I learned in science class that no two snowflakes are exactly the same, their uniqueness owed to the fact that tiny droplets in clouds cool and freeze at different rates. Determining whether a student understands this or any other level of science is where tests come into play and for students considering college, the standardized admissions test has an important role.

For college admissions officers, it can be difficult to gauge what a student has actually learned in their previous education and even harder to determine whether they have the right core of knowledge and reasoning skills to succeed in a university environment. Standardized tests, while only one of many criteria for college admission, level the playing field among students from diverse backgrounds and guard against grade inflation. But what if the test doesn’t serve that purpose or worse, skews the perception of a student’s ability?

Those are among the questions raised in a study of the ACT admissions test by the National Bureau of Economic Research (NBER). The NBER study, published in May, concluded that the science portion of the ACT is ill suited to determining whether high school graduates will succeed in finishing their college careers. In fact, the NBER researchers concluded that the science and reading portions of the ACT, “provide essentially no predictive power regarding college outcomes.”

According to the NBER report, much of the problem with the ACT as a reliable predictor of college success lies with how a student’s test scores are presented to admissions officers. There are four subject areas covered by the ACT; English and mathematics, science and reading.

But rather than reporting the individual scores from each of the four subject areas, all four scores are combined into a single composite score. Hypothetically, a student who answers all of the ACT science questions right but flunks the math section may well be judged the same as a student who does moderately well in all four subjects. By comparison, the SAT, another popular college admissions tests, provides three separate scores for that exam’s language, mathematics and writing sections.

Compounding this flaw is the ACT science test itself and what it identifies should make all parents of school-aged kids nervous. The ACT purports to assess a student’s understanding of biology, chemistry, physics and the Earth sciences, including astronomy, geology and meteorology, and claims to do so in 40 questions.

In fact, the ACT science section doesn’t test much science at all. For lack of a better term, it looks more like a comprehension test, featuring lengthy, science-based reading passages followed by multiple choice questions.  A few questions require actual knowledge of science beyond what is already laid out in the passage. But if you can read and understand what is right in front of you, you can score pretty well. Small wonder researchers found this test to have “essentially no predictive power” in measuring a student’s prospects for success on a university campus.

The ACT science test also stands as something of a contradiction. While the ACT largely promotes itself as a test that measures what a student has learned, the company’s website says, “The intent (of the science test) is to present students with a situation to engage their reasoning skills, rather than to invite their recall of a classroom activity.” What then is the ACT, a knowledge test or a reasoning test?

Facts matter and the ACT could be more forthright in how it portrays itself. If it claims to be a knowledge test, then it should test actual knowledge, whether it’s science or any other subject. As for the reporting of test scores, college admissions officials will be hamstrung by the ACT composite score and students will have to risk their college admission on a flawed test.