by Ian Angus
When I wrote my article about the use of “per capita” statistics to defend populationist ideas, I wasn’t familiar with the work of U.S. environmental sociologist Allan Schnaiberg. If I had been, I could have saved myself a lot of time and effort, because he confronted the same question and came to similar conclusions, thirty years before I did. 
Schnaiberg, who died last year, developed what he called the treadmill of production framework for understanding the rapid expansion of environmental degradation in the United States after World War II. He first presented that framework in detail in 1980, in The Environment: From Surplus to Scarcity.
A full chapter of that book was devoted to examining and rejecting populationism, a term he seems to have been the first to use. He defined it as “a social ideology that attributes social ills to the number of humans.”
Discussing the “populations problems model,” Schnaiberg pointed out that it is remarkably easy to make any social problem look like a population problem:
Consider any event that society has defined as a problem. Suppose that N number of events occur in a given year, in a society with a population sized P. With no other information, we can immediately construct the following analytic model:
Where R is the problem rate. From this purely arithmetic operation, devoid of any substantive meaning, we can see the problem as a “population problem.”
The formula seems to show that in order to reduce problem N, we must reduce population. That’s misleading, Schnaiberg explains, because R isn’t independently measured — it is calculated using the other two terms.
R is computed by taking the relevant population total and dividing it into N, the total frequency of social ills. … R is typically not a rate but a computed ratio between two social accounts: a population account and a social ills account. 
So if we know total pollution (N) and Population (P), then dividing N by P produces a per capita Rate (R) — which leads to the tautological claim that total pollution equals population times per capita pollution. Many populationist arguments are no more sophisticated — and no less vacuous — than that.
A book co-written by Schnaiberg summarized this argument again in 2009.
Some analysts claimed that it was the growth in population that required a production increase. As a sometime demographer, it was clear to Schnaiberg that, while there had been a baby boom during this period, the rise in energy and material use vastly outstripped population growth.
Demographic explanations, he argued, were often appealing mainly because we had detailed records of population characteristics. Thus, we could trace the rise in population along with the rise in some forms of pollution .…
Ignoring the methodological dictum that we need to distinguish causation from mere correlation, it became easy for many early analysts to take an environmental statistic, divide it by the level of population, and come up with a “per capita environmental impact” assessment. This ratio was treated as if it were an analytic rate of how much each individual actually added to environmental degradation. 
That approach makes no more sense than dividing the number of Justin Bieber recordings by the total population to find the number of recordings per person — and then concluding that the best way to prevent him from singing is by reducing the global birth rate.
So the next time someone quotes a “per capita” statistic to support the “too many babies” argument — or any other political position — check whether the number is a rate (which is directly measured) or a ratio (which is calculated from other measurements) — the distinction is critical.
Notes Many thanks to Byron Clark, who drew my attention to Schnaiberg’s work in his recent talk on population policy in New Zealand.  Allan Schnaiberg. The Environment: From Surplus to Scarcity. Oxford University Press, 1980, p. 69-70  Kenneth A. Gould, David N. Pellow, Allan Schaiberg. The Treadmill of Production. Paradigm Publishers, 2008. I have added several paragraph breaks for on-screen readability.