Percentiles for CLEP / DSST

matt · #1 (**permalink**) 09-03-2008, 08:46 PM

Guys and Gals:

Does anyone have any information on score percentiles for CLEP / DSST exams? Are they published anywhere?

Best,

Matt

cookderosa · #2 (**permalink**) 09-04-2008, 12:57 PM

Quote:

Originally Posted by matt

Guys and Gals:

Does anyone have any information on score percentiles for CLEP / DSST exams? Are they published anywhere?

Best,

Matt

>>

Do you mean the raw score conversion to scaled score percentages? You won't find them- they are unpublished and top secret. If you want to know percentages of pass rates, look in my sig for the IC Works link- follow that about half way through and you will find a few great links that show a lot of pass rates/percentages.

matt · #3 (**permalink**) 09-04-2008, 04:58 PM

Thanks cookderosa, appreciate it, I reviewed both the DSST military stats and the CLEP study from the link you sent.

I don't get why they can't give you your scaled score with a percentile on exam completion. I understand why they would want to protect the raw score data, and any scaling formulas / metrics .... but they should be able to tell you score 60 (70th percentile of all examinees) for instance.

Matt

cookderosa · #4 (**permalink**) 09-04-2008, 11:02 PM

Quote:

Originally Posted by matt

Thanks cookderosa, appreciate it, I reviewed both the DSST military stats and the CLEP study from the link you sent.

I don't get why they can't give you your scaled score with a percentile on exam completion. I understand why they would want to protect the raw score data, and any scaling formulas / metrics .... but they should be able to tell you score 60 (70th percentile of all examinees) for instance.

Matt

>>

Well, I think (totally guessing based on what I have read the past couple years) it is because each question is probably weighted. What I know: I know some questions are being tested and won't count on your score. CLEP tells you this. How many? 1? 10? This must change as new questions are being challenged. 1 for me in March, 3 for you in April? Next, I believe based on the specific exam feedback here, that there are multiple versions of the exams. Lastly, what I know is that the performance of "C" students must vary wildly! For example, to get a "C" in sociology MUST be easier than scoring a "C" in calculus. I think the raw scores to scaled score calculates this in based on the testing they have done of actual college students (like in the article).

So, even if you say that you score a 50, so you "should" be 70% like a C student, that's not really accurate because you still don't know how many raw answers were necessary to score a 50, or how many raw answers were equal to the raw answers scored by C students.

Yeah, sometimes it hurts my head too.

gus · #5 (**permalink**) 09-09-2008, 07:03 PM

In a way, I believe Matt is correct as far as percentiles are concerned. Although we don't know the exact percentile ranking (not the percentage of correct answers), we could still estimate it on the previous DSST format (I don't have enough information to say the same about CLEPs).

To get a little technical, the previous DSST scoring format was a standardized score ranging from 20 to 80 with a standard deviation of 10 and a z-score=0 at 50 (the mean). Assuming the shape of the score distribution of the initial population of exam takers is a normal, bell shaped curve (which is to be anticipated given how large the initial population was) and not skewed, we can easily estimate percentiles. Since the standard deviation is 10, a score of 40 is a z-score of -1 and likewise a 60 is a z-score of +1. All we'd have to do is convert the original score to a z-score and look it up in the standard normal probabilities table to determine the estimated percentile. InstantCert had the table in their Statistics notes. To calculate the z-score divide the DSST score by 10 and then subtract 5 from the result. For example if you scored 65 your z-score = (65/10) - 5 = 6.5 -5 = 1.5.

There's an online calculator that may be easier to use than looking up values in z-score tables:

Z table - Normal Distribution

Using the curve on the top of the page:
1. Enter 50 under Mean
2. Enter 10 under Sd
3. Select 'Below' and enter your DSST score in the corresponding field

The answer should be under 'Shaded area'. For a score of 65 you should get 0.933, in other words, an estimated ranking at the 93rd percentile.

Again, this only works assuming that we are working with a normal distribution, which may or may not be the case.

Barcotta had suggested a while back that we start a thread on the topic but now that the scoring method has changed on the newer version I thought let's wait until we get more information.

Here are some Wikipedia links which may be of interest for those of us who haven't touched Statistics yet:

Standard score - Wikipedia, the free encyclopedia

Normal distribution - Wikipedia, the free encyclopedia

Standardized test - Wikipedia, the free encyclopedia

Percentile rank - Wikipedia, the free encyclopedia

gus · #6 (**permalink**) 09-09-2008, 07:08 PM

This is off-topic but I just noticed your updated signature, Jennifer, you start Harvard in around 5 days! Best of luck!

I hear it's a lot of work so I hope you're ready.

Have you decided what degree you plan on pursuing?

cookderosa · #7 (**permalink**) 09-09-2008, 11:38 PM

Quote:

Originally Posted by gus

This is off-topic but I just noticed your updated signature, Jennifer, you start Harvard in around 5 days! Best of luck!

I hear it's a lot of work so I hope you're ready.

Have you decided what degree you plan on pursuing?

>>

Gus, I have nearly 200 pages of reading due before the first day

so you may notice I'm not here as much as before. BUT, I am quite excited and ready for the class (Abnormal Psych, PSYC1240, Carson) and I am always excited to share with others how this board has helped me, and give back any way that I can. 19 months ago, I would have laughed off the idea of earning a bachelor's degree.

I will wait to decide if/which degree to pursue, as you may know, I can't apply to Harvard's ALM (Master Liberal Arts) unless I complete 3 courses inside my concentration with a B or better, and one must be the proseminar. I'm not 100% sure of my concentration, so I may take a few just to "see" which area really lights my fire. My interest (from the choices) include psychology, anthropology, and English. My real interest (in general) is in sociology, specifically food-related sociocultural issues.... so that is a slight sticking point. In the end, I will have to choose between many pros and cons no matter where I end up. I do have 3 grad schools hanging- they all have me planned for Spring 09 start. All in different programs with very different price tags. I have not done anything with them yet- I want to see how this class goes first. I may leave HES with my tail between my legs. LOL Since there isn't a "perfect" program for me, I will have to do my best to find the best fit. You can be sure that I will keep the board abreast of my Harvard experience. So far, I think it's awesome, and I am simply giddy!

twosidneys · #8 (**permalink**) 09-10-2008, 01:08 AM

I will be very interested to hear about your Harvard experience. I had toyed with the idea myself. Tell all!

Sidney

gus · #9 (**permalink**) 09-15-2008, 10:40 PM

I can sense the excitement, it surely is going to come in handy along the way. I also know what you mean about the dilemma of choosing the concentration. I find myself toying with the idea and not finding the exact fit, but it's just too tempting I think I might at least go for one course to see if it's feasible.

By now you must've already started and I really wish you the best of luck!

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode