So in response to a couple of posts, this is (mostly) the way that your exams are graded: Your exams are marked by computer with numerous quality checks and then you are given a numerical score that is simply the number you got right out of 240. You will never, ever see that number and CFAI would do everything short of contempt of court to stop you from seeing it. The only thing you really care about is pass/fail, anyway. The pass/fail mark is set by a group of people called standard setters. These are charterholders who have graded for many years and are part of the in-grader clique (you probably don’t want to be there). These people get together and decide what a “minimally competent” LII candidate should know by looking at a variety of tests and much secret discussion. In the end, they decide on some number that shouldn’t depend on a passing rate, just a passing score. Then there’s this thing about “if you’re borderline, they weigh ethics more”. I’ve heard this is true from some people who ought to know but I’ve never seen it on anything official. As the number of candidates increases, I believe it less and less. Anyway, I don’t really know about that part. Next month you will get a score report like that posted all over AF that gives you score ranges. People try to estimate their scores using this 40/60/80 rule which means nothing except that if you line up people on 40/60/80 it corresponds reasonably well with pass/fail. In past years, it seems that people <60 on 40/60/80 almost always fail, > 70 almost always pass (maybe always), and in between is grey with a 40/60/80 passing score somewhere around 63-65 is usually about the boundary. However, since this should have no bearing on setting this year’s passing mark this is sort of irrelevant (except that the passing score this time will be the same because it always is).
Thanks Joey D for the nice explaination.
Joey - good to see you back… was wondering where you were!! “These people get together and decide what a “minimally competent” LII candidate should know by looking at a variety of tests and much secret discussion.” so are there some “need-to-answer-right” questions that are taken into consideration to decide if the candidate is a minimally competent LII candidate? I remember reading this in a post somewhere ( I think before June 07 L1 results) but no one had validated this statement… Thanks, Ruckmani
I know this is the Analyst Forum, but be careful of over-analysis. Just learn the material well enough to get 70%, and you will pass! All this stuff about “you’ve got to pass ethics/fsa” or “If you spin round 15 times before handing in your paper, they’ll let you skip straight to level 3, but only if you’ve written the secret code” reminds me of my favourite ever psychology paper (pigeon superstition paper for those of you who have seen it before): http://psychclassics.yorku.ca/Skinner/Pigeon/ The CFAI let us know as much as they want us to know, everything else is speculation. They won’t let us know our actual marks because you might end up “Super CFA” (>90%), which isn’t what they (or people with <90%!) want. In addition, if they published actual marks, then many borderline candidates would challenge the result. Therefore they give you the 3 ranges of marks, which (it seems) by design doesn’t provide enough information to derive marks, apart from through rough approximations like the 40/60/80 rule - which is too imprecise to be useful in borderline cases, even with chaps with PhDs in Stats on the board 
ruckmani Wrote: ------------------------------------------------------- > Joey - good to see you back… was wondering > where you were!! > I’ve been workng on a project pretty hard… > > “These people get together and decide what a > “minimally competent” LII candidate should know by > looking at a variety of tests and much secret > discussion.” > > so are there some “need-to-answer-right” questions > that are taken into consideration to decide if the > candidate is a minimally competent LII candidate? > I remember reading this in a post somewhere ( I > think before June 07 L1 results) but no one had > validated this statement… > Probably not explicitly because they are just coming up with a number, but for sure there are questions that nearly everyone with a passing score answered correctly and those must have influence. > Thanks, > Ruckmani
hmmm… that makes sense… I was hoping that it would not be something like pre-decided Q23, Q56, Q74, Q89, Q105, Q115 should be right for a candidate to pass… If it were questions that most people would have gotten right, it would not be a terribly tough one…
Sorry Joey. I’m slightly confused. The whole concept of a minumum passing score means that it is not possible to discriminate between which questions you did and did not answer correctly. The only way this could be done is if questions were given different weights. Assuming that each of the CFA questions is worth 1 mark and there are 240 marks in total - the concept of a minimum passing score will not differentiate candidates in the way you appear to imply.
Robert0s - I dont think that this will be used to differentiate candidates… like Joey said the numerical value of the no. of Q right over total questions will be the base number, on top of which the slicing and dicing will happen…
Except for the ethics part which I don’t know about, all that matters is the number of questions you got right. There just isn’t that much slicing and dicing that can be done. In particular, there aren’t that many people working on grading these exams for CFA Institute (a bunch for L III, but only a handful for L I - like they can all go out to dinner together).