While there absolutly is a US News stats component to it, I have been collecting the numbers from this year’s admissions cycles (so far only EA has been processed) and while it is unlikely you will be offered NU.in for certain high stat levels (only 3 NU.in acceptances with ACT over 33), NU.in decisions are very mixed together from other ranges below that on particular line. I’m going to be releasing a lot more synthesis later, but for now, I’d note that there are NU.in acceptances that have both higher ACT/GPA than over 1/3rd of the posted admits on CC that are also in the middle 50% for both GPA and ACT.
While there’s absolutely reporting bias from collecting data from CC, I’m not sure how much that bias would affect comparing NU.in vs traditional acceptances. It would absolutely show through in admit rate (which EA has at over 50% for example) and likely in the profiles of rejected students, but the acceptance data should be relatively balanced IMO.
Based on that, the data basically showed that if you are at a 32 ACT/4.5W GPA or lower and are accepted, it’s a 50/50 chance you get NU.in and that chance doesn’t actually increase with lower stat ranges. We’ll have to also see about RD but that’s held for EA. A 32ACT is under the middle 50% but 4.5W GPA is above the top 75%. So as one would expect, since US News only officially pays attention to test scores, NU.in acceptances are most often high GPA applicants with relatively lower test scores. Still though, there is no hard line based on ACT/GPA that defines an NU.in admit vs a Fall admit.
It’s also worth noting that the stats of NU.in admits are not staggeringly lower than fall admits - The average NU.in acceptance for EA was 31.3ACT/4.19W GPA compared to a CC reported fall acceptance profile of 33.75ACT/4.28W GPA. The closeness of that acceptance profile to that reported to US News seems to validate the lack of bias in this particular view of the data as well.
It’s worth noting that the profile for NU.in is pretty tight. On CC with about 200 data points there were 14 acceptances with ACT’s under 31. Of those only 6 were NU.in, making up less than 5% of the admitted student pool from CC. If you extrapolate the data above, here is a more accurate profile including NU.in (focusing on SAT over GPA given the variance in that data and the lesser correlation of it with NU.in acceptances:
Acceptance Rate:
- 19% acceptance rate from 2018
- 62K applicants this year
- assuming the same yield in both pools given lack of data
Estimate: 24.5%
ACT Range:
- Percentage of acceptances estimate: 25%ish = (1000ish/yield)/(3800/yield) = 1000ish/3800
- assume a 25% yield (about the number needed to make last years acceptance rate/applicant number work)
- Using approximate distribution of NU.in acceptances based on 2019 EA data
- 10% of top 50% and 40% of bottom 50% = 25% of total acceptances
- assuming somewhat even distribution of middle 50% given that no one here has that data
If 80% of NU.in students are bottom 50% (based on EA data that means 30-33.75) we add 3200 students to that bottom 50% and add 800 students to the top 50%.
New Top 50% number: 6400
New Bottom 50% number: 8800
New Denominator: 15200
The 4000 mark is about the 75th percentile, so that would put it 2/3rd’s the way through the top 50% (36-33.75). We know that distribution will be lower than even, so I think 34.5 is a fair guess. For the 25th percentile, we’re looking at 11,500ish for the line. That puts it right in the middle of the bottom 50% at about 31.9.
Adjusted Middle 50% Accepted ACT Range: 34.5-31.9
So, if we use that, let’s compare to the reported numbers:
Reported: 19%, 35-33
NU.in Adjusted: 24.5%, 34.5-31.9 (would be reported as 35-32 or 34-32)
After all this, the selectivity and student profile really doesn’t change much - certainly not to an almost doubled acceptance rate. The idea that NU.in is a hard line that drags down the middle 50% drastically really doesn’t check out. You have to remember that US News is a game of inches and while those profile differences may help with US News test score ranges, it doesn’t mean they have to be drastic differences.
This is all just a preview of future data to come though as I’m hoping to get a lot more interesting data once I add RD and look at factors such as major and if you apply for FA 
Note: I used ACT to standardize the data between SAT and ACT with the least loss/assumption of data, so it made sense to standardize to the least granular, ACT. There are also absolutely assumptions made in this data in an attempt to standardize many GPA formats and to estimate conclusions where insufficient data is present. While I had the data parsed, much of the calculations are ad-hoc and subject to rounding / error