Tikalon Header

A Hierarchy of the Sciences

March 28, 2012

There's a feeling among some physicists that all other sciences are derivative of physics. Chemistry is explained by quantum mechanics, biology is just chemistry, and so forth. The mathematicians, of course, feel that physics is a subset of mathematics, and the string theorists have bought into that theory. Although I'm a proud physicist (who defected to materials science, so maybe not as pure as the rest), I think a lot of these arguments are similar to saying that the literary classics are derived from the dictionary.

August Comte (1798–1857), who is probably the first philosopher of science, was the first to publish the idea of a hierarchy of the sciences. Perhaps influenced by the ancient concept that the higher things are the most pure, Comte thought that astronomy was at the highest level, and things like mechanics were a step below. Things were lower when you got down-and-dirty at Earth's surface and the abode of biology. Comte further believed that sociology was the least developed science.

Camille Flammarion, L'Atmosphere: Météorologie Populaire (Paris, 1888), p. 163.The peregrine philosopher discovering the secrets of the universe by peeking through the sphere of the fixed stars.

Camille Flammarion, L'Atmosphere: Météorologie Populaire (Paris, 1888), p. 163.

(Via Wikimedia Commons).

Modern scientists tend to think of this hierarchy more on a hardness scale. Physics is a "hard" science, and the social sciences are the "soft" sciences. The more advanced thinkers tend to follow Karl Popper's idea that falsifiability is the essence of science. Hard sciences, then, would be those whose theories are quite directly falsifiable by experiment. The soft sciences are those with all the hand-waving.

An interesting 2010 paper by Daniele Fanelli of the Institute for the Study of Science and Technology and Innovation of the University of Edinburgh (Edinburgh, UK), lists the following qualities of a hard science:[1]
• A high level of agreement on a single set of theories and methodologies.
• The rigor with which data is related to theory.
• The extent to which the choice of problems and decisions made in solving problems are based upon cognitive as opposed to non-cognitive criteria.
• The level of consensus on the significance of new knowledge.
• The continuing relevance of 'old' knowledge.
• Its explanatory success.
One telling observation is the following: Data and theories speak more for themselves in the hard sciences; whereas, things such as the prestige of a scientist within his discipline and other sociological factors are more important in the soft sciences.

Fanelli's paper, entitled "'Positive' Results Increase Down the Hierarchy of the Sciences," examined 2434 published papers in many disciplines to examine the hypothesis that there are more positively supported hypotheses in the soft sciences; that is, the soft sciences are less critical of their hypotheses. Space science had the lowest percentage of positive results, 70.2%. Psychology/psychiatry had the highest percentage of positive results, 91.5%.[1]

A statistical analysis revealed that a positive result was five times more likely for a paper in the areas of psychology/psychiatry and economics/business than in space science. A paper in the social sciences was 2.3 times more likely to convey a positive result than a physical science paper. Behavioral and "people studies" were 3.4 times more likely to have a positive result than physical and chemical studies on non-biological material. Not surprisingly, the biological sciences ranked between these extremes. These statistical "odds" were calculated after adjustment for papers that tested multiple hypotheses.[1]

Figure captionPositive results by discipline.

The error bars are the 95% confidence interval.

(Fig. 1 of Ref. 1, Creative Commons Attribution Licensed).

One interesting statistic in this study is that publications in materials science have nearly as many positive reports as papers in psychology/psychiatry. This would score materials science as a soft science, but my own observations of the field point to a different effect. In materials science, and in other applied science and technology fields, results are often unexpected, and the "hypothesis" is generated after the fact. We all know that the "scientific method," as taught to schoolchildren, is not really how things happen.

One possible source of error in this study, as identified by its author, is the tendency of scientists to keep negative findings unpublished; and for journal editors to reject such papers. Also, results that rely on statistics, as are commonly generated for soft science papers, are more publishable.

Although the study places the social sciences at the bottom of the hierarchy, the results are encouraging in that they show that these sciences are not purely subjective. They just differ from the hard sciences by a matter of degree.

Reference:

  1. Daniele Fanelli, "'Positive' Results Increase Down the Hierarchy of the Sciences," PLoS ONE, vol. 5, no. 4 (April 7, 2010), Document No. 10068.