Most universities are not transparent with their data, reporting only the minimum information required of the Common Data Set. Incomplete data forces ambitious families and educators to find self-reported data or access Naviance scatterplots. Naviance is a tool at many high schools to help counselors and families navigate the college admissions process. Naviance “scatterplots” show the high school’s recorded history of admitted and rejected applicants to a given university based on GPA and test score.

Students plot their academics compared with their school’s previous applicants to estimate their chances. Scatterplots can be an excellent way to avoid the Lloyd fallacy if no student with your academics has been even close to gaining admission. However, most students will fall somewhere within the range of previously admitted applicants.

Looking for a signal in the Naviance noise will not help you make sense of the admissions madness or assess your chances. Because Naviance doesn’t break down admissions outcomes by major, it provides an inconsistent comparison. At universities such as UT-Austin, UC-Berkeley, or Carnegie Mellon (CMU), admissions rates for in-demand majors such as Engineering and Computer Science differ drastically to non-STEM programs. CMU admits only 7 percent of their computer science applicants compared with 31 percent to their music program.

Even when universities are transparent with their overall admission statistics, if they don’t provide information about admission to specific majors, especially for STEM, it’s of little help to applicants. There is increasingly a bifurcation within most prominent universities between high-demand STEM and business with non-STEM programs. STEM or Business at some universities is substantially more competitive than Communications or Liberal Arts. Computer Science applicants to Stanford or Princeton confront a steeper hill to climb than someone applying for History.

Some websites, Niche, for example, offer an “admissions calculator” that even breaks down admission by major, but the data is self-reported, allowing for, at best, incomplete comparisons. Relatedly, many applicants, despite taking AP Statistics, are insensitive to sample sizes. Niche’s self-reported data size for UCLA is around 30,000 applicants dating back a few years, but UCLA receives over 110,000 applications annually. Even the largest sample sizes will only ever capture a small percentage of total outcomes. Without a comprehensive data set, comparisons can never be accurate. Accessing complete data sets for most universities is impossible; therefore, dissecting scatterplots is an exercise in futility.

Examining scatterplots is like gamblers at the horse track who look for patterns in previous results to gain an edge. Acquiring and analyzing mountains of data gives the bettor an illusion of knowledge and control. Although a gambler might find some marginal advantages through sophisticated machine learning systems, few gamblers will reliably turn profits. Gamblers have access to many more known variables than college applicants. At least in horse racing, you can see the horses, jockeys, track conditions, and weather. In college admissions, you have access to almost nothing beyond your own application. College applicants are akin to gamblers throwing their money on the table without knowing the rules, odds, or who the “house” is.