Based on that logic they shouldn’t have released a sample with percentile scores compared to the index score. (ETA: out of curiosity I looked at historical data ranges to see if those ranges were pretty normal, bc I just assumed they were, and they were. Raw scores in those range give approx those precentile results.
Fwiw, CB provided scoring for the practice test. No one has to guesstimate their score or index for that test. What no one can predict without internal data is if the percentile data will follow their previous tests. In those terms, even trying to predict a NM score is pointless.)
“Based on that logic they shouldn’t have released a sample with percentile scores compared to the index score. (ETA: out of curiosity I looked at historical data ranges to see if those ranges were pretty normal, bc I just assumed they were, and they were. Raw scores in those range give approx those precentile results.”
The sample should have been released showing something for percentile scores because that’s what shows up on the student’s score report. It’s for illustrative, not analytical, purposes.
As for scores in those ranges - yes, it’s the CONVERTED score that is going to be consistent - that’s why it’s called a converted score. You can’t really say how the raw score is going to look - there is no historical data on that yet.
So, all this talk about raw scores had ME curious about something - we have four new practice SAT’s and conversion tables for each of them. How much do they vary?
Using Math as a good example (Raw score max = 58 in all cases):
Raw score = 29:
Test #1 520
Test #2 540
Test #3 560
Test #4 570
Raw Score = 51:
Test #1 710
Test #2 730
Test #3 720
Test #4 750
If you miss 1, the converted score is 790 for tests 1 - 3, and 800 for test 4
If you miss 4, the converted score is 750-760 for tests 1 - 3 and 780 for test 4
If you miss 11, the scores swing from 690 to 740
Keep in mind that the percentiles as determined by test score (i.e. converted score) really don’t change from test to test. What particular raw score converts to that test score certainly can. For instance a score of 700 is at the 93rd percentile: That might mean a raw score of 48 - 50 or a score of 43, based on these conversion tables.
So a lot depends on the individual test. And golly but these norms aren’t even based on historical data exactly! (for the PSAT they say they used “research data” to determine the norms and I’m guessing it’s the same for the SAT). So for a real test I’d say at this point there is even more a priori uncertainty.
So, @ro2400 - does that answer your question? I wish it were a bit more consistent. (And I realize that you were referring to a real PSAT not the SAT practice tests but hopefully you can glean some insight from this). Anyway, If you only missed a couple then you are probably ok regardless.
I am not saying there is a conspiracy. I am saying this entire test appears to have been rushed into place and thrown together in too much of a hurry and that the scaling and grading are somewhat suspect.
The college board says, " Raw scores are converted to scores on a scale of 160 to 760 using a process called equating. Equating adjusts for slight differences in difficulty between various versions of the test (such as versions taken on different days).
The College Board uses equating to make sure there’s no advantage in taking the test on a particular day. A score of 400, for instance, on one day’s test means the same thing as a 400 on a test taken on a different day — even though the questions are different."
But where is the percentile information for the practice test? Absent percentile information, scaling is arbitrarily mapping one number into another.
Let’s look at the scoring for a kid who missed 5 questions in each section of the practice test. After clicking around a little, you can find the scoring booklet at https://collegereadiness.collegeboard.org/pdf/scoring-psat-nmsqt-practice-test-1.pdf Using the chart on page 7, it says that the raw reading score is a 42, that becomes a 35. The raw writing score is a 39, that becomes a 34. The raw math score is a 43, that becomes a 740.
Now I’m going to follow the instructions in my daughter’s sample test booklet to calculate this student’s “selection index”. The booklet reads (on page 55): "“The selection index is calculated by doubling the sum of the reading, writing and language, and math test scores.” Hmm. What is a “score”? Is it the raw score or the scaled score? Doesn’t say. Ok, let’s assume they gave us the scaled scores because we are meant to use them. The sum of the scaled scores is 35+34+740= 809. This student’s selection index is 809x2=1618.
I was looking through this forum for discussions and I came across a thread about an exact PSAT question…posted 5 days before the PSAT was administered…
“But where is the percentile information for the practice test? Absent percentile information, scaling is arbitrarily mapping one number into another.”
-The percentile information and the scaled scoring are pretty much one and the same. Not sure if that will hold for the new test but guessing it will.
"What is a “score”? Is it the raw score or the scaled score? "
It is the “test score”; that is, the scaled score, by definition.
“The sum of the scaled scores is 35+34+740= 809. This student’s selection index is 809x2=1618. Defend that.”
You have the 740 wrong. It’s 740/20 = 37. Her Selection Index as calculated for the new PSAT is (35 + 34 + 37) * 2 = 212.
The confusion is that the College Board doesn’t give us the first scaled Math score (called the “test score” for the other two subjects). But the scale is 38 for both reading and writing and for each to weigh in at 1/3 of the total that means the Math score must be scaled at about 38 as well. That implies that you divide by 20 (760 / 20 = 38).
@mathyone I noticed that typo too. They never updated that part of the language for the new test. I bet that paragraph was identical on the old book, and since the selection index calc wasn’t changing, the editor in charge of updating the materials missed the fact that math has to be scored back from the 740 by dividing by 20.
So you are correct - NO WHERE in the official docs does it say to divide the math by 20. We are going by internet chatter and what makes sense.
@picuberoot I saw that too. Really ticks me off. How do these people get their hands on the test ahead of time. They had the gall to come here to get answers. Sloppy on the part of the CB - somehow someone got their hands on the material.
@Mamelot Thank you for this analysis. I’m really tired right now and my brain is kind of dead - help me interpret. Is test #1 the ‘hardest’ according to the data you extrapolated, or is #4 the hardest? I can’t wrap my mind around it right now…
@suzyQ Test #4 must be the hardest because you can miss the most to get the same score.
And I have to disagree that there is a typo and that someone missed converting the Math raw score to a 38. It doesn’t need to be converted to a 38 for purposes of “PSAT” (as opposed to “Section Index”) scoring. Math is one section. Conversely, Reading and Writing are, together, ANOTHER SECTION with parts added up. So you convert to something simple (38 for PSAT 40 for SAT) add them up and multiply to get the 760-based score (or 800-based for the SAT).
There are two purposes to the PSAT. 1) NMSQT qualifying test - for that you kinda need to know how to calculate that simple converted Math score but obviously people figured it out. and 2) An indicator of how you will do on the SAT. In the recent past the two sets of scores were equivalent. But no more. And you’ll all be interested to know that back in the “olden days” of late 70’s / early '80’s when I was in High School the PSAT Qualifying Index was computed as “twice the reading score plus Math”. Math has always been 1/3 of the total qualifying score. At least for the past 35 years or so.
OK, why on earth do you have to multiply the verbal sections, but not the math? That chart on page 7 of the scoring information is garbage. There is no reason to give the math scores in one format and the verbal scores in a different format. It’s just one example of how poorly put together this test is. Of course I know that you have to get the math and the verbal scores on the same scale before you can calculate a selection index. But does the immigrant first gen kid know that? Why do you have to consult two different publications, and put the instructions together? And then the instructions don’t work–it’s wrong. Incredibly unprofessional.
If they can’t show a percentile chart for the scoring chart, then the scoring chart was just pulled out of thin air. Meaningless.
I’m the one that mentioned the dividing the scaled math score on the practice PSAT by 20 because that’s the only reasonable way to get it to the scaled score they want for the selection index. I’ll admit that NO WHERE in any official College Board document that I could find does it mention how to deal with the math score so I could be wrong.
Does anyone have any better ideas? I also showed DH Ima B Student’s sample score report and he agrees that the numbers are totally made up and don’t fit in with any real scaled scores.
This just goes to show your that the CB can manipulate the numbers any way they want.
@Mamelot, OK, I showed above how National Merit states that the qualifying index is computed but we all know that’s not what they are going to do.
Now as you said the other purpose of the test, indeed, the purpose for 99+% of test takers is to predict their SAT scores. OK, then why oh why is the PSAT no longer being scored on the same scale as the SAT? Since they are no longer on the same scale, where does the college board provide the conversion table that students can use to predict an SAT score from a given PSAT score? Are we meant to just guess? Then what’s the purpose of 9 pages of scoring instructions, if at the end of it we just take a guess?
I don’t see how anyone can defend anything about this.
If what the posters above are saying is true, then the test was leaked and the results aren’t valid. Talk about unprofessional.
"OK, why on earth do you have to multiply the verbal sections, but not the math? "
-Both the “simple” scale of 38 and the scale of 760 are similar. 38, after all, divides into 760. When you multiply a distribution by a constant, you don’t change the relative relationship of the percentiles LOL.
“There is no reason to give the math scores in one format and the verbal scores in a different format”.
They aren’t.
Not sure how to answer your question about the first gen. immigrant kid. I know several who would have no trouble understanding that 760 is simply a multiple of 38.
They will certainly make percentiles available because that’s part of the Score Report. I have never seen a percentile chart for PSAT. CB releases percentile charts for SAT and I’m not sure they will change all that much. The scaling is still 800, after all. The Reading and Writing Chart might be an average of the separate charts from earlier history. Of course this is speculation on my part but it’s one way to use historical data to arrive at percentiles that apply to the current test. The SAT hasn’t changed THAT much, after all.
@mathyone if you can show that the scoring chart was pulled out of thin air then perhaps you should take up your complaints to the College Board. There’s serious money riding on these scores, after all. No one can argue with a slam dunk case. Good luck with that!