A while back I read an article (Turner 2010; available here) that explained the rapid growth in the number of international Non-Governmental Organizations (INGO‘s) as a function of of demographics and the structure of economic systems in developed countries.
Essentially, the authors argue that an over-supply of cultural and economic elites led to the creation of an ‘auxiliary vehicle’ for maintaining wealth and status (such as participation in social organizations). In other words, rising inequality that produced an ever-increasing number of highly-educated geographically mobile young people without productive employment opportunities and thus without a way of asserting–socially or economically–their elite status. Competition for work and status drove what Collins termed a ‘credential crisis’–too much education and not enough employment (a powerful signifier of status). According to the authors, the causes of an over-supply of elites include:
“a rise in the numbers of individuals with advanced degrees, a baby boom, cultural trends
(such as targets to employ more females in top positions), economic
displacement of middle class workers due to globalization and technological change.” (Turner 2010)
The average age of leaders of any large organizations could tell us something. When competition is very high (i.e. in times of overabundance of educated elites) “advancement to the rank of the leaders is slow, and leader age should increase.” The credential crisis, for example, manifests itself in law school applications: the number of students entering law school in the past couple of decades has far outpaced the number of new jobs, a trend reflected to a lesser (or maybe just more uncertain) extent with MBA students.
The reason I am writing about all of this here, in particular, is one of the only two graphics that accompany the Turner text.
Labeled ‘Cohort dynamics and increase in a)INGO and b)age of IGO leaders,’ it demonstrates what I think is a very powerful tool in data visualization. This tool is what I call implied correlation.
Interestingly, at least to someone used to seeing regression tables and the like, there is no other tabulation than a brief summary of the statistics of one group (MBA graduates) within the cohort. No other statistical analysis, no regression models, residuals, goodness of fit, etc. Just two graphs. What I think is so interesting here is that I was convinced. I would have liked more information, but at least intuitively (and I’ll get back to this) it seemed right. And this intuition is whats important: a good data visualization is able to encode information (by displaying it visually) in order to reveal something deeper about your data that tabulation or textual explanation just cannot efficiently provide. As I have argued before in my post on the brain as a pattern-recognition machine, data visualizations tap into the pattern-recognition powers of the human mind–it is what we are good at evolutionarily. I call Turner’s example implied correlation because you don’t need to provide any data analysis because it employs our intuitive grasp of visual information to recognize patterns.
But it also opens us up to manipulation. Without even summary statistics, how can I be sure that this is an accurate representation of the data?
At the very least, I need to trust that their numbers aren’t fudged, but beyond that how can I be sure that what I am being told is true? More importantly, how can I tell if there are other lurking variables, other factors that may actually tell a different story (like, for example, reverse causality, where the INGO growth rate effects cohort size. Obviously there is almost no way this could be true, but you get my point)? The thing is, you can’t. And this is where the danger of compressing information comes in.
Whatever the reason for INGO growth actually is, this serves as an example as to how effective implied correlation can be. The most basic of graphs can tell an impressive story and provide a large amount of pattern information. And I think implied correlation is something we use all the time in different areas, but when we see it rhetorically, it just kind of looks like fallacious and invalid logic (e.g. “We are increasingly reliant on foreign oil. Iraq has oil. Thus we invaded Iraq because of oil.”) That is not to say that rigorous quantitative analysis of social science events cannot be methodologically both sketchy and just a little lazy. Rather, we need to be aware of what goes into building the infographics and charts that are now so ubiquitous we may be a little careless in how readily we accept their veracity.
Collins R. 1979. The Credential Society: An Historical Sociology of Education and Stratification. Academic Press, New York
Goldstone J. 1991. Revolution and rebellion in the early modern world. University of California Press, Berkeley
Turchin P. 2003. Historical dynamics: Why states rise and fall. Princeton University Press, Princeton
Turner, Edward. 2010. “Why Has the Number of International Non-Governmental Organizations Exploded since 1960?” Cliodynamics, 1(1). http://escholarship.org/uc/item/97p470sx