04-13-2006, 05:35 AM
ShotoJuku Wrote:So riddle me this then: Based on the chart, how many questions (raw score) does one need to answer correctly in order to receive the lowest passing (scaled score) in the Rise & Fall of the USSR????? THANKS!!
Hi ShotoJuku,
I'm not a statistician by any means. But I don't think the chart provides enough information to backwork the formula. I could be wrong. But they do say in the pdf file that:
"The calculation of a standard score requires a different conversion/formula for each test. Without the formula, one cannot work back from a standard score to learn the number of correct responses."
[SIZE=2]Without knowing the specific formula DANTES uses for EACH exam, and for EACH form of each exam, it is not possible to convert raw scores to scaled scores.[/SIZE]
[SIZE=2][/SIZE]
[SIZE=2]Probably the best we can do is note what the mean raw score is for any particular exam.[/SIZE]
[SIZE=2][/SIZE]
[SIZE=2]If we look at the "100 item" version of the Soviet Union exam: The average raw score of the sample test-takers was 50 correct answers. A fair assumption might be that you would likely need to score higher than the average to pass the exam.[/SIZE]
[SIZE=2][/SIZE]
[SIZE=2]The standard deviation is also provided (13.9).[/SIZE]
[SIZE=2][/SIZE]
[SIZE=2]If the distribution is a standard normal distribution, we would expect that 68% of all sample test-takers scored between 36.1 and 63.9 correct questions out of 100.[/SIZE]
[SIZE=2][/SIZE]
[SIZE=2]We would also expect that 34% answered between 50 and 63.9 correctly (1 standard deviation above the mean).[/SIZE]
[SIZE=2][/SIZE]
[SIZE=2]Further, we could assume that 13.5% answered between 63.9 and 77.8 correctly (2 standard deviations above the mean).[/SIZE]
[SIZE=2][/SIZE]
[SIZE=2]The top 2.5% will have answered between 77.8 and 100 correctly (3 and 4 standard deviations [/SIZE][SIZE=2]above the mean).[/SIZE]
[SIZE=2][/SIZE]
[SIZE=2]As you can see, without the necessary formula to convert these raw scores to the DANTES scaled score system, the raw numbers are not that useful. But they might give you a general feel for where in the pack a particular raw score lies.[/SIZE]
Hope that helps,
Snazzlefrag
[SIZE=2][/SIZE]
[SIZE=2]
[/SIZE]