SAT Score Trends: Five Things Enrollment Leaders Need to Know
SAT Score Trends: Five Things Enrollment Leaders Need to KnowCollege Board Communications
Recently, the College Board hosted a series of webinars that highlighted key trends in SAT scores, what these trends mean for the current application cycle, and what enrollment leaders can expect in 2018 and beyond.
Below are five of the most important takeaways from these webinars. Enrollment and admission staff who have more questions about this information can reach out to HEDreports@collegeboard.org at any time.
1. The Class of 2017 was the largest cohort in SAT history, in part due to more states and districts offering the SAT during the school day at no cost to their students.
The increasing number of states and districts offering the SAT during the school day at no cost to students (SAT School Day) means that we’re seeing an increase in SAT test takers with a greater range of academic preparation and college planning experience—this has the potential to impact how students are applying to college in terms of:
- Geography: Increased numbers of test takers in states and districts using SAT School Day may yield changes in student behavior that will impact colleges that recruit them.
- Volume and Timing of Score Sends: The first August administration of the SAT may impact where and when students send scores.
- Score Distributions: Percentiles and means reflect the composition of the SAT test-taking population. As that population changes, campuses need to thoughtfully research and consider how changes may impact reporting of SAT scores and setting policy scores.
Recent research shows that offering the SAT at no cost to students during the school day propels more students into college.
2. Overall, performance differences among subgroups on the new SAT are similar to those on the old SAT.
When we look at subgroup performance, generally we’re seeing similar performance differences by race/ethnicity and gender on the new SAT compared to what we’ve seen in prior cohorts for the old SAT.
We have done a series of analyses at both the administration and cohort level. It is important to note that making direct comparisons year over year is difficult due to not yet having a complete cohort that has taken the new SAT, as well as significant changes in the cohort composition and size, changes in the test, changes in the way we collect race/ethnicity data, and changes in score scales—but the key points are included below.
Performance differences by race and ethnicity are similar on the old and new SAT and across PSAT-related assessments. It is important to note that the way we collect race/ethnicity data changed in the 2015-16 school year, so we need to use caution in interpreting changes by race/ethnicity.
Performance differences on the SAT decreased slightly in nearly all race/ethnicity categories, except for American Indian/Alaskan Natives, where we saw a slight increase. We have two years’ worth of comparative data on performance by race/ethnicity on the PSAT-related assessments that show improved performance in both math and Evidence Based Reading and Writing for nearly all demographics and grade levels.
Gender performance differences decrease slightly across the board for the new SAT when we break down the data at the test-level. The graphs below depict the slight decrease in performance differences for males and females on the SAT Math section from the 2015 and 2017 cohorts.
Most colleges consider SAT scores for their applicants at the section level. Old SAT scores were reported based on three section scores—Math, Critical Reading, and Writing. Most colleges summed only Math and Critical Reading sections to determine a total score on a 1600 scale. Some colleges sum all three sections to determine a total score on a 2400 scale. These different processes mean that math is weighted differently depending on how a college sums the sections (i.e., 50% when two sections are summed vs. 33% when three sections are summed), resulting in greater or lesser differences in gender performance.
When we look at the old SAT on a 2400 total scale for the 2015 cohort compared to the new SAT for the 2017 cohort, we see that gender disparities increased slightly. When we look at the old SAT on a 1600 scale (old SAT Critical Reading + Math to new SAT total, we actually see a decrease in the disparities between men and women.
Every question on the SAT is researched and pre-tested with a representative sample of students to ensure it meets our rigorous test design and development standards. We examine performance among males and females and among race/ethnicity subgroups, and if we find any evidence of a disparity, the item is discarded and never used. We will continue to analyze test taking performance, participation, and behavior and will provide timely advice and assistance for using College Board data.
3. Changes in test-taking behaviors caused changes in score distribution during the 2016-17 admission cycle. For students in the class of 2018 who took the SAT on a Saturday, these trends appear to be returning to normal.
During the 2016-17 application season, we saw some expected changes in test taking behavior associated with the redesign of the SAT, which impacted some colleges’ applicant pools; some students chose to sit for the old SAT or take another college entrance exam. As a result, some colleges saw differences in the score patterns of their applicant pools, particularly at the top end of the score range, during the last application cycle.
Overall, we’re seeing these trends return to normal in terms of score distribution and participation. However it is important to remember that some colleges may still see some of these changes in the test-taking population impact the means and percentiles within their applicant pools for the class of 2018; this is our “new normal”.
4. Colleges should continue to use concordance tables to compare old SAT scores to new SAT scores during the transition.
Colleges should rely on the old SAT to new SAT concordance tables when making admission decisions.
Given the significant changes to the test-taking population due to SAT School Day, we re-weighted the 2015 concordance study sample to reflect the characteristics of the 2017 cohort who took the new SAT. There were no significant differences between the re-weighted table and the original concordance table.
We are working closely with ACT to update the SAT/ACT concordance; our joint study is underway and concordance tables will be released in the summer of 2018. The current ACT/new SAT concordance is derived from older concordances and is as accurate as the last direct concordance.
5. The new SAT is as predictive of college success as the old SAT.
Our pilot predictive validity research shows that the SAT continues to effectively predict success in college. Colleges can participate in the Admitted Class Evaluation Service™ (ACES™), a free online service that predicts how admitted students will perform at a particular institution. We’re enhancing the tool to make it easier to upload data, request research studies, and get more reporting functionality than before.