Traditionally, anthropometric criteria are set for two different measured populations, which together are taken to represent most of the intended Users of a design. For example the range '5th percentile UK Female to 95th percentile UK Male' is taken to represent 95% of UK Adults.
However most designs are used by a less simple mixture of measured populations, in which the gender split is uneven, and the age and nationality mix is different from any one national or measured population as a whole. This can cause significant errors in the estimations, and prevents design decisions being based on simple percentage exclusions among the target user group.
The mathematically correct technique is to calculate the percentile separately for each constituent population, add up the included proportions, and then progessively adjust the dimension until the specified overall percentile value is reached. This technique always produces the correct result, even with diverse mixtures of measured populations. As an extreme example, 5% of 5-year-old children in an otherwise adult user group will be handled correctly by this method. Desktop computers have now become so fast that this calculation can be performed instantly, and in PeopleSize Professional all calculations act on composite populations in this way.
By naming a set of constituent populations in a meaningful way, you can clarify the anthropometric decisions in a project. For example if your design is a mobile phone, and you can obtain market research data about customers of mobile phones, you can create a User Group and call it 'Mobile Phone Users'. Then you can correctly discuss, for example, a second percentile hand grip diameter, or state that 97% of mobile phone users can grip both sides of the phone.
This method is more accurate than using a composite distribution.
A real measured population is handled as a one-element composite population.
Because of the benefits of this technique, PeopleSize Professional only addresses one population, and you specify the makeup of your composite population in a User Definition dialog. Some composite populations are included ready-made with the software.
One important effect of this approach is that (for example) 5th to 95th percentile no longer accommodates 95%, but only 90%. This is because instead of two different populations which are mutually exclusive at the extremes (male and female), you are now specifying for a single population from which 5% is being excluded at both ends.