Some Tips for Further Exploration and Drawing Conclusions
Models are a way to discover new ways by which one might explain things,
and to create new questions. They are NOT representations of the "real world",
nor are they intended to be.
Humans are subject to a wide array of influences, and may, even were all
of the somehow taken into account, display some remaining unpredictability.
Correspondingly, clustering/segrgation seen in any given situation may reflect,
among other things, racism, economic constraints, and a host of other possible
factors not included in the models here. It is not in general possible to
build models that correspond to the entirety of the real world. Nor would
it be helpful. The point of a model is not to "explain" any particular situation
but rather to allow one to discover and test possible explanations
that one might be ignoring in the complexity of the real world.
In the present case, one is not entitled to conclude from the model
that a preference for being surrounded by similar people is the single factor
that accounts for segregation/clustering in all cases. Nor is one entitled
to conclude that other factors are irrelevant or even unlikely, either in
individual cases or generally. One can conclude that a preference
for being surrounded by similar people would suffice to account for segregation/clustering
in a lots of different cases and, perhaps more importantly, that it could
be a contributing factor of the sort that helps to explain why similar behavior
occurs a wide variety of circumstances. And that a particular sort of change
rather than another sort is more likely to be effective in a wide array
of circumstances in bringing about more integrated communities (assuming
one had that as an objective).
Careful and deliberate collection of observations is, in many cases, as
important in drawing conclusions from models as it is in drawing conclusions
about the world.
Single observations are rarely adequate to draw conclusions about real
world phenomena, and the same holds for models that, like the present ones,
incorporate random factors. Since both the initial population distributions
and the movements of unhappy people have a random element, the observed behavior
is different in at least small ways every time the model is run. One needs,
at a minimum, to have run the model enough times to develop a feeling for
how much variation there is from trial to trial in order to be able decide
what observations are meaningful and what ones are not. In the present case,
for example, the trial to trial variation in percentage of similar numbers
is at least a percentage point, so one would not want to draw conclusions
that rely on differences between situations in that range of values.
See Do Your Own Research
for some additional general ideas about observations and conclusions that
hold both for research on models and research on the world.
Good models are at least as important for what they make one wonder about
that one might not have wondered about as they are for what they allow one
In this case, if you find yourself generally frustrated by the model because
it isn't "reality", go back and re-read point 1 above. If, on the other hand,
you're wondering whether the phenomena observed in the simple model depend
on particular characteristics of this particular version of the model and
what would happen if you could vary some of them in particular ways then you're
well on the way to becoming a productive modeller yourself. The Advanced
Model is set up to allow you to vary some additional features. Of course,
it too has some things you can't vary. If that bothers you in general, go
back and re-read point 1 again; there is no way that a model can incorporate
variation in all possible features and its not clear what one would like for
in it if it could. But if you have a reason to be interested in what would
happen if you altered something unalterable in the Advanced Model, you're
ready to try some model-building yourself. You can download the NetLog modelling
environment and you're off and running. Let us know if you find something
interesting and we'd be delighted to include it here.
Modelling, like science in general, is fundamentally a social activity
One can make observations and try and interpret them oneself, but its much
more effective to pool observations with other people, and see what kinds
of observations make sense over more observations looked at from more perspectives.
Let us and others know what you're seeing and thinking in the Segregation/Integration
on-line forum area.