Difficult moral/ethical decisions occur in the intersection of conflicting moral/ethical principles. These intersections are generally fine-grained and self-similar, with differing decisions hanging on small differences in personal values and the relative force that these conflicting principles have in a particular situation. This "fractal" complexity presents peculiar problems to the measurement of the effectiveness of ethics instruction that are difficult to resolve with standard paper and pencil testing. A computer-based approach is proposed which enables measurement of these conflicting principles and decision variations within layered and situationally adaptive moral/ethical test cases.
Principles are, in their effect, the forces of morality and ethics. Most people understand moral and ethical principles and have an easy time living with them. Most of us try to be truthful and honest, avoid harming others, be loyal to our friends, live up to our commitments, and be supportive of our families. As long as these principles don't conflict (or if we are unaware that they do), our moral/ethical decisions are easy. They only become difficult when principles collide such that we cannot satisfy one without compromising another.
It is a characteristic of complex moral/ethical decisions that they are made in an infinitely finely grained decision space that is sensitive to small variations in the values of the decision maker and the specifics of the situation, and the decision maker's knowledge of those specifics, including the history that led up to the decision. There will almost always be a set of decision possibilities that are wrong from any perspective and a set of decision possibilities that are right from certain perspectives. There is no guarantee that there will be any decision that is right from all or even most perspectives or that any of the "right" answers will not simply be the least wrong answers. The fine grain of moral/ethical decision spaces emerges from the conflict of moral/ethical principles at least two ways.
First, the relative force of each conflicting principle is infinitely variable such that small changes in personal values and/or the situation can make huge differences in ones decision. Consider, for instance, a person who strongly values both family and personal accomplishment. For this person, a choice between remaining at work to complete an assignment on time and going home to spend time with the family would often prove a difficult one. Indeed, either decision would probably entail some level of guilt. Both decisions are, from a certain perspective, right. Neither decision avoids wrong. It may not be possible to find a usable compromise like taking work home to do after the kids are asleep. The force of each principle (accomplishment at work versus support of family) is, moreover, likely to vary considerably from day to day. An event at school may shift the balance to going home on one occasion. An impending work evaluation may shift the balance to staying at work on another. Many other factors, including a sick child, an impending business trip, a good football game on TV, or guilt feelings about previous evenings at work may shift the balance toward home. Examples of the many factors that might shift the balance toward work include immediate deadlines, changes in management, commitments to team, and concerns about rumored impending layoffs.
A second source of fine structure in moral/ethical decision spaces is found in the processes of precedent and exception. It is in the nature of precedent that there is at least one exception hiding inside every precedent. When, for instances, a parent is faced with a sick child and a school day, the parent is caught between the twin principles of the importance of education and going to school and the importance of the child getting better. Depending on the illness and other circumstances, other principles may also come into play, including the importance of not spreading the illness to others and the importance of getting to work. The decision to keep a child home from school when sick is a precedent that is recognized as such by both the sick child and their siblings, but one that may not be as readily accepted by the parent on subsequent occasions. First level exceptions (that might cause the child to be sent to school) might be based on such issues as the mildness of the fever, the absence of a rash, and the child's appetite. Second level exceptions (that might cause the child to be kept home anyway) include excessive worry, the child's history of illness, or the appearance of either being in great pain or very lethargic. Third level exceptions (that might cause the child to be sent to school anyway) include the conviction that the child may be faking it, test dates, and paper due dates. One could easily go on through additional levels of exception, many of them similar in some regards to the exceptions at previous levels.
|Figure 1: Layered Sand subjected to three linear stresses, including a fixed end, a shaped base, and a movable end that is pushed in to emulate tectonic motion. Note the complex "fractal" patterns that occur at the intersection of these three forces. Photograph, of an exhibit at the Rose Planetarium of the American Museum of Natural History in New York, by the author.|
This combination of infinitely fine-grained resolution and self-similarity is generally referred to as being "fractal". Fractals are a special type of chaotic system in which the "chaos" is infinitely complex (i.e. no matter how closely you resolve it, there is always another layer of complexity to be found) and self-similar (i.e. patterns recur within and between levels of resolution). Fractal complexity occurs routinely in nature when "simple" forces are set in opposition to one another, much as they are in the figure 1 "sand painting" of the collision of two tectonic plates. The "collision", in this case, is simply pressure applied to one end of the evenly layered sand painting. The result is a "mountain" in which the horizontally layered sand has been redistributed into complex and sometimes vertical structures. Collisions of moral/ethical principles have similar effects on the complexity of moral/ethical decision spaces.
How, then, do we measure the effectiveness of our ethics instruction? While there will, in the intersection of two or more moral/ethical principles, certainly be obviously wrong decisions, the "right" answer may be highly dependent on details of the situation and the individuals forced choices between otherwise equally important ethical principles. There will be multiple right answers, including both answers that are more right than other answers and otherwise wrong answers that are simply more right than any other answers. In our usual testing practice there is one right answer to any question. In some kinds of testing we may rank answers from best to worst. Creating tests that allow for multiple right answers is outside of the norm, however, except perhaps in open ended essay style questions. It appears that this is exactly what we need to do if we are to measure complex moral/ethical decision-making.
If there is no single right answer, and the rightness of the various answers change according to the individual's value system and the specifics of the situation, one might approach the problem by measuring the students ability to identify the following for a given case:
This kind of testing will clearly require the use of layered or serial case studies in which a basic situation is initially described and tested, and then simple variations on this basic case are introduced and tested. It would be best if introduction of these variations were flexibly administered such that a student decision that favors a particular principle leads to a strengthening of the force of one or more of the other principles. These adaptive situational variations would allow the test to measure the student's ability to recognize changes in the situation and adapt decisions to changing conditions.
Any such test would need to be computer administered. Paper and pencil testing is unable to restrict the visibility of case variations, score the multiple answers that would arrive at each step in the near real time required if the exam is to be adapted, or to adapt as needed.. Open-ended answers with automated computer content analysis and classification might be a useful possibility in testing such cases, but it should also be possible to administer the above with variations of multiple choice testing. It is likely that such an instrument, and the requisite supporting computer software, will be developed in the near future.