First page Back Continue Last page Overview Graphics

Simulations vs. Surveys


Quantifying this type of infrastructure is much harder than its physical counterpart.

One option is the ‘bottom up’ simulation approach – designing numerous agents who interact in a virtual organization –
Another alternative is survey-based analysis – with which I’m sure you are all quite familiar.
The simulation can supply a theoretically unlimited supply of detailed data – but it’s unclear if it accurately reflects the reality of an organization of actual people.
Surveys provide a more convincing picture of the organization – yet there are inherent biases in self-report data and most importantly the data is quite sparse.
Now, I got the pleasure of meeting Tom Sniders yesterday during his workshop where he introduced Siena – a program which seems pretty good at fitting this type of sparse survey data, and I would like to create something similar, but for much from sensors.

Current research on modeling organizations:

U Michigan – agent based models of competition and cooperation
Organizational Evolution Lab
Ross Hammond's corrupt/honest A-society

“inference by abduction”