A Study On The Factors Affecting The Difficulty Of Physics Examination Questions In Senior High School

As the very important means of education assessment, the examination exists incontestably and develops constantly, though it had rise and fall in its developing process. Now in China, the number of examinees who attend the national matriculation examination is closed to 10,000,000 at one year. Obviously, it's very important to improve the quality of such a large-scale examination which involves thousands of families and various aspects of society.The difficulty is one of the very important aspects to evaluate the test quality. In the process of setting a paper, to get a full understanding to the difficulty of the test questions is the guarantee to the promotion of the test quality. Nevertheless, the estimation and adjustment to the item difficulty is not so satisfied. The main reason is that we don't get a sufficient grasp to the affecting factors and lack in-depth research in this field. For instance, there is no systematic study on the affecting factors of physics test items. So to study the affecting factors, from theory analysis to the demonstration, is extremely significant to the test paper designing and the classroom teaching.With the method of questionnaire, this research investigate the students and teachers' understanding to the difficulty of physics test items, and probe into the affecting factors of item difficulty. On the data of the questionnaire and the former research result, we bring forward the hypothesis of 15 factors affecting test items of physics in senior high school, that is: the number of involved characters, the insufficient or unnecessary resource, the number of involved knowledge points, the number of modules, the complexity of physics processing, the complexity of mathematics processing, the opening extent of the question targets, the physics modeling difficulty of context features, the scoring probability of guess, the degree of hint, the novelty of context, the expression way of question, the status of involved knowledge in daily teaching, the steps of asking and the familiarity of background information.Using the statistical method, which employed by the researchers of Educational Testing Services while they investigate the item difficulty, for reference and bringing the specific features of the test items of physics in senior high school, this research establishes the evaluating criterion of the individual affecting factors, and, based on the criterion, provides a statistical analysis to the 15 hypothesis of affecting factors and statistically discusses the affecting tendency to item difficulty of these factors. The main findings are as the following:Statistically, there are four factors which significantly affect the difficulty of physics test items, i.e., the number of involved knowledge points, the complexity of physics processing, the physics modeling difficulty of context features and the complexity of mathematics processing. These factors generally exist in the physics test items and the very factor which affects most to the difficulty is the physics modeling difficulty of context features. The following factors take turns, i.e., the complexity of physics processing, the complexity of mathematics processing and the number of involved knowledge points. The affecting tendency is, the more difficult the physics modeling of context features, the more difficult the test questions are; the more the number of physics reasoning (or physics equations), the more difficult the test questions are; if the complicated mathematic operation or reasoning the question employed, the question is more difficult; the more the question involved knowledge points are, the more difficult the test questions.On the basis of statistical analysis to the 15 hypothesis of affecting factors, this research also uses the parallel testing method, which employed by the researchers of UCLES while they investigate the validity of contextualized test questions, for reference to give a making-up study to the affecting factors. The main results are:On the features of the questions context: if we present a real, non-artifactitious or abstract context to the examinees, with more non-related information and the more implied model, the physics modeling will become more difficult. In most of the cases, the familiarity of context can reduce the difficulty of physics modeling; but on the other hand, just because the familiar context, the examinees are inclined to the traditional thinking and answer the question with their familiar models, as a result, the examinees neglect the analysis to the new context and give a wrong answer to the question. Hereby, to the questions which surface features of context are familiar with the former but its substantial context has changed, the more familiarity the surface features employed, the more misleading will be to the physics modeling, and then the more difficult the question is. Due to the different contexts to the same question, the difficulty will be likely to vary a lot. On the expression way of the questions: The habit of wording and the diction all possibly affect the difficulty of the test questions. The graphics may facilitate or embarrass or make no difference to examinees' answering process. To the questions which got a much complicated context and a comparatively nonfigurative space-time relation, the attachment of graphics will markedly facilitate to answering questions for the examinees having secondary ability level. At the same time, if the context and space-time relation are very simple, the graphics don't work. If we didn't pay attention to the details of the graphics, it may lead to the examinees' misunderstanding to the questions and increase the difficulty. With the aim of providing an accurate, brief and much clear expression of the questions, the designer can choose the literal, graphic expression or the combination of these two types to assist examinees' understanding, in this occasion, the different expression methods have no markedly difference in difficulty of test questions.In some test questions, there is more or less resource than needed. This research found that, in such occasion, the difficulty of questions is inclined to increase. When there is much more redundant information in the question, the examinees need to exclude the redundant and to distill the useful. Obviously, in this condition, the requirement to examinees' ability to distill information is enhanced and the question difficulty is increased. When there is no enough information for examinees to use, i.e., the designer didn't provide all the necessary information of the question, it needs the examinees supply the necessary information. However, maybe such information submerged in the large quantities of information in their brains. Consequently it essentially enhances the requirement and increases the question difficulty.Some questions are employed opening-answer way, so the correct answers are not sole. This research result assumes that the opening-answer questions are likely to get the tendency of increasing the difficulty.While presenting the target of a question, designing the ladders for answering is different with asking in steps. If the sub-questions in asking steps are not related with each other, then there is no essential difference between this and the one-step question. Statistically it doesn't affect the question difficulty. Contrastingly, if the sub-questions in asking steps are related with each other, and the answer of the previous sub-question can give a clue to the answer of the next one, i.e., the sub-questions are in proper sequence or ladder-like, it can help examinees understand the question and decrease the difficulty. The result of parallel testing research shows that designing ladder-like sub-questions is likely to have the tendency of reducing the difficulty.Finally, an identification survey was carried out to the experts and teachers. The results indicate: Most of the experts/teachers agree with the conclusions drawn from this research. Using the four significant affecting factors to analyze the examination papers, the result indicate that the sequence of the questions difficulty scale which determined by these four affecting factors has a significant correlation with the teachers' forecasting difficulty and the real testing difficulty of the test items. It indicates that the four affecting factors and their measure criterions proposed by this research employ a significant meaning in the forecasting of test items difficulty.