The ACT and SAT are standardized tests that are hardly standard; it may behoove you to take one test over the other depending on your own skills and which schools you want to apply to. In today's Noodling, Randall Malcolm takes a look at the numbers and discusses which test is preferred by which school.
Normally, when we compare, we want to compare apples to apples and oranges to oranges. However, what happens when we want to compare apples to oranges? You standardize what you want to compare, or put more plainly, you convert the apples and the oranges into a common value.
Can we do this for the ACT and the SAT? And what will these comparisons tell us? Let's try and find out.
Step 1: Examining the data before conversation. Is a comparison viable?
Let's first focus on the SAT. Just counting the math and verbal sections, the SAT has a maximum score of 1600 and a minimum score of 400. The increment from one score to another is 10. That means that we have 121 possible scores.
The ACT has a maximum score of 36 and a minimum score of 1. The increment is one. This means that there are 36 possible scores on the ACT.
This is our first problem: The SAT offers nearly four times the number of scores than does the ACT. This means that the SAT gives you a more specific score.
Is there any way to equalize the number of scores on both tests? No.
So, before we even try to normalize the scores from these tests, we have to accept that the scores from these tests are not perfectly comparable? Yes, it's not one to one, but roughly one to four.
Also, 1,568,834 students took the ACT and 1,547,918 students took the SAT in 2010. So, in terms of the number of students who took the tests, it's a good comparison.
Step 2: How can we standardize SAT and ACT scores?
Standardizing means you change the scores into the same type of values. In our case, we can change the SAT and ACT scores into percentages. Collegeboard and the ACT give us the percentages that correspond to each score.
But isn't there another problem? The SAT attempts to maintain evenly distributed scores, or put more mathematically a bell curve. This means that they distribute the scores so that the number of people who score a 400 is the same as the number of people who score a 1600. The ACT doesn't attempt to distribute scores in this way and therefore doesn't have a bell curve.
In the charts below, you'll see that the SAT distribution looks like a bell. The ACT distribution looks a little lop-sided below 18. (The charts below are based on the 2010 graduates who took the SAT and ACT. The data is from the CollegeBoard and ACT, Inc.)
What this means for us is that there will be more error when we compare the data below 18 for the Act and below 1000 for the SAT.
What the numbers say:
The general (not so exciting) calculations:
When we subtracted the SAT percentiles from the ACT percentiles across 300 of the top colleges, we found that the ACT was generally higher. In the 25th percentile range the ACT was, on average, 7.75% higher than was the SAT. In the 75th percentile, the ACT was, on average, 2.175% higher than the SAT.
This makes sense. As we discussed, the distribution of the ACT is grouped more toward the higher scores. This means that the 25th percentile of the ACT falls around 16. This is pretty far from 8--what it'd be if it were evenly distributed.
The ACT is also higher in terms of the 75th percentile, but not to same degree. The 75th percentile on the ACT falls at about a 24, instead of a 28--what it'd be if it were evenly distributed.
What this means is that generally the schools judge the SAT and the ACT more or less equally. So, equal in fact, that the mathematical differences between how the scores are distributed are shown. So, that's good.
The (exciting) specific calculations:
But are there schools that don't follow the general patterns? Yes.
The schools that report a significantly positive ACT/SAT difference (expressed by being at least one standard deviation above the mean) are schools that judge the ACT more harshly, or put very plainly, are school for which you should take the SAT.
The schools that report a significantly lower ACT/SAT difference (expressed by being at least one standard deviation below the mean) are schools that judge the SAT more harshly, or, again, put plainly, are school for which you should take the ACT.
For which schools should you take the SAT? (based on the 75thpercentile data)
For which schools should you take the ACT? (based on the 75thpercentile data)
Important note: In these school specific calculations, we used the 75thpercentile. We did this because the 75th percentile is more reliable and is the percentile you should look at when applying to colleges. There are always exceptions in the admissions process--great athletes, artists, etc. These scores influence the averages. Therefore, to be safe, always look at the 75th percentile and make sure you're competitive on that level.
What does this mean in terms of your college admissions process?
If you're looking at schools that are not on these lists (with the exception of 1st tier schools--see Do Ivy league schools prefer the SAT? What the Numbers Say), then you should take the SAT or the ACT, whichever you prefer and perform better on. If you are looking at schools that fall onto one of these lists, then take the test the list indicates and increase your chances at admission.
In terms of how colleges interpret this data, it's important to note that the colleges understand that there's a difference between the tests. They also know that the tests aren't perfectly comparable. In fact the UC system has its own way of standardizing ACT and SAT scores. So, when they are looking at your test scores, they're faced with these same comparison difficulties. They know that these tests are far from perfect.
What does all this say about the SAT and the ACT in general?
Why do we use these tests if they aren't comparable and aren't perfect?
The easiest answer is that the colleges want a standard way of measurement. Each high school, each teacher, and each student is different. To that end, grades, essays, extra-curriculars do not offer a standard way of comparing students.
Why do we need a standard way of comparing?
Think about it like this, these tests are rulers. Each time they are used to measure, the same scale is used. They offer good comparisons among students.
But what do they measure? I mean the tests measure different things, right? They're different rulers. I don't get it.
The tests most accurately measure how you perform on standardized tests and how much test prep you received.
That doesn't seem like a good system? Aren't they supposed to be about how smart I am?
It's not a good system and they aren't about how smart you are. If a student can go up 300-400 points on the SAT with a tutor in six weeks, the test measures how well you take standardized tests. Let's be honest, not even the best tutor in the world can teach you that much in six weeks. Schools know this. Consequently, some schools have done away with standardized tests altogether, (here's 27 of them). Others have asked that the tests be revised.
And this takes us back to the first point made. Think about the schools to which you'd like to go and then think about the tests that you need to take. According to these preliminary decisions, make sure that you are as prepared as you can be for the test that will serve you best.
These tests do not test your intelligence, they test how well you take standardize tests. They are a hoops to jump through. Practice jumping through the hoops and you'll do just fine.