The administrator of an existing school or of a school being started must make many decisions with regard to curriculum and methods. His goal is to make education better. But how can he choose the right means to this end? How can he recognize which means work and which do not? There are basically two ways of knowing which ones work. Since these two ways are like the two ways of knowing anything, we will give them labels borrowed from epistemology. The historic rivals in epistemology are rationalism and empiricism. Rationalism supposes that knowledge arises from a priori thought. Empiricism supposes that experience is the basis of knowledge.

Rationalistic Assessment

Educational innovation usually begins with someone conceiving an idea of how to improve education, or with such an idea arising anonymously from current educational thought. Then, after the idea has become widely known, some educators with authority over schools decide to implement it. Only later, at a time well after its introduction, is the innovation evaluated for effectiveness.

The rationalistic way to evaluate it is simply to ask whether it makes sense. And, frankly, most innovations in schools are never subjected to a more rigorous test. They continue in use if administrators and teachers, relying solely on subjective impressions, judge them to be appropriate and successful. But to view innovations so uncritically leads to dead weight in educational practice. An attractive new way of doing things may have hidden drawbacks.

For example, it may depend on the wrong model of human learning. Secular educators, seeing learning as somewhat analogous to evolution, have always been inclined to view the human mind as a tabula rasa, capable of progress to knowledge only in logical increments beginning at ground zero. Yet the mind is a creation of God, who fitted it to reach great heights of understanding within a few short years. Pardon the digression, but my granddaughter could read, truly read, before age four. No tabula rasa could do that! The brain comes prewired to perform many tasks without instruction, or with no instruction except the common experience of mankind. Thus, a plausible proposal of something new to teach, or of a new way to teach something, may generate an innovation that wastes everybody's time.

A topic presently emphasized in many lower-level textbooks in math is pattern recognition. Pattern recognition is indeed assumed by higher-level math skills. But does it need to be taught? Does instruction in pattern recognition actually build useful connections between brain cells? Perhaps these connections exist already in most children, or will develop naturally in the course of ordinary experience. Although the schools of yesteryear never taught pattern recognition, the omission seemingly did not retard student progress in math.

Another hidden drawback to some innovations is that they unintentionally subvert one or more factors supporting successful education. One example is the detailed lesson plans that teachers must now prepare for all courses. Such plans may be helpful to some teachers of some courses, but for other teachers and courses they may not be helpful, and the overall effect of requiring them may in the long run be deleterious.

I have taught math for many years with good success, as judged both by student achievement and student morale in my classes. Yet I rarely do lesson plans beyond writing down the assignment, and my preparation for a class rarely involves more than a quick glance at the topic. In other courses, I often do more elaborate preparation, but in math I proceed not according to a plan, but according to a strategy, which is always the same. When I enter a classroom, I know the subject inside out. That is a given. I do not need to teach myself the subject every time I teach it to others. After outlining basic concepts, I proceed quickly to problems (because what students must know, both for homework and tests, is how to work problems), and in presenting my solutions I follow a procedure, refined through long practice, that seeks to draw students into my way of thinking. Whether I do a problem or ask a student to provide the steps (there are also other possibilities), whom I call upon, what I emphasize, whether I go back to something earlier, what I do next, whether I keep the original assignment—all these decisions depend on feedback from the students. The multitude of options at each choice point fall within my teaching strategy, but could never be anticipated in a lesson plan. I teach math pretty much the way I was taught it, back when education was both less complicated and more successful. My approach might be considered a variation on the time-honored Socratic method, which uses spontaneous dialogue to shape the thinking of students.

It is doubtful that doing fuller lesson plans would measurably improve my math instruction. What it would do is adversely affect both my work load and my satisfaction in teaching. Not being overworked and not resenting or dreading my job contribute more to my success as a teacher than fuller lesson plans ever could.

Empirical Assessment

The best way to find which innovations work is empirical. Some effort must be made to gather objective data that will shed light on whether the innovation is truly effective.

Pilot trials

Ideally, an innovation should be tried first in a few schools or classrooms and evaluated there before it is disseminated. Measures of its effectiveness in moving children toward some desirable goal cannot be definitive, because confounding variables always intrude. Nevertheless, such measures should be able to detect a notably useful innovation.

Field trials

The usual practice today, however, is to take an innovation straight from the ivory tower to the wide world. Empirical evaluation of an innovation after it has become a common practice of real schools is extremely difficult. The schools adopting a particular change differ from other schools in so many indeterminate respects that it is impossible to tell whether any observed difference in achievement is due to the innovation or to confounding variables. Before and after comparisons are likewise meaningless, for they too allow confounding variables: namely, all those changes associated with the mere passage of time.

The popular strategy in educational research for circumventing these problems is to use statistical controls. Comparisons are restricted to groups perfectly matched on presumed confounding variables, or the contribution of these variables to observed differences is partialed out by mathematical techniques. Yet, this strategy is just another brand of futility, because measurement is inherently imperfect. Two groups may seem identical in some characteristic (socioeconomic class, for instance), but if they were drawn from larger groups differing in the same characteristic, the smaller groups still differ, because the measurements used in selecting them contain error.

In a study comparing matched groups, whether actual or mathematically contrived, it is sometimes possible to estimate the reliability of the devices used to measure confounding variables, and thus to assess whether the amount of error is sufficient to account for observed differences in the effect variable. Yet even this more sophisticated approach to educational research is doomed to failure, because, although we might be able to determine a measure's reliability, we almost never can determine its validity. To the extent that it is invalid, it is in error. For example, who knows, or who can find, the actual correlation between income and the ill-defined continuum known as socioeconomic class?


The most powerful research tool is experimentation. Yet experimentation on human subjects, if both the treatment and the effect involve mental processes, leads to doubtful results, especially in real-world situations. Indeed, we can say that such research posing as experimentation is probably bogus. Why? Because the human mind is creative rather than mechanistic in its operations. Being capable of unpredictably adding thought to thought, it can always respond to an experimental treatment in such a way as to generate confounding variables. For example, suppose I wish to determine whether spanking is a superior form of discipline. I might choose a representative sample of schools, randomly assign them to treatment and control groups, carefully design the treatment so that it introduces no change except the consistent use of spanking, and in like manner design the control so that it introduces no change except the consistent disuse of spanking, and then, after a time, compare the two groups on relevant variables, such as order in the classroom. This experiment is impeccable on the surface, but worthless. School discipline cannot be divorced from its social context—in other words, from all those minds in the community who think about what the school is doing. Even though spanking might be a superior form of discipline, students and parents ideologically opposed to it might react to its introduction in such a way, with anger and resistance perhaps, that the new policy produces a deterioration in classroom behavior. The right control for this experiment is difficult to imagine.

Wilhoit has observed,

A closer examination [of social science research] reveals that experimental results are more like the mythical nine-headed serpent Hydra. When one of its heads was cut off, two appeared in its place. Likewise, social-science research typically leads to more questions and research, not to definitive answers (1).

What he says is true, but he does not quite hit the nail on the head. Although some empirical studies in social science (including education) are better than others, almost none are methodologically adequate to give the desired answers. They fail because they cannot cope with the free will of man and the complexity of human society.

The difficulty of evaluating innovations argues against introducing them too quickly, without testing them first in small laboratories of learning. A change in methods or curriculum may be costly, and once it enters real schools it may quickly become entrenched, so that to eliminate it may require overwhelming evidence of its failure. Public schools today are full of innovations—operational definitions of objectives, look-say reading instruction, social promotion, to name a few—that persist mainly because nobody has been able to show conclusively that they do not work.

Regression Analysis

The best way to set school design on a solid empirical foundation is to broaden the scope of research beyond evaluation of particular educational practices. As I have shown, the evaluation of any practice old or new is fraught with difficulties, because of the limitations in educational research. It is more profitable to look for the variables that fully describe the difference between good classrooms and bad classrooms.

The procedure is simple. A battery of measures with proven value in assessing educational success is administered to a large number of classrooms. The very best and the very worst, by these measures, are subjected to detailed objective observation and study, the results of which are submitted to regression analysis, using a single score of success as the criterion. By this procedure, it should be possible to find factors accounting for nearly all the variance in success. Once they have such information, educators will know what factors are important in designing a good classroom. Whether these factors are the actual causes of educational success is beside the point. (Again, achieving perfect separation of causes from confounding variables exceeds the power of educational research.) What we need for good education is not complete understanding, but enough understanding to build classrooms that mimic successful ones in every important respect.

It is inconceivable to me that such a regression analysis would discover anything surprising. The factors essential to good education are doubtless the ones long recognized by common sense and traditional practice. The best setting for learning involves a demanding curriculum well tailored to present achievement levels. Moreover, it involves good order in the classroom, as well as a sufficient and reasonable amount of meaningful homework, a discriminating grading system, and a climate of parental support for high academic standards. It is also helpful if the teacher likes his job, knows his subject, and has experience to draw upon. Reasonable limitations on teacher work load and class size are beneficial as well. Schemes for school improvement yield the greatest benefit when they concentrate on honing these factors to greatest advantage.

Outcome-Based Education

Just as there are two ways to evaluate curriculum and methods, there are two ways to evaluate schools. The rationalistic way judges how closely a school conforms to a preconceived model of a good school. This is the way taken by accrediting agencies everywhere and at all levels. Yet the empirical way makes better sense. A school should be judged by whether its graduates are well educated. Outcome-based education—that is, education judged by its success in producing desired outcomes—is good in principle. The problem is, who decides what it means to be well educated? Public education may come to the place where it recognizes that schools must be held accountable to tests of achievement. But achievement means one thing to the world, another thing to Christians. The world wants school graduates to be good global citizens, in tune with prevailing secular myths and philosophies, equipped to join the labor force and support a consumer-driven economy. A Christian who holds old-fashioned views on salvation and creation and who wishes to spend his life serving God is not a well-educated person, in the world's view.

Thus, the main battleground for Christian education in the future may be the question, which tests will be used to evaluate schools? Wherever Christians wield power in the courts and legislatures, they may soon find it necessary to insist on the right of Christian schools to evaluate themselves with tests that are neutral ideologically and sympathetic to traditional educational goals.

Teachers should also be judged empirically, not rationalistically. Whether a teacher holds the credentials we might presume necessary, or whether he uses certain methods, or whether he subscribes to certain educational theories is really a poor test of his competence as a teacher. A far better test is actual performance. Assessing his performance requires only a handful of bottom-line measures. His students must show good progress on relevant and sensitive tests of achievement. He must issue a satisfactory distribution of honest grades. He must promote good order in his classroom and in the school as a whole. He must be generally well liked. And his spiritual influence on students must be constructive.

Academic Freedom

When I was young, teachers were very protective of academic freedom. At all levels, but especially at the collegiate level, they resisted regimentation. They felt that as professionals, they deserved the right to design, organize, and implement their courses according to their own best judgment, in light of their own interests and abilities.

Although there have been many abuses under the banner of academic freedom—abuses arising from neglect of the responsibilities that true freedom always entails—academic freedom is in itself a worthy ideal. It is one of those freedoms like political freedom, economic freedom, and religious freedom that we should work to preserve. The great danger in not keeping academic freedom alive is that the state, with its many self-serving agendas, will gain absolute sovereignty over the education of children.

We live at a time when all freedoms are under attack. As the institutions of business, government, and education become bigger, they accumulate more power to limit our choices and channel our decisions. Self-determination is rapidly shrinking to a narrow range of consumer options. Americans once believed that the defense of freedom was a patriotic, even a Christian, duty. Is that duty any less incumbent upon us today?

Schaeffer agrees that the greatest evil we must fight at our moment in history is the loss of personal liberty.

I wonder if Christians of the future will be thankful that in our day we spoke out and acted against . . . the special sickness and threat of our age—the rise of authoritarian government? That is, will we resist authoritarian government in all its forms regardless of the label it carries and regardless of its origin? The danger in regard to the rise of authoritarian government is that Christians will be still as long as their own religious activities, evangelism, and life-styles are not disturbed (2).

I do not mean to sound like some sort of militiaman. Mustering in a cornfield will do nothing to further the cause of freedom. In today's world, violent dissent is an invitation to be crushed. We must understand that God grants us freedom when we convert it into greater good than we would achieve under persecution. Therefore, the road to freedom is repentance, integrity, and service—in other words, revival.


  1. Jim Wilhoit, Christian Education: The Search for Meaning, 2nd ed. (Grand Rapids, Mich.: Baker Book House, 1991), Wilhoit, 120.
  2. Francis Schaeffer, How Should We Then Live? The Rise and Decline of Western Thought and Culture (Westchester, Ill.: Crossway Books, 1976).Schaeffer, 256.