiting the number of leading and trailing zeroes is.
I have implemented an algorithm based on these results. It seems to work pretty well. It’s one limitation is that its results are based on a static number of significant figures that may not be the same number of significant figures as the user wishes to use. However, this point is not particularly relevant if the code base is used specifically for cooking as most cooking statistics have only 1 to 3 significant figures.
Thanks to everyone who completed the form for me!