Pat Naughtin wrote:
>>Some of the values were rounded to the nearest 5 mm.
> 
>This must be very annoying to those who want to use this data.

It is very annoying. Most people don't do it, but Pheasant did. I remember
being surprised at the time he presented the data set.

A rounding of 5 mm is quite a lot given what often happens with the data. If
the numbers were used 'as is', then it would not be so bad. Pheasant himself
said that he did the rounding because, for example, stature is not a precise
dimension. I agree with him on that.

However, I rarely use the data without some calculation. One of the most
common calculations is to obtain other values (e.g. 99th percentile) using
the mean and standard deviation. As the value gets smaller, the error in 5
mm rounding increases. A few mm does not matter much for stature but it
matters a lot for hand dimensions.



>Especially if you need to use national data to produce
>international amalgamations � say when you are designing door handles
>for an international airport.

Yes. Things can get pretty bizarre. Particularly if you think about
different nationalities, age groups, wheelchair users etc. Then you start
factoring in the fact that each groups will be present in different numbers,
and that some groups may be regarded as more in need of design effort than
others.

Then you may need to consider future sales potential to user groups that are
not currently planned. For example, I worked on a missile system that is
manually reloaded. We were not certain that our eastern customer would be
able to give us the big contract to buy the system because the missiles were
designed for our soldiers and appeared too heavy for their soldiers. You can
imagine that this was no small detail.


>I was surprised to see that the standard variations were remarkably
>variable � the French females had an SD of 2.5 but the Sri Lankan
>females had an SD of 5.0 � double!

The reference quotes 4 mm and 5 mm respectively, so perhaps that sounds more
consistent. Remember, you started with data that contains rounding and you
can't reverse the calculation. In any case, I am not surprised by variations
in data from different researchers.



> �Which size 12 are you?

That is interesting and entertaining. Thanks.

Reply via email to