Measuring psychological variables more effectively by direct comparisons?

[Akusius]

Hello!

Just would like to share an idea about measuring variables in the social sciences (especially in psychology) more efficiently.

E.g. let's try to measure the extroversion trait in a group (or in the population).

Today our method is to construct a survey, which tries to measure the various aspects of this characteristic numerically (e.g. with the help of 1-5 scales). Then we apply a numerical model to the results, and get a final value for the extroversion quality of each individual.

The problem with this approach is, that the end result is too closely attached to the actual survey and numerical model, and it cannot be generalized easily. Furthermore it is already questionable, whether the usage of numbers for psychological qualities can be justified at all, since it seems not to have much sense of speaking about e.g. twice as large extroversion, etc.

So we should obviously find another measuring method, which does ensure accuracy but without the necessity of using numbers or numerical models for this purpose.

Would not a sorting algorithm based on multiple direct comparisons be a more effective way of measuring these variables (similarly to that used e.g. in the object-oriented programming languages for sorting the objects according to a certain aspect)?

E.g. for measuring the extroversion quality only one single comparison function would be needed, which would provide a <, = or > result (consistently) for comparing two individuals based on their extroversion.

Performing this comparison multiple times we would get a perfectly defined scale at the end, and so we could measure this trait accurately without the use of any numbers.

Further advantages of this method could be (described in a more elaborated manner on this page):

- the comparison function is much simpler and more elementary, than the number scales used in the questionnaires;

- the brain is much more consistent in providing such types of comparisons, than assigning numbers in a scale;

- the "brutality" of the comparison function enforces the formation of conscious judgments;

- the comparison function itself is already verified technically by the sorting algorithm;

- this measuring method could facilitate considerably the exchange of the scientific results between various groups and laboratories (because the comparison functions on their own could already be exchanged without any theories);

- with the help of well-functioning comparison methods reference lists could be constructed, which then could be used to measure other items;

- a well-behaving comparison method could also be used in reverse mode, i.e. to define new qualities and variables.

Would this idea make any sense, or there were perhaps already attempts in that direction previously?

Thank you!