Comment Most engineering degree programs - including those labeled 'human factors engineering' and "human factor psychology'- do a poor job of teaching students about human variation in general, and anthropometry and biostatistics specifically. As a result, experts spend a large portion of professional time sorting out well meaning (but dangerously ignorant) scientists and engineers. What can standards and guides to to improve this situation?

---

-- BrianSherwoodJones - 06 Jul 2005

> >
Have to agree Brian. My experience after doing workstaion assessments for the past 5 years is that there is a lot of ignorance from furniture/equipment designers on what designing for the population means. There appears to be a universal assumption that 5th or 95th percentile measurements (for chairs for instance) are only concerned with height. Any practising ergonomist meets users with short/long backs, short/long legs and all measurements in between, any combination of which can produce a 5foot 10in person. OK, for ergonnomists that is a duh! moment, but try explaining that to a client who thinks that setting up a workstation demands no more than the purchase of an "ergonomic" chair and desk combination. Couple that with typical Bridge layout problems of visual arcs of sea and ship, radar, alarm panels, etc.,etc., and a dirth of anthropometric data for the mariner population, then we have a real basic problem.

--TWiki.TomMayfield - 14 Jul 2005

> >
To my mind, these problems are made worse by the data provided (or not provided!) in some of the more readily accessable anthropometric databases:

• The data collected are not always those most useful for design purposes. A case in point was a major survey which provided stool height, sitting height (seat to top of head) and pupil-top of head distance. Designers will normally need to know such things as seated stature (which is not nth percentile stool height plus nth percentile sitting height) and - crucial for display design! - seated eye height (which is not nth percentile stool height plus nth percentile sitting height minus nth percentile pupil-top of head distance)
• Few databases provide correlations between dimensions. The naive designer assumes that r = 1.0. Of course, as soon as you work with correlated data you find your confidence intervals widening, reflecting population variability.

Yes, designers at using stature is better than nothing. I believe that I was hired by one consultancy on the basis that I am 97.5th percentile male UK stature - and on one occasion fairly recently I found it highly convenient to have a short female colleague present during a mockup review. Such 'quick and dirty' methods can be useful as a first pass. But it's encumbent on the Human Factors community that we should a) caveat our anthropometric data heavily and b) provide useful, non-misleading data in the first place.

A moan: My recollection is that the UK 'Adultdata' survey (and related surveys for children, the elderly and people with disabilities) was one of the few that did supply correlations. It's been out of print for a long time. The raw data has not, as far as I'm aware, ever been in the public domain - although I gather it's used as the basis for one commercial product. This is work that I, as a taxpayer, funded!

-- DavidCarr - 22 May 2006