Have there been studies about how to order the difficulty of questions in tests to get the best scores? The questions could be easiest to hardest, hardest to easiest, or random, and it would seem that there would be a difference in how the students did on the test.
What got me thinking about this is something my daughter said about her math final. It was in two parts. The first part was questions from the college AP exam and the second part was from class work. The AP questions were unfamiliar and difficult, and she blanked on how to solve them. She got anxious and thought she would not be able to do any of them. She went to the second part and was able to easily complete them because they were more familiar. Then she went back to the first part and was easily able to solve them. So it would seem that the harder initial questions gave her anxiety, which could have caused her to do poorly on the test. If the easy questions were first, she likely would not have had the anxiety at all.
Then I was thinking it might even be better if there were some trivially easy questions at the beginning of the test. This would serve to give the student confidence and reduce test anxiety. Even if the easy questions didn’t count for score, it seems the confidence would allow the student to do better on the other questions.
I always took this advice as a time management thing, “don’t spend so much time fighting with a tough question that you miss an easy question you could have done easily except that you ran out of time”. In other words, if you going to miss/rush questions because of a lack of time, miss the questions that you weren’t likely to do well on anyway.
The OP suggests a different, purely psychological, effect independent of time pressure.
Reflecting on what the prior posts have said, it occurs to me that many students may not use the technique of flipping through the test to find the early questions at the outset, but may simply start with question 1 and proceed in order. In that scenario, if the easy questions are front-loaded, scores are likely to be higher, if for no other reason than that the student is likely to complete more questions during the allotted time for the test. If the test is not timed, there may be a tendency for scores to flatten out somewhat.
That’s a good point. It’s not always clear which questions are easy just from looking at them. If the easier questions were first, the students who completed the test sequentially would get through more of the questions in less time.
I think one key factor would be how susceptible the student is to test anxiety. Some students are very intelligent and know the material, but freeze when it comes to taking the test. I would expect those type of people would have a significant difference in scores if the initial questions were easy. The easy questions would reduce their anxiety and they would be better able to proceed. Students which don’t have test anxiety may not show as much of a difference.
Yes, however, there are many instances where high test scores benefit the test giver. For example, many school districts receive funding based on how well the students do on standardized tests. They are highly motivated to make sure the students get high scores. During testing week the school won’t assign homework and there will be extensive test preparation. So if they could structure the test questions in such a way that the scores are optimized, I’m sure they would want to use it.
Besides, the test should reflect how well the student knows the material. If the student freezes up because of anxiety and fails, their score doesn’t reflect their true level of knowledge. In some disciplines you may want to weed out people with that type of anxiety (doctor, lawyer), but in most cases suffering from test anxiety is irrelevant to performing your job.
I don’t think building a test that is designed to have some students fail is a particularly effective pedagogical tool. Now, some students WILL fail–not bright enough, didn’t study, material bores them–but the intent of the test is not to have X percent fail. As Filmore says, it’s to find out how much kids have learned.
The OP’s point is about how to better assess students’ learning, specifically by making kids more confident in their test-taking ability.
I wonder though if “easy questions” might vary from student to student. Some might find analogies very easy, others questions about the parts of speech, or reading comprehension, while finding the others hard. How do you determine what “easy questions” will be easy for all?
The very point of standardized tests is specifically that those that would benefit by the higher scores are not preparing the exam…
I don’t think test anxiety is really a major problem. For students that truly have anxiety, they can work with the school to get extended time or extra breaks to help cope. For the vast majority of students, it is their level of preparation that governs their score.
An alternate explanation would be that your daughter read the questions, didn’t immediately know how to solve them, and her brain had time to work them out a little while she did the other problems. When she went back, it “clicked” on how to approach them. It’s possible that if she had come across them second, she would have done worse without the extra time between first seeing them and then coming back to them.
The point would be to structure the test to neutralize the effect of anxiety. Even though everyone may take the same test, they may not all react to it the same way. For example, pretend that test anxiety was inversely related to income. Then the rich kids do better on the test because they have less anxiety and their districts get more money. If there was a way to structure the test such that it produced the least anxiety, the results would be a more accurate representation of the students’ knowledge.