Artificial Qualitative Data: From Quants to Quals?

We often go through processes of producing quantitative data from qualitative data - many social science methods essentially aim to do this through coding of data, frequency analysis, etc. But we never think of going from 'quants' to 'quals'. One of the reasons why the question is never raised is that somehow 'quants' are considered more 'solid' and trustworthy than quals: if you can fit an equation to it, it must be right! But in reality, it is the quals which are the most fascinating.. the "any other comments" fields which capture the essence of how people are feeling. But we struggle to abstract this into meaningful and defensible conclusions.

One of the most significant aspects of agent-based modelling approaches to 'artificial psychology' is the prospect of producing a kind of rich 'qualitative data' from quantitative data. How would that work? Well, what is needed is an approach to categorising the data in the first place. I had fantastic meeting with Leydesdorff in Amsterdam yesterday, and he showed me how he looks at quants specifically to identify social dynamics. That stage is simply a statistical stage. But the more interesting thing is to be able to understand the intentional forces behind the dynamics. That, for Leydesdorff, has to do with 'anticipation', or (in his language) a 'hyper-incursion'. This is where the state of the system in the present is dependent on the state of the system in the future. He argues that hyper-incursion amounts to 'social structure', and he writes an equation to express this:

This hyper-incursive dynamic can be compared to what Leydesdorff calls an 'incursive' dynamic, where each state is dependent on the previous state (this is a bit easier to understand that the hyper-incursive dynamic). In effect, this amounts to 'agency'. It is expressed:

Structure and agency operates within a continually emerging information domain, where each state of information is dependent on the previous state. This dynamic Leydesdorff calls 'recursion' and he expresses it as: 

I'm still thinking about this. But I am also thinking about Kadri's artificial psychology and his approach to hysteresis (see http://www.emeraldinsight.com/journals.htm?issn=0368-492x&volume=40&issue=7&articleid=1949246&show=abstract) How do Leydesdorff's equations relate to Kadri's use of the Volterra series?

Basically, using Leydesdorff's approach to identifying social dynamics, it might be possible to reproduce a dynamic using either a formula (which is what Leydesdorff's formulae do) or using agents. And if the dynamic which produced the data can itself be reproduced (and maybe even the data reproduced), then I think some fascinating opportunities present themselves for using data as a means producing defensible judgements and effective coordination, because what will have been identified is the dynamic mechanism which sits behind the emergence of data. This, in fact, is an approach very similar to Realistic Evaluation.