Curving, and What It Means for You as a Student

If you're an AP student, you might have heard that some AP tests have a "good curve" while others have a "bad curve." But what do people mean when they talk about the curve? Let's take a look at what a curve is, and then look at examples of curves to explain this concept.

A curve is essentially changing the grade to form a certain distribution. Teachers may want to do this when the scores are too high or too low. This is to ensure fair grading between different sections of the same course.

Curving will make the scores among different students, teachers, and schools easier to compare. In turn, this will increase the score's legitimacy.

In the following two AP examples, we look at the scoring distribution of the pass rate, 5 rate, and the percent correct needed to pass or get a 5. The College Board gives the tests to college students ahead of time. Then, it uses these scores to set the curve.

The two AP Physics C tests are examples of the few tests with a "heavy" curve, as demonstrated by the scoring distributions. A large majority of test takers pass the exam every year, and almost 50% get a 5. Compared to other tests, the percent needed to get a passing score or to get a 5 is also relatively low, with around 50% correct needed to get a 5.

However, don't get fooled by this! This is not an easy test to get a 5. As I have learned from taking AP Physics C: Mechanics test this year, the questions are still very hard. Moreover, the FRQs can easily make matters worse.

Tests with a similar curve include the AP language tests and AP Calculus BC, which are also considered to be tests with hard content.

There is also another related reason for this curve. These classes self-select, meaning that most students who take it are already highly-qualified for the course. This is achieved through prerequisites or prior experience. Thus, these motivated students will be more likely to score higher. The "good" curve is there because the test is so hard that even the top-performing students perform poorly on a raw scale. Be thankful for the curve!

At the other end of the spectrum, there is the Physics 1 test. AP Physics 1 consistently has the lowest pass rates of all AP exams. Only about 5% of students get a 5, and less than 40% pass. A higher raw score is also needed to score well, with 70% or higher needed to earn a 5. Relative to tests like AP Physics C, the content is easier. This will result in highly-performing students getting a higher raw score.   

Tests with a similar "bad" curve include AP Psychology and AP Human Geography, which also have relatively easy content.

These tests don't have a self-selecting population. Instead, many students have these courses as their first AP classes. As a result, they are less likely to be prepared than more experienced AP test-takers. In conclusion, we can see that the easier the test, the harsher the curve will be. Inversely, the harder the test, the lighter the curve will be.

The SAT is another curved test. Unlike the AP tests, we compare the curves between SAT dates by seeing how many points a test-taker drops if they miss a single question in each section.

The College Board curves each test through a process called equating. Equating compares the difficulty between tests and adjusts the scores accordingly. Thus, two equal scores on the same tests have the same meaning. The College Board does a similar process for the SAT Subject Tests as well.

The May 2017 SAT was one of the better curves in recent history. According to CollegePanda, a -1 in Reading resulted in a 40 Reading score, a -1 resulted in a 39 Writing score, and a -1 resulted in a 790 Math score.

This was a very good curve because missing 1 question each on Writing and Math usually drops your score more than the May 2017 curve did. However, the good curve meant that the test was harder. Students lost less for missing a question that might have been harder compared to other tests' questions.

On the other hand, there is the infamous June 2018 SAT. This test's curve was reportedly so bad that dissatisfied students started a petition for a recurve. As a student who took this very SAT, I know how harsh the curve was for this test (I got 1540 with a 540 EBRW score and an 800 math score).

The harsh Math curve resulted in a -1 giving a 770. The EBRW did not receive any slack either, as missing one question each on Reading and Writing led to a 39 and 37, respectively. These point deductions were much harsher than those of other tests, leading to high-performing students scoring much lower than they expected.

This example serves as a reminder that, in general, you don't want an easy SAT. Getting relatively easy questions wrong could cause you to lose more points than expected.

----

Curving is very different in a classroom setting compared to curving on a standardized test. Teachers are concerned about their own classes, and standardization for millions of students is unnecessary.

To curve in a classroom situation, teachers normally pool the class scores and order the grades from lowest to highest. After that, there are two main ways that they curve:

The first way is the simpler way, which requires teachers to find how many points it would take for the highest score to become 100%. Then, they add that amount of points to everyone's score. For example, if the highest score on a 100 point test was 80 points, then that 80 would become 100, and 20 points would be added to everyone's score. This method is mostly used in high schools.

The other way to curve is more complicated. Teachers find the distribution of the class scores, then they determine the percentages of students to give each letter grade. In a 10-20-40-20-10 curve, the top 10% of the class would receive an A, the next 20% would receive a B, the next 40% would receive a C, the next 20% would receive a D, and the last 10% would receive an F. As opposed to the first method, this method is commonly used in colleges and universities. 

We are going to look at three scenarios here.

The first scenario is where a class' scores are evenly distributed from 0-100% with the mean at 50%. There would be no need to curve, as the score distributions are already even. Curving would not affect the distribution or scores.

The second case is where all scores are high, with the average score being around 80-90%. Curving would be bad for students because even if the lowest students scored around 70%, which is normally a C, they would be curved down to an F.

The final scenario is the opposite of the previous case, where the class average is much lower than 50%. Now, curving would be in the students' favor. Even though the whole class failed the test, a majority of the class would still pass after the curve.