







Subject  % ........S  % ........S  % ........S  % ........S  % ........S  % ........S 
Maxwell theory  38.....307  47......259  83.....269  87....230 


Vibration and waves 


87.....106  87.....122  93.....100  84....115 
Optics and Maxwell theory 


76.....165  65....180  56....134  57....190 
Mechanics and kinetic theory of gasses  42.....72  60......84  93.....116  96.....112 


Kirchhoff, Mechanics and vibrations  34......148  39.......124  95.....107  84......98 


Energy management 


82.....270  74....265  68......85  61....255 
Statistical thermodynamic 


98.....249  88......261  53.....390  54....578 
B. Lecture format
At the beginning of each lecture course the students are given a short talk on how the system works, to ensure a minimal level of confusion in its subsequent operation.
At the start of a lecture, the students are given photocopies of the overhead projections (OHPs) to be used. These OHPs use a basic color coding scheme: green to signify that the material is being repeated or is expected prerequisite knowledge, red to show questions, blue for the possible answers, and black for new information. The lecture itself begins, as many do, with a recapitulation of what was said during the previous one. Questions are asked of the students to test the understanding and recall of the matter, with response via APF.
Throughout the lecture, a difficulty in comprehension is signaled by there being more than 30% incorrect answers to a question of type a, b, or c; or if the time taken for students to signify their readiness is too long for a question of type d (the average wait is around 1 1/2 min, but this, of course, varies with the complexity of the question being posed) , When this occurs, the lecturer takes the students through the problem step by step and then asks a supplementary question (often on the blackboard) to check the new level of student understanding.
If it becomes obvious that the students are completely comfortable with a topic,'for example by questions being answered very rapidly and correctly, the redundant questions are discarded.
At the end of a lecture, the teacher asks "Who thought that was too fast/too slow/about right?" A desired distribution would be approximately 20% "too fast," 20% "too slow," and 60% "about right," as it is impossible to match everyone's ideal speed of presentation.
Overall, a lecture consists of around 20 min of APF functions interspersed between 25 min of conventional lecturing.
IV. RESULTS
It is difficult to measure the interest level of students in a lecture empirically, but there is some evidence that they prefer the use of APF. In addition, there are data showing that the use of the system improves examination results: while it is generally accepted that examinations are not a perfect measure of student comprehension, they are a reasonable indicator of understanding.
These results are not immune to any Hawthorne effect with respect to the studentsthey enter a new exciting lecture format and there may be bias from this. However, the system has been in use in Eindhoven for a considerable time (since 1966), so the lecturers using it have become very familiar with the system, mitigating any effects on them.
The 288 APFexposed and 19 790 "ordinary" students were asked on a Likert scale from one to nine (nine indicating a very strong positive): "Do lectures contribute much/ little to a better understanding of the subject?" The mean score for nonAPF students was 5.1; for APF students this rose to 6.7. This indicates a preference for the lecture when APF is present, and a positive reaction to it.
The endofcourse examination pass rate for a given course was measured over four years: either two nonAPF years followed by two with APF; or two with feedback and the two subsequent traditional years. In each case the academic year of each of the four years' students remained the same. The data are drawn from the period 1979 to 1992, and cover the faculties of industrial engineering and management science, electrical engineering, chemical engineering and chemistry, and applied physics.
To ensure consistency in the standard of understanding represented by an examination pass in APF courses during the periods considered, an independent supervisor from each faculty was appointed. It was felt vital that the course present the same amount of material, the supervisor be closely consulted when examinations were designed to ensure their consistency, and he/she be ultimately responsible for the courses' yeartoyear equivalence. rate Table I shows that the use of APF improves the pass in most of the variety of physical science lectures in which it has been used. The mean pass rate, Fig. 1, of the APF lectures is significantly higher than that where conventional methods have been employed. Of equal importance is the reduction in the standard deviation of this average, indicating a more consistent level of comprehension throughout any given class, and year on year. This in turn means that decisions on the required level of understanding assumed for future courses can be made with more confidence.
Fig. 1. The mean and standard deviation of the individual courseaggregated (data of Table 1) pass rate of 2550 students attending lectures with APF (shaded) and 2841 from traditional lectures.
V. CONCLUSION AND DISCUSSION
These results demonstrate that the application of APF in the lecture theater has been of significant use in the students' learning process; both increasing the mean pass rate of individuals exposed to it and reducing the variability between the achievements of different students.
These are being tentatively ascribed to four effects, which are, in order of decreasing importance: the removal of the "house of cards" effect, the negation of the inherent passivity of students in lectures, studentstudent teaching, and a mild Hawthornelike effect.
In a traditional lecture it can be extremely difficult to measure the students' comprehension of a topic. This can lead to a house of cards effect, where the lecturer is explaining a subject to students who have yet to understand its precursor. APF allows the lecturer to ensure that the majority of the student body has understood the material before moving on.
In addition to this, the students are given a role in the lecture, and play an active part in it. This increases their cognitive engagement and so material taught to them is considered more closely.
The students also spend some time discussing each problem; during this period there is some element of studentstudent teaching, or consolidation of material presented by the lecturer between students.
Finally, there is likely to be a residual Hawthornelike effect: the students are presented with a situation in which the lecturer has prepared a clear set of OHPs, and where they are given the special attention of the handsets. This cannot be completely discounted as an explanation until far larger longitudinal studies are undertaken.
With the results of other independent studies involving student feedback also showing promise, further exploration of the efficacy of lecture feedback will be conducted, with the aim of improving this teaching form.
The authors are grateful to Professor Sir Eric Ash for a number of helpful discussions.
Corresponding author.
Why do we still teach classical mechanics to graduate students in physics? It has been remarked...that when scholars in the humanities use the terms "classic" or "classical" they mean that something endures or is archetypical, but when physicists say something is classical, they mean it's wrong. Stephen Renolds, in a review of Florian Scheck, Mechanics: From Newton's Laws to Deterministic Chaos, American Scientist 80, 391392(1992). 