<p>What does it mean when a college class has a "curve?" I heard that means only a certain number of students will get A's, B's, etc. Isn't that unfair? Why the hell would colleges do that? But if they do, I kinda understand why it would be so competitive at top schools like Berkeley.</p>
<p>They curve grades so that if the class average was a 60, the smartest person, who, let's say, got an 80, gets an A. Or vice versa with a high class average. It's the colleges' way of trying to make it so that people with a difficult professor don't get screwed with their grade.</p>
<p>This reminds me of a class I had in high school that was graded on the curve. Guess what? I had both the valedictorian and salutorian in that class...so the curve didn't even exist!</p>
<p>People confuse a curve with a scale. It's impossible for a curve to not exist.</p>
<p>A scale would be a class in which the top grade is an 80% and that gets made to be an 100% and thus everyone's grade would move up 20%. Thus in a class in which someone some one gets a 95, and everyone else only gets a 75, people get angry...In an intro sociology course freshman year which was going to be scaled, I managed to earn a 55/50 on the first test (the Grad student teaching the class not only was going to scale it, but also gave extra credit on the exam). Needless to say, people were pretty upset, and the prof ended up using the next highest person's grade as the scaled score. I was told that I wouldn't get any extra points...</p>
<p>A true curve is based on a standard bell curve in which the class average is set to some pre-determined letter grade. Usually a C+ or a B- depending on the prof. A true curve in which statistical analysis is done to determine the standard deviations and the Z-score for each student will assign grades based on how many standard deviations away from the mean you are. In a class of 200, that would mean that about 5 people will get A's and 5 will get F's. As an undergrad the few courses I had that were on curves didn't do a full statistical analysis, but the profs looked for natural breaks in the groupings of students and assinged grades starting from the top on down. It's pretty much the same thing though not as precise.</p>
<p>I see curves of the first type Bigredmed described all the time, even in high school.</p>
<p>Well I know there's definetely no curve at my high school, and I go to a private school</p>
<p>During the course of all of my schooling, I never took a class which had a curve. </p>
<p>I do not know if that was a good or bad thing. </p>
<p>One time, I did totally run out of time taking a History test. I guess the prof felt sorry for me, because I got a +5 points added to the test grade. Maybe that is because I go balls to the wall while taking tests? I do not know.</p>
<p>So, sometimes curves do not exist. I am living proof of never having taken a class which had a curve. The normal grading scale is better and easier.</p>
<p>True bell is pretty rare, at least where I go. One class I had with a pretty intensive lab component had the lab grades on a true bell curve normalized so the average was 75% and the standard deviation was 10%, and another class, the results were roughly true-bell-shaped, but it wasn't deliberately calculated in that way.</p>
<p>More frequently, the grades are calculated so that the average or median are a B-, with whichever is lower chosen to be the B-. That means that the number of A's and B's is roughly equal to the number of C's, D's, and F's (with relatively few, like 5%, getting F's)</p>
<p>
[quote]
What does it mean when a college class has a "curve?" I heard that means only a certain number of students will get A's, B's, etc. Isn't that unfair? Why the hell would colleges do that? But if they do, I kinda understand why it would be so competitive at top schools like Berkeley.
[/quote]
</p>
<p>Curves in college usually mean a bell-like distribution of grades, with Cs being the most common grade, and the frequency of grades decreases as the letter grade approaches A+ or F.</p>
<p>In layman's terms, the grade distribution looks something like this:</p>
<p>10% As
20% Bs
40% Cs
20% Ds
10% Fs</p>
<p>(note: actual bell curves at most universities have more As and Bs, and fewer Cs, Ds, Fs, than in the example above)</p>
<p>This is used, from what I've seen, mostly so the professor has less pressure in creating tests that are fair to the class. So, if the professor slips up and makes a test that is too hard (i.e. top students gets 40%...and that happens), or a test that is too easy (i.e. 93% is the class average...that happens too) then at least the grades are still evenly distributed and students aren't hurt too much by the former example.</p>
<p>So in effect, the curve sometimes makes it MORE fair to even out some tests that are too difficult for the class.</p>
<p>It is unfair in that there is some randomness or bias to the strength of students in your class. What I mean is, you could just happen to be stuck with several really brilliant students in one class, effectively killing off your chances of getting an A. Or, in competitive courses, the environment can turn ugly quick. But these two cases are still in the minority.</p>
<p>And yes, curves are pretty common at Berkeley, and probably other similar schools. They are less common at many private schools, as jbusc pointed out.</p>
<p>When my classes are curved, they're usually about half A/B and half C/D with just a few F's. It also depends on the department; pre-med and science grading is more strict, but liberal arts is sometimes curved to higher grades. </p>
<p>Some classes don't have any curves, but the tests are generally designed to have an average in the low B area. Larger classes are much more likely to be curved and smaller classes are often based on individual performance.</p>
<p>so that means that if i got say, a 92 on the midterm and the average was a 69.5, i'id be in good shape for an A for the quarter ya think?</p>
<p>^^^ Yes, you should have an A unless the standard deviation is really screwed up.</p>
<p>A 92 is already a raw A, and it looks like if he/she wants to make a general curve to 69.5 = 75, then that percent my be boosted further (assuming he/she does not use a graduated curve, where those at the top recieve no boost, but those in the bottom and middle recieve moderate boosts).</p>
<p>A 92 isn't always a raw A. Look at this Physics Midterm's distribution. Class average of 83, Std Deviation of 15. 92 is not a raw "A" here.</p>
<p>Sometimes utes overthink these things. I was a teaching assistant when I was in grad school and participated in LOTS of grading. First, few classes have enough students in them to represent anything like a statistically valid population (I didn't TA the classes with 200 students). Even in larger classes, the grade distribution tended to be lumpy and not nicely bell-shaped. You can see that an exam that combined a big essay question, some short-answer questions, and perhaps a multiple choice section that you would get patterns -- maybe everyone had trouble with the essay for example, except some students who aced it.</p>
<p>What we did, as another poster suggested, is look at the final distribution. If a group of students was clustered through the scale, it's pretty easy to distribute grades. What would not be "fair" would be to have 10 students with an 89 and 10 with a 90 and say "those are Bs and those are As." Is there any real difference in understanding of the material between the 89s and the 90s? I don't think so, and that is why we tried to put the grade lines in empty or thin spots in the distributions.</p>
<p>Class grades rarely follow an approximately normal distribution. More often it's clearly a mixture of two or three populations.</p>
<p>/it's all about location-scale families, using order statistics and probit is just wrong and evil...</p>
<p>What I'm not entirely sure about is why are engineering and sciences classes graded on such harsh curves while the humanities courses if graded on curves at all, are graded to B+ or A- range and then if not curved than the average grade in the class comes out to a B+/A- anyway. Why is it that Engineering students absolutely must have lower grades? Did I miss the memo?</p>
<p>I could write a book on the strange aspects of "grading." When I was setting grades with the professor I worked for, we would look at the score clumps and then pick grade dividers. If half the students in the class were closely clumped at the top of the range, they would probably all get "A"s. If only 5% of the class was in the top clump and then there was a big gap to the next bunch, then only 5% of the class got "A"s. If you assume that the test was a valid indicator of learning (another good topic), then it's usually pretty obvious who "gets it." If everyone in class "gets it" and masters the material, I can't see any reason everyone shouldn't get an "A." Curving scores just for the sake of creating winners and losers is -- stupid. I'd give teachers like that an "F."</p>
<p>curves are a godsend...atleast here where no one gets about a 40 on a test</p>