Segregation-&-norms-2N5

view/download model file: Segregation-&-norms-2N5.nlogo

WHAT IS IT?

Here is a simple model of segregation: each cell in a grid represents a house occupied by either red or green individual in equal numbers. Because this is like a checkerboard grid, each person is next to up to eight others. You can try this as an exercise with checkers on a checkerboard, or use pennies and dimes on a grid. Be sure that about 10% to 20% of the houses are empty, and that you have equal numbers of each of two types (pennies and dimes, or red and green checkers etc.)

There is one simple rule each individual follows. At any point in time, an individual might be satisfied or unsatisfied, depending on how many neighbors are of the same type. Try this first with a rule that is very tolerant of difference: individuals are satisfied if at least 30% of their neighbors are like themselves, and unsatisfied if less than 30% of their neighbors are like themselves. If they are satisfied they stay where they are, if not satisfied they move, e.g. each individual implements the following simple rules:

IF < 30% neighbors are the same as you, THEN not satisfied, move to an empty cell at random

Each individual repeats the above rule repeatedly, iteratively. So if it is not satisfied after randomly moving once, it will move again.

The flip side of this rule is: IF % similar neighbors (neighbors of same color as me) >= 30% THEN condition = satisfied. When satisfied, the individual does not move.

Now imagine many individuals repeating this rule in a large grid such as a checkerboard with a large number of red and green circles. Will segregation occur? What patterns do you predict?

Let’s change the rules slightly so that each individual will tolerate 50% being different. So, even if half of the neighbors are of a color different than itself, a person will be satisfied and not move. Will segregation occur? Try this on your grid with checkers, or pennies and dimes. What patterns did you find?

My norms version of this model:

The segregation model is was created by Schelling. I added a version in which individuals do not move but rather change color according to a collective threshold. That threshold is the same as that for segregation, e.g., if < 30% of neighbors are the same as you, THEN not satisfied and CHANGE COLOR (instead of moving). The results are very similar. This tells me that it is not the moving but rather the collective threshold that is the essential component. My change color version is like social norms. It is essentially a norms threshold, e.g. IF >=70% are the other color, then change to fit the norms.

HOW TO USE IT

Click SETUP, then GO.

The %-same-wanted slider allows you to choose the percentage of same-color neighbors that each individual tolerates. Start with 30%.

NETLOGO COMMANDS

The basic procedures for individuals moving in netlogo are as follows:

The program sets up the model by placing individuals on spaces, assigning each randomly to a space. Half the individuals are red and the other half are blue.

The next step is to execute individual rules as in our exercise above, except using Netlogo’s language. When the individuals execute these commands, the individuals move until they are satisfied.

to move-unsatisfied-individuals
ask individuals with [ not satisfied? ] [ find-new-spot ]

This is just like saying: IF not satisfied THEN move to find a new spot. Then the program defines the find-new spot action:

to find-new-spot
rt random-float 360
(right turn, at an angle randomly selected from 1 to 360 degrees).

fd random-float 10

(This means individuals move to some random space up to 10 spaces away).
if any? other individuals-here [ find-new-spot ]

(individuals keep trying until they find an unoccupied space).

Remember we had set the rule for each individual’s satisfactory condition as having >30% similar neighbors. This netlogo program uses the following command to set whether or not a individual is satisfied:

set satisfied? same-nearby >= ( %-same-wanted * total-nearby / 100 )

This is Netlogo way to say: IF % same neighbors >= 30% THEN set condition = satisfied.

There are other Netlogo commands to make the program work smoothly, and to show us the proportion of individuals satisfied and so on. You can study these details at your leisure by experimenting with the Netlogo programming (see note). Here we want to focus on the essential underlying logic. In terms of the underlying logical grammar ‘IF condition THEN action’, all these netlogo commands essentially amounts to the rule we wish, simply:

IF < 30% of your neighbors are the same as yourself, THEN move.

The individuals can only detect the condition surrounding it. The individuals can’t look in some distant part of the neighborhood. Rather all the individual can do is move, then detect its immediate surrounding eight neighbors. For this reason the program has each individual move to an empty cell at random before checking its new condition. individuals do not know anything about other spaces in the grid, nor the conditions of other individuals. This is like a trial and error approach with feedback. The feedback is a simple comparison of the wanted condition with the actual current local condition.

NETLOGO FEATURES

N-OF and SPROUT are used to create turtles while ensuring no patch has more than one turtle on it.

When a turtle moves, MOVE-TO is used to move the turtle to the center of the patch it eventually finds.

CREDITS AND REFERENCES

This is based on Thomas Schellings famous model of segregation. Schelling originally developed the model without a computer, in a manner much like the pennies and dimes exercise described above, and as we could try in a classroom exercise.

I added a version in which individuals do not move but rather change color according to a collective threshold. That threshold is the same as that for segregation, e.g., if < 30% of neighbors are the same as you, THEN not satisfied and CHANGE COLOR (instead of moving). The results are very similar. This tells me that it is not the moving but rather the collective threshold that is the essential component. My change color version is like social norms. It is essentially a norms threshold, e.g. IF >=70% are the other color, then change to fit the norms. I call this the norm-driven segregation.

I discuss the segregation model at length in my book, Keane, C. 2013. Modeling behavior in complex public health systems. Springer. There I discuss some recent empirical tests of this model.

Schelling, T. (1978). Micromotives and Macrobehavior. New York: Norton.
See also a recent Atlantic article: Rauch, J. (2002). Seeing Around Corners; The Atlantic Monthly; April 2002;Volume 289, No. 4; 35-48. http://www.theatlantic.com/issues/2002/04/rauch.htm

To refer to this Netlogo model in academic publications, please use: Wilensky, U. (1997). NetLogo Segregation model. http://ccl.northwestern.edu/netlogo/models/Segregation. Center for Connected Learning and Computer-Based Modeling, Northwestern University, Evanston, IL.