A Note on Measurement in the Social, Behavioral, and Natural Sciences
Measuring the Changes in Our Culture
“I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind.” - Lord Kelvin
Science, as distinct from other forms of human activity, is characterized by measurement. Measurement may be defined as the act of assigning numbers, and often units, to objects or events.
Measurement in the social sciences dates back to biblical times when simple counts of individual livestock served as the basis for taxation. Through the ages, counts of assets, including soldiers, allowed leaders to assess the likelihood of prevailing in conflicts with other tribes or cities. In the 16th century, the practice of recording births and deaths began and the insurance industry was born. In the 19th century, governments began counting their citizens and the census was born. Today, social scientists count a huge variety of things including economic measures and political phenomena (polls). These count-based data form the basis for evaluating the effectiveness of policy prescriptions in all areas of cultural concern. For example, traffic fatalities in relation to mandated safety improvements in the automobile industry is an example of a social scientific measurement product.
Measurement practices in the natural sciences evolved with the development of units to describe the dimensions of interest. To be useful, such units had to possess three properties: they had to be standard, absolute, and universal. The centimeter as a unit of length meets these criteria, for example. Everyone agrees it is the standard, its value remains constant from measurement to measurement occasion, and it describes every instance of length. The centimeter-gram-second (cgs) system of units provides such measures of distance, mass, and time.
Progress in the natural sciences can be measured by the appearance of new units as new phenomena were discovered and investigated. Electricity provides a good example with the watt, ohm, and ampere as basic units describing electrical phenomena were devised as they were discovered.
In the history of measurement in the social sciences, a singular event occurred around 1830. A Belgian astronomer named Adolphe Quetelet became fascinated with the normal law of error as described by Frederich Gauss in 1800. This curve described the distribution of random events like dice throws, coin tosses, and estimates of the time a star crossed an astronomical observer’s field of view. Standard practice was to calculate the mean of such numbers to get an estimate of a “true” value around which errors were distributed. Quetelet thought that by analyzing the distribution of chest circumferences of Scottish soldiers, for example, he could discern the intent of the Creator! e reasoned that God was trying to make the perfect one but kept missing in a fashion described by the normal curve. In other words, from variability in a set of observations, one could arrive at a measure of the underlying cause. All that remained was to name that cause.
This practice became the foundation of a new discipline known as psychophysics- the study of the relation between physical stimuli of known properties such as light or sound and a human’s perception of that stimulus. The JND, or just noticeable difference, was the unit created to describe the amount by which a physical dimension of a stimulus must be changed in order for a human to detect a difference. The JND was closely followed by the concept of the threshold, the limiting quantity of the physical stimulus an observer could detect in 50% of its presentations. Examples include the faintest light, highest pitch of a sound and smallest diameter of a palpable breast lump in vivo. These psychophysical measures and their units have great practical value in everyday life such as when fitting glasses or hearing aids. Note that measuring a threshold requires several exposures and analysis of the variability of the observer’s behavior.
Quetelet’s discovery reached majestic proportions when Alfred Binet and Francis Galton in the late 1800s sampled the behavior of large numbers of French school children doing tasks and solving problems. Measures of this behavior were distributed in accordance with the normal curve. Galton and Binet divided the scale along which these measures were distributed into 12 units and called it a measure of intelligence! In other words, the children’s performances varied because of their underlying level of intelligence. The variability in behavior served both to define its cause and provide a measure thereof. Any analysis of the actual causes of the behavioral variability is precluded by this action. Intelligence testing became a large industry, supported in part by the need of the military to sort and classify recruits during wartime.
It is useful to contrast this use of variability with the approach taken by Gregor Mendel, a monk who raised peas in a backyard garden. The peas exhibited variability in color (green or yellow) and texture (smooth or wrinkly). By controlling the parenting of successive crops, he was able to isolate the appearance of each characteristic in successive generations and thus gave birth to the science of genetics. Had he followed Galton and Bibnet’s practice, he might have developed an 8-point scale of peaness and thereby provided a measure of a basic Freudian construct.
Psychology, the traditionally designated science of behavior, had as its subject matter phenomena defined essentially by the practices just described. To be sure, behavior was frequently measured but the measures supported inferences about the causes such as learning, motivation, cognition, and so on. Experimental analysis of the causes of behavior and behavior change only began in the first half of the 20th Century and the resultant technologies have just begun to emerge in the last half of that century. They are now beginning to flourish.
The natural science and technology of behavior owe a great debt to early adherence to the principles of measurement characteristic of the natural sciences. Behavior occurs in repeatable instances and these instances are countable. These instances also occur in time so count per unit of time, or frequency, is a measure that is standard, absolute, and universal. Thus, all behavior has a frequency and that measure is not defined by the behavior itself. Instances of behavior can be directly observed or transduced by mechanical or electronic means. Moreover, behavior often leaves products that can be counted independently of their actual occurrence. Arithmetic problems solved correctly are an example. Some behavioral instances leave chemical traces in the body that can be detected by sensitive devices such as the breathalyzer or chemical analysis of blood samples.
The challenge as we seek to deploy behavioral tactics to guide the evolution of our culture is to remember to describe and observe behavior directly as we search for the variables that cause it. We must not assume, as Quetelet did, that when we see variability, we have identified the cause. We must use changes in the variability to guide us in our search for the causes that we can then reframe.
Happy holidays and see you in two weeks!
Hey gents. I'm a free subscriber but nowhere on your site can I find a button to become a paid subscriber, or find out the price. Can you help me out? -- Mike