Modelling Complex Behaviours (Week 3)
'A public opinion poll is no substitute for thought.'- Warren Buffet
While Steve Cork considered the complexity of uncertain futures, Martin Richardson showed that uncertainty may derive from complex behaviours. This week's tool is thus an understanding of the use and limitations of behavioural models. Predicting behaviour is a good way to get what we want, whether that be a gambling win, business success, or electoral victory. I would have liked to ask Richardson just what he meant when he commented that we were 'always modelling' and why we do so. Perhaps it is because modelling behaviour offers high rewards but involves complex calculations of uncertain human motivations?
Politicians are of course obsessed with modelling the voting behaviour and preferences of their constituents. They wish to understand their concerns and, more transparently, to garner their votes. The most obvious political modelling occurs through the complex business of political polling. The tutorial trading game offers several insights into the political practice of polling. It showed that it is difficult to perceive a pattern in complex behaviours without understanding the intention of the actors. Perhaps focus groups were developed in order to explain the intentions manifest in voters' polling behaviours.
Politicians are of course obsessed with modelling the voting behaviour and preferences of their constituents. They wish to understand their concerns and, more transparently, to garner their votes. The most obvious political modelling occurs through the complex business of political polling. The tutorial trading game offers several insights into the political practice of polling. It showed that it is difficult to perceive a pattern in complex behaviours without understanding the intention of the actors. Perhaps focus groups were developed in order to explain the intentions manifest in voters' polling behaviours.
The tutorial game also confirmed Richardson's point that a model is only useful when understood within its own parameters. We were operating only as a caricature of a regulated financial market. For example, the government sought to secure only the votes of the trading firms, and did not have to consider the concerns of small business or workers. Political modelling is similarly limited by its parameters such as the nature of the questions asked. I wrote an essay in US Government and Politics which found that presidential polling or even elections do not found a mandate for specific policy stances. Rather, they provide a broad trend of aggregated popular preferences on a particular day, for example towards stronger national security or away from welfare spending.
Importantly, political polls do not take into account 'cross-cutting concerns'. These concerns are individual preferences on a range of issues which are on opposing ends of the liberal-conservative spectrum. They are important because voters who chose a conservative party may only like conservative policies in aggregate, but be more liberal on some points and more conservative on others. Such contradictions often come into play in politics and must be resolved by the political leaders of the day. An immediate example provided by Xuemai Bai comprised the opposing imperatives for Chinese economic growth and land development against basic food security through preserving farmland.
Importantly, political polls do not take into account 'cross-cutting concerns'. These concerns are individual preferences on a range of issues which are on opposing ends of the liberal-conservative spectrum. They are important because voters who chose a conservative party may only like conservative policies in aggregate, but be more liberal on some points and more conservative on others. Such contradictions often come into play in politics and must be resolved by the political leaders of the day. An immediate example provided by Xuemai Bai comprised the opposing imperatives for Chinese economic growth and land development against basic food security through preserving farmland.