doggy 说的很有道理。偶刚看到一个CDER的东东,也是关于抽题和判分模型的,是一个网友给GMAC写的询问信:“是否15秒内算猜题并不给高分”可能对大家有用,尤其在实力一般的情况下(既不是几乎没有不会的题,也不是基本不会的)策略有时就更重要了,赶快粘过来了。 俺在一战后,曾经就15秒的问题发邮件问过GMAC,俺的问题是: The second qestion is: My friend just told me that the "pace" is very important to the GMAT test. For example, if you done one question with a extreme quick speed just as within 15seconds to click the "next"and"yes", the system will make a judgement that you just guess this question and there will be no score or very low score for this question..Is it right? 得到的答复如下,红色部分是俺自己做的提亮: 2. For each multiple-choice section of the GMAT® exam, there is a large pool of potential questions ranging from a low to high level of difficulty. Each section of the test starts with a question of moderate difficulty. If you answer the first question correctly, the computer will usually give you a harder question. If you answer the first question incorrectly, your next question will be easier. This process will continue until you complete the section, at which point the computer will have an accurate assessment of your ability level in that subject area. In a computer-adaptive test, only one question at a time is presented. Because the computer scores each question before selecting the next one, you may not skip, returning to previous question later. Questions late in the test count just as much as early questions do in determining your final score. In general, it is true that the computer’s estimate of your ability changes more in response to a question early in a section than to questions later in the section, but that is because the computer has to evaluate you on fewer questions in the beginning. Near the end of the section, this estimate is based on a greater number of questions and is therefore a more reliable estimate. If you answer the first question correctly, for example, the computer can infer only that your ability is probably at or above the mean for the population taking the GMAT® test, but it cannot tell how far from the mean. It will then deliver a more difficult question, corresponding to a higher ability level, in order to begin the process of narrowing in on your true ability. If near the end of the section the computer delivers a question that is comparable in difficulty to question one, it does so with the benefit of a considerable amount of information about your ability level. A correct response in this instance would not cause the computer to make as large an adjustment to its estimate of your ability, precisely because it is taking into account more than this single question in computing its estimate. 个人感觉,虽然他们没有明确答复关于15秒的问题,(几乎都有点答非所问...)但是可以看到,Verbal的前10~15题真的非常重要,也许几乎就决定你的分数段。 In a computer-adaptive test, only one question at a time is presented. Because the computer scores each question before selecting the next one, you may not skip, returning to previous question later. Questions late in the test count just as much as early questions do in determining your final score. In general, it is true that the computer’s estimate of your ability changes more in response to a question early in a section than to questions later in the section, but that is because the computer has to evaluate you on fewer questions in the beginning. Near the end of the section, this estimate is based on a greater number of questions and is therefore a more reliable estimate. If you answer the first question correctly, for example, the computer can infer only that your ability is probably at or above the mean for the population taking the GMAT® test, but it cannot tell how far from the mean. It will then deliver a more difficult question, corresponding to a higher ability level, in order to begin the process of narrowing in on your true ability. If near the end of the section the computer delivers a question that is comparable in difficulty to question one, it does so with the benefit of a considerable amount of information about your ability level. A correct response in this instance would not cause the computer to make as large an adjustment to its estimate of your ability, precisely because it is taking into account more than this single question in computing its estimate. 个人感觉,虽然他们没有明确答复关于15秒的问题,(几乎都有点答非所问...)但是可以看到,Verbal的前10~15题真的非常重要,也许几乎就决定你的分数段。 |