<p>How does Collegeboard curve the subject tests?? how do they decide what the curve will be for each particular test?? and is it based on other people within your state or is it a nationwide curve?? please answer and thanks a lot!</p>
<p>they look at everybody else who took the same test and see what the average score was (to judge how difficult the test was). based off of this, they make a table to convert your raw score into a scaled score out of 800. a generous curve would be a hard test where you can get more wrong and still get an 800. </p>
<p>i saw one person mention that it might be statewide curves, but in my opinion thats just not possible because it would defeat the purpose of a curve (which is to make different test day test scores comparable). people shouldn't be punished if they live in a "smart" region and shouldn't gain an advantage if they live in a "not-smart" region. it most likely is a national curve.</p>
<p>how do they decide where to cut off the 800?? do they have to meet a certain quota persay of people who get 800s?</p>
<p>No, they don't curve the test based on the people who took that test. They have a rubric that determines the difficulty of the test, and they curve the test before it is given. The raw scores are converted into the scaled score before the test is given. That's the only way that tests can be compared to each other across time, and it does not depend on who else took your test. </p>
<p>There is no "quota" of people who get an 800 on any particular administration of the test.</p>
<p>do they really have athe curve made up before the test is administered? how would the accurately determine how hard the questions (and overall test) are unless they see a general trend in students who actually take it? a large number of students take it and the data from such a large set would be more conclusive than maybe a few hundred question makers deciding on how a question is.</p>
<p>they don't have a way to give the questions "trial runs", like they do for sat 1s and practice sections, do they?</p>
<p>does anyone else know?? because there seems to be a conflict lol</p>
<p>Well some tests, like the Chinese tests have averages in the 768s, while some have averages in 602's, and others in the 658s. </p>
<p>I really doubt that they allow only a certain amount of people to get 800s.</p>
<p>yes, the curve is preset based on the type of question and the percentages of test takers getting it right. they take those stats for every single administration of the test they do...you can see percentage of people you get the question right in the answer section of the real sat u.s. history tests</p>
<p>Here is how the CB does it:</p>
<p>Quick summary: repeated questions from previous tests are used to determine the difficulty of the new test (the repeated questions appear in those "extra" sections which don't count toward your score). With repeated questions, the CB can distinguish between, say, a test which was too easy and an unusually strong bunch of students (in both cases, the raw scores would be higher than usual).</p>
<p>In this way, the curve is independent of how good a particular bunch of test takers is. The curve does NOT attempt to keep percentile ranks constant. So, for example, it would be possible for 2% of test takers to get a math 800 in June, and 5% in October.</p>
<p>See also the thread: </p>
<p>fignewton, isnt that only for the SAT I??</p>
<p>A similar "equating" is done on the subject tests to compensate for test difficulty. Here, though, there isn't a separate section that doesn't count. So, questions that were previously used have to be mixed in with new ones. This is why there isn't a "QAS" for the subject tests. When you get a QAS test back for the SAT, the "equating" section is missing.</p>
<p>If anyone knows more specific info on the subject test equating, please post some links!</p>