资料题：According to the passage, which of the following experimental variables is most difficult to evaluate in experiments comparing programmed instruction to classroom teaching（  ）。

Do students learn form programmed instruction? The research leaves us in no doubt of this.They do, indeed, learn.Many kinds of students learn-college, high school, secondary,primary, preschool, adult, professional,killed labor,clerical employees, military, deaf, retarded, imprisoned- every kind of student that programs have been tried on.Using programs,these students are able to learn mathematics and science at different levels, foreign languages,English language correctness, spelling, electronics, computer science, psychology, statistics,business skills, reading skills, instrument flying rules, and many other subjects.The limits of the topics which can be studied efficiently by means of programs are not yet known.For each of the kinds of subject matter and the kinds of student mentioned above, experiments have demonstrated that a considerable amount of learning can be derived form programs; this learning has been measured either by comparing pre-and post-tests or the time and trials needed to reach a set criterion of performance.But the question, how well do students learn from programs as compared to how well they learn from other kinds of instruction, we cannot answer quite so confidently.Experimental psychologists typically do not take very seriously the evaluative experiments in which learning form programs is compared with learning from conventional teaching.Such experiments are doubtlessly useful, they say, for school administrators or teachers to prove to themselves(or their boards of education) that programs wok.But whereas one can describe fairly well the characteristics of a program, can one describe the characteristics of a classroom teaching situation so that the result of the comparison will have any generality? What kind of teacher is being compared to what kind of program? Furthermore, these early evaluative experiments with programs are likely to suffer from the Hawthorne effect: that is to say, students are in the spot-light when testing something new, and are challenged to do well.It is very hard to make allowance for this effect.Therefore, the evaluative tests may be useful administratively, say many of the experimenters, but do not contribute much to science, and should properly be kept for private use.These objections are well taken.And yet, do they justify us in ignoring the evaluative studies? The great strength of a program is that it permits the student to learn efficiently by himself.Is it not therefore important to know how much and what kind of skill's concepts, insights, or attitudes he can learn by himself from a program as compared to what he can learn from a teacher?Admittedly, this is a very difficult and complex research problem, but that should not keep up from tying to solve it.