A Social Decison model for the stem cell research issue

This page shows the results of running decision models composed from opinions collected from Mechanical Turk and Survey Monkey.

Questions posed to surveyed individuals and their numeric values:

The data was fed into the following model to determine the best decision based on submitted opinions

Rectangle "nodes" represent the decision and its options, oval nodes represent "variables" that have a probabilistic value. Diamond nodes are called "utility" nodes and represent the value (positive=good, negative=bad) of an outcome. The supplied values are "propagated" through the model to determine the overall utility based on which decision option is selected.

The following image shows the probability distribution of the beliefs collected from 293 people. Each question is represented by a clump of bars. The label under the bars corresponds to the Key in the first figure.

If we average all the opinions, we get the following utilities for each decision option:

Given these results, the best option would be to fund both embryonic and adult stem cell research. However, if we group the individuals into groups with more similar beliefs first (using clustering techniques), we see that a significant portion of the population surveyed would actually find that option to be the worst case scenario. The following table shows the utilities of the different groups (or clusters). A negative utility value means that the option is undesirable. The first group supports funding both kinds of stem cell research and has a high personal stake. The second group would prefer stem cell research to be funded by private entities. The third group slightly favors funding only adult research and the fourth favors both, but with a lower personal stake than the first group.

Social decision models are intended to help decision and policy-makers make the most appropriate decision given the direct input of their constituency. This work is described in a paper that will be presented at IEEE Social Computing conference, in the Social Intelligence in Applied Gaming workshop.

K. Greene, J. Kniss, G. Luger and C. Stern. "Satisficing the masses: Applying game theory to large-scale, democratic decision problems"