<p>I’m sure somewhere in the vast reaches of CC there’s a great discussion of what “safety” and “match” (and “safe match”, etc.) mean. I don’t know where it is. But I think people sometimes get confused by the standard data about admitted students’ SATs, GPAs, class rank, etc., as well as the percentage of applications accepted. </p>
<p>College A and College B may both have a 25%-75% SAT range of 2150-1900, 80% of entering students in the top 20% of their HS class and 50% in the top 10%, an average GPA of 3.6, and a 40% admit rate. But that does not make them both safeties or even matches for a student with 2200 SATs, 3.8 GPA, and top 10%. One college may essentially admit everyone with a clean rap sheet that clears a certain set of statistical hurdles, and then pick and choose among applicants in the next band down; it could be a safety for an applicant who clears those hurdles. The other may pick and choose applicants from a much wider statistical band, and may also have characteristics that limit the number of applications received (thus reducing selectivity). At the first college, a student with those stats may have a 95% chance of acceptance; at the second, a student with those stats, or better, may still only have a 50% chance of admission.</p>
<p>You can see that really clearly on the scattergrams that float around here. On some of them, a simple regression would give you a set of SAT/GPA functions defining a 90%, 75%, 50%, etc. chance of admission. On others – like HYPS – the best you could do would be to define an area with essentially NO statistical chance of admission, but even applicants with “perfect” numbers are being rejected in droves. Then there are schools where a meaningful percentage of applicants are accepted, but it’s clear that fewer numbers-based decisions are being made compared to the first group. The band at which 50% of the applicants are accepted may be relatively generous, but there may not be ANY band where 90% are accepted. I suspect schools like Wellesley and Barnard are in that category.</p>