ACT Scoring System
The broadest indicator of overall ACT performance is the total or composite score, which is reported from 1 to 36. The ACT composite score is the average of the four multiple-choice ACT section scores (one each for the English, Math, Reading, and Science tests), given on the same scale of 1 to 36. Each of these four tests includes multiple reporting category scores, which are rendered in raw score format (x correct answers out of y total category questions) and as percentages (e.g. 6 out of 8 correct answers is 75%). The 1-36 scale is the basis for test-takers' STEM and ELA scores. The STEM score is the average of the Math and Science test scores, while the ELA score is the average of the English, Reading, and Writing scores.
Because these sections of the ACT are entirely multiple-choice questions, scoring is done electronically. Total section scores appear at the top of score reports, and below each section score is a graphic that compares the student's performance to ACT's College Readiness Benchmark for that section. The English and Science tests have three reporting categories each, and the Math test has eight. The Reading test has three standard reporting categories along with a fourth area, Understanding Complex Texts, for which there are only three possible outcomes: below, proficient, and above. For all standard reporting categories, score reports include numerical representations of student performance and graphics that indicate whether or not the student met the ACT's Readiness Range for that reporting category.
The Writing test is scored by at least two trained ACT readers, each of whom evaluates four areas on a scale of 1 to 6. Students therefore receive four Writing test subscores of 2-12 and a total Writing test score that is the average of these four subscores (also 2-12). There are no benchmarks or readiness ranges associated with the ACT Writing test, but student Writing test performance is ranked at the national and state levels.
ACT score reports include percentile rankings that compare the student's scores with "recent high school graduates" in the test-taker's home state and across the country (ACT does not specify exactly what constitutes a "recent high school graduate," but ACT scoring data is typically reported in sets of three testing years). Students receive national- and state-level percentiles for eight areas: Composite, Math, Science, STEM, English, Reading, Writing, and ELA.
The latest statistical summary of all students who have taken the ACT shows the following averages: English (20.3), Math (20.8), Reading (21.3), Science (20.8), Writing (7.1), STEM (21.0), ELA (21.3), and Composite (20.9). The ELA and Writing scores indicated above are based on about 2.83 million test-takers in a single year. All other scores were calculated from approximately 5.86 million ACT-tested students over three years.
Composite ACT scores of 30 should be considered the absolute minimum for students wishing to gain acceptance to the most selective American universities. At Dartmouth College, 75% of new students scored at least 30, with a middle-50% range of 30-34. The same ranges for students at Harvard (32-35) and the Massachusetts Institute of Technology (33-35) were even higher. ACT standards at the so-called "public ivies" (the best flagship state universities) were not much lower, such as the University of Virginia (middle 50% of 29-33), the University of California-Berkeley (29-34), and the University of Michigan (29-33).
Students taking the ACT are asked questions about their college plans when they register for the test, one of which concerns their intended college majors. This allows ACT to connect interest in academic subjects with attainment of ACT College Readiness Benchmarks. The intended majors with the highest percentages of students meeting all four sectional ACT benchmarks (English, Math, Reading, and Science) include chemical engineering (64% of test-takers declaring this major met all four benchmarks), aerospace/aeronautical engineering (58%), biochemistry and biophysics (55%), general finance (54%), and computer science and programming (50%). The lowest percentages of benchmark attainment were associated with students claiming interest in medical assisting (10%), criminology (14%), athletic training (15%), and nursing (15%).
The ACT scoring system provides students and educators with a large amount of information. With composite scores, section scores, reporting category scores, percentile rankings, and benchmark indicators, college admissions officials are able to thoroughly review their applicants in terms of general and specific academic skills. Students learn their academic strengths and weaknesses, which can help them make informed choices about which fields of study to pursue. The ACT scoring system also contributes to educational policy. In many states, the ACT is given to all high school students, which allows policymakers to closely track student performance in the subjects tested. The complex scoring system helps administrators identify areas in need of improvement and evaluate the success or failure of policy initiatives. Though there are legitimate criticisms of the ACT and standardized testing, the ACT scoring system is helpful to many aspects of secondary and postsecondary education.
Independent researchers have examined how well ACT scores predict college grades. A 2015 study conducted at the Northwest Missouri State University, for instance, used 200 randomly selected students to examine correlations between high school grades, ACT scores, and college grades. The researcher concluded that while both high school GPA and ACT scores were linked to college GPA, high school grades were a more reliable predictor of performance in college courses (correlation coefficient of 0.639 versus 0.383 for ACT scores). ACT has admitted that this type of research does appear to show that high school GPA more accurately predicts college GPA than ACT scores. ACT nonetheless claims that test scores have "incremental usefulness beyond high school GPA," and in "situations involving high selectivity and high academic performance," test scores are more accurate than high school GPA in predicting grades. In the final analysis, both sides are probably at least partially correct, and postsecondary institutions benefit from as much relevant information about their applicants as possible.
Bias in standardized testing is a well-documented issue. ACT data show that white and Asian American students meet the ACT College Readiness Benchmarks much more frequently than other groups. White test-takers achieve the ACT English and Reading benchmarks at rates of 74% and 58% respectively, while for African Americans, the numbers are 33% (English benchmark) and 20% (Reading benchmark). Among Asian American test-takers, 70% meet the benchmark in Math and 58% in science, compared to respective rates of 26% and 22% for Hispanic students. ACT has also examined the impact of "underserved criteria" on testing performance (these criteria include low-income households, low levels of family education, and minority status). The statistics show that 81% of students who met all three underserved criteria achieved either zero or one ACT College Readiness Benchmarks, compared to 32% of students to whom no underserved criteria applied (68% of non-underserved test-takers met at least two ACT benchmarks).
Mean ACT scores at the state level show significant differences in overall performance, ranging from a low of 17.8 (the average in Nevada) to a high of 25.5 (New Hampshire). However, 100% of high school students in Nevada take the ACT, while in New Hampshire, the figure is just 18%. This suggests an easy (though partial) explanation for the discrepancy: Because the ACT is not required in New Hampshire, students who choose to take the test are almost certainly college-bound and are thus a highly motivated and self-selected group. Among the states that administer the ACT to 90% or more of their students, the highest average scores were received by test-takers in Minnesota (21.5), Illinois (21.4), and Colorado (20.8). In addition to Nevada, the lowest-performing such states were Mississippi (18.6), South Carolina (18.7), and Hawaii (19.0). It should be noted that the states with the lowest ACT testing rates (and generally the highest scores) are mostly clustered in the Northeast and on the Pacific Coast, including California, New York (both at 31%), and Maine (the lowest in the country at 8%).
Before September 2016, the ACT Writing test was scored on the same 1-36 scale as every other section. The change to the current 2-12 scale was likely motivated by vast discrepancies between Writing scores and composite scores, as well as substantial score changes for students who requested rescoring of their essays. An official ACT statement released in June of 2016 cited "confusion among students" and "perceptual problems" as some of the reasons for the implementation of the new Writing test scoring scale.