by Judith Curry
This past week, legislation was introduced in the U.S. Congress to cut NSF funding of political “science” . This amendment . . . brings to light an important debate on the more fundamental nature and scope of science itself.
There are two interesting posts at Real Clear Science (h/t Kip Hanson). The first article by Thomas Hartfield is entitled ‘NSF Should Stop Funding Social ‘Science‘. Some excerpts:
This past week, legislation was introduced in the U.S. Congress to cut NSF funding of political “science,” to the dismay of some. Politicians cherry-picking projects to be allowed and funded is certainly not a pleasant notion. That said, this amendment does not target any specific study or research; it actually brings to light an important debate on the more fundamental nature and scope of science itself.
“Urban food system alternatives,” “how power affects empathy” and “outlook on life and political ideology” are all recent projects to gain NSF funding.
This is not to discount the value of work in this field or others like it. The study of life and society as well as art, literature, history and other things unquantifiable certainly has value and has a place in our consciousness. I simply contend that it does not fall under the jurisdiction of science. It does not and cannot follow the rigorous requirements of reproducibility, testability and objective truth required of science.
Why should the National SCIENCE Foundation fund non-science?
This was followed by an article by Alex B. Berezow & Tom Hartsfield who ask the question “What Separates Science from Non-Science?”
The term “science” carries a centuries-long aura of legitimacy and respectability. But not every field of research can rightly call itself scientific.Traditionally, fields such as biology, chemistry, physics and their spinoffs constitute the “hard sciences” while social sciences are called the “soft sciences.” A very good reason exists for this distinction, and it has nothing to do with how difficult, useful or interesting the field is. Instead, it has to do with how scientifically rigorous its research methods are.
What do we mean by scientifically rigorous? Let’s start by discussing what we don’t mean.
Using statistics doesn’t make a field scientifically rigorous. Baseball players and gamblers use statistics everyday. They are not scientists. Even using extremely complicated math and statistics doesn’t make a field scientific.
The mathematically intensive field of economics is largely preoccupied with determining correlation and causation. In order to do so, economists employ a statistical technique, multiple regression analysis, which is every bit as complicated as it sounds. But, as the authors of Freakonomicswrite, “[R]egression analysis is more art than science.”
So, if mind-bending statistical analysis doesn’t make a field scientifically rigorous, what does? Five concepts characterize scientifically rigorous studies:
- Clearly defined terminology.
- Quantifiability.
- Highly controlled conditions.
- Reproducibility.
- Predictability and testability.
Admittedly, this is a tough list. But, it’s supposed to be. The standard for rigorous science should be very high. Even some fields widely accepted as scientifically rigorous don’t always measure up. Particle physics – most notably, string theory – sometimes makes predictions that are not testable with modern technology. Epidemiology often cannot perform controlled experiments, both for reasons of ethics and practicality. Epidemiologists can’t lock 20,000 people in a room for 20 years to determine if force-feeding them hot dogs will cause cancer. Instead, they rely on observational studies.
But, clearly, some social science fields hardly meet any of the above criteria.
So, returning to the question posed in the title of this article, “What separates science from non-science?” It’s hard to say. There isn’t a crystal clear dividing line between the two. But, what can be definitively said is this: A scientifically rigorous study will meet all or most of the above requirements, and a less rigorous study will meet few if any of those requirements.
As useful and interesting as the social sciences are, they usually fall into the latter category.
JC comments: If Berezow and Hartsfield are correct in their criteria for what is science vs what is not, it seems that all of the geosciences and environmental science is not really science, not to mention cosmology and string theory. We’ve had this discussion before at Climate Etc., see especially Mike Zajko’s essay on Scientific Method (see also the comments). Another implication is that the methods of science cannot be applied to complex systems. IMO the issue of what separates science from non-science is a much more complex issue than Berezow and Hartsfield articulate.
So where does that leave the social sciences in terms of NSF funding? Information on the NSF Directorate for Social, Behavioral and Economic Sciences is found [here]. The mission of the SBE Directorate is:
There is consensus among economists and policy researchers that public investments in science and engineering yield very high annual rates of return to society. Furthermore, the activities supported by the National Science Foundation (NSF) – fundamental research and education based at academic institutions – are generally viewed as among the most productive of all Federal investments.
The Directorate for Social, Behavioral and Economic Sciences at NSF (SBE) supports the research that underlies such findings, as well as other research that builds fundamental knowledge of human behavior, interaction, and social and economic systems, organizations and institutions. It does this through its Division of Behavioral and Cognitive Sciences (BCS), Division of Social and Economic Sciences (SES), and SBE Office of Multidisciplinary Activities (SMA).
To improve understanding of science and engineering, SBE provides tools for tracking the human and institutional resources vital to building the nation’s science and engineering infrastructure. It does this through its National Center for Science and Engineering Statistics (NCSES), which is the nation’s primary source of data on the science and engineering enterprise.