DX is really hard to evaluate because it basically has two main aspects: manual dexterity and bodily dexterity. In the real world, few people have high scores in both. Magicians, for instance, have a high manual DX but should not be especially good in football or gymnastics. On the other hand, athletes have high bodily DX, but should not have steadier fingers than the average guy.
Until we can come up with a good way of evaluating DX (maybe by using Basic Move and other derivates of DX?), I can only treat it through the game designer's lens, i.e. use the same score range as for IQ. This is because IQ and DX both have the same cost, and are both widely used for buying skills. According to IQ Reality Check, 99% of people are in between 8 to 12, so we'll assume it's the same with DX.
For very good characters, you can buy (or create!) Advantages or Disadvantages that modify manual or bodily DX. You can also add a few Talents to the mix. This should model pretty realistic characters.
In GURPS, "Dexterity" covers overall body agility, reflexes, flexibility, balance, and manual dexterity. These can all be tested, as outlined below:
This test is known as the 3 Cone Drill and is used to assess the agility of players of American Football.
Required Resources
To undertake this test you will require
The three cones are set five yards apart on a straight line.
The coach starts the stop watch on giving the 'Go' command and stops the watch when the athlete touches the middle cone. The best of two trials in each starting direction, right and left, are recorded and the best score in each direction is used for scoring.
Top times for recruits from a region's high schools are about 7.0 seconds Top times for players in the NFL draft are about 6.7 seconds Minimal data appears to be available on the distribution of times across the set of all players, however. One completely ad-hoc guess at a time-to-DX correspondence is:
In this test, a person must catch a metre ruler in their hand, the quicker they catch it, the quicker their reaction speed.
How to conduct the test
The ruler is held by the assistant between the outstretched index finger and thumb of the athlete's dominant hand, so that the top of the athlete's thumb is level with the zero cm line on the ruler. The assistant instructs the athlete to catch the ruler as soon as possible after it has been released.
The assistant is to record distance between the bottom of the ruler and the top of the athlete's thumb where the ruler has been caught.
Incidentally, the abilities of the off-hand relative to the first can be measured. A truly ambidextrous person will have the same ratings for their off-hand as for their dominant hand.
This is the "sit and reach" test. The person sits on the floor with their legs outstretched. They reach forward their arms, bending forward, trying to touch and then pass their toes. The coach measures how far before or after their toes they pass with the tips of their fingers. The test is repeated three times. A better performance will be achieved if the person jogs or walks for five to ten minutes before the test, so most tests are conducted without warmup. According to this site the distribution of flexibility in the population is:
The following is the "Standing Stork Test."
On command from the coach:
Repeat the test for the other leg, and take the average of the two times. A guess at DX values, based on this site for measuring athletes' development:
This section remains to do. Most manual dexterity tests require specialised kits.
By testing these five areas, people can get an idea of their own DX score in GURPS terms. Many people will say, for example, "oh, I play computer games, so I have good manual dexterity, therefore good DX," or "I play sports," etc. But DX covers such a broad area, which is why we present these five tests here. A person of a certain DX level will have about that performance in all five tests.
Note, however, that most available tests have limited information available on the distribution of their scores, and what is available is most often for trained athletes. Set the "average" values for such tests (e.g., these at the desired average for trained athletes in your game.