Clearly the 167 cm person could vary enough to fall in either the 165 cm bin or the 170 cm bin. So now he he has two heights, 5 cm apart instead of of varying ± 1 or 2 cm. Has this really led to improvement? I think restricting the least significant digit (LSD) to particular values is an added complication that adds little value (it looks ominously like a "half" of something, perish the thought).
Given the variation, measuring to millimeters or finer is clearly "decimal dust." Measuring to centimeters results in more than ±0.5 uncertainty in the LSD, but not so much that I would entertain measuring only to the nearest decimeter. I would measure to the centimeter and accept that I might have variation of as much as 3 in the LSD. Precision comes in decades when you use decimal fractions, and your choices are a little more or a little less than what you really want. Usually, the best choice is a little more, but avoid 10X more or greater. If I were using height as one of several biometric markers for ID, I would need to be aware of personal variation within a day, but I would prefer to measure with more accuracy than that "tolerance." Then, at some future time, instrument error would not compromise hypothesis testing of whether a person is of such different height that he could not reasonably be some particular person of "known" height, using classical quality control concepts. ________________________________ From: Pat Naughtin <[email protected]> To: U.S. Metric Association <[email protected]> Sent: Sun, December 12, 2010 1:20:35 AM Subject: [USMA:49169] Human height and centimetres Dear All, As you know your height varies through each day. Some heights vary only by about 10 millimetres (1 centimetre) while others vary by as much as 38 millimetres (3.8 centimetres that looks a lot like an inch and a half). The average variation is probably about 20 millimetres (2 centimetres) although I cannot find clear data for this. I was brought up to the idea that whatever you measured could vary by half a unit, or half a division, up or down. So if someone says, "I am 167 centimetres tall", how should I respond? Is the centimetre suitably accurate and precise for measuring human height? Is "What time did you measure your height?" a suitable response? We tend to lose more height between rising and 10:00 a.m. than we do for the rest of the day. Some time ago, I was involved in training police in my home state. We confronted this problem head on when we measured the height of each police trainee before and after exercise only to find that their height varied (probably due to intervertebral disc compression). Eventually we decided that the vest we could do to record and report the height of criminals was to use metres with two decimals rounded to end with a 0 or a 5 so the person quoted above would become 1.65 metres. Any thoughts? P.S. For the Body Mass Index (BMI) watchers among you, it is probably best to record your mass and height as soon as you rise in the morning -- you have less mass before breakfast and you are taller! Cheers, Pat Naughtin LCAMS Author of the ebook, Metrication Leaders Guide, see http://metricationmatters.com/MetricationLeadersGuideInfo.html Hear Pat speak at: http://www.youtube.com/watch?v=_lshRAPvPZY PO Box 305 Belmont 3216, Geelong, Australia Phone: 61 3 5241 2008 Metric system consultant, writer, and speaker, Pat Naughtin, has helped thousands of people and hundreds of companies upgrade to the modern metric system smoothly, quickly, and so economically that they now save thousands each year when buying, processing, or selling for their businesses. Pat provides services and resources for many different trades, crafts, and professions for commercial, industrial and government metrication leaders in Asia, Europe, and in the USA. Pat's clients include the Australian Government, Google, NASA, NIST, and the metric associations of Canada, the UK, and the USA. See http://www.metricationmatters.com/ to subscribe.
