How much do GPA and GRE Scores Matter?

[Left Skew]

Hello all,

I'm back again with some more data. For those of you that are unfamiliar with my original post, I create a procedure in R that scrapes the results survey and output some pretty-ish graphics from the data. 

After a bit of cleaning and consolidating, I looked into things like: common interview days, most popular schools, and most popular program types...

The data is based on self-report and isn't always the cleanest, so I have't taken the initiative to make more profound "predictions".

But here I am, one last time, to give some info on the qualitative metrics behind an application (i.e., GPA, GRE Verbal, GRE Quant, GRE AWA) to see if there is really a difference in those that get accepted and those that get rejected. 

I split the data further by program type; however, some of the programs (e.g., Developmental Psych, Quant Psych, Neuro) had such a small sample of candidate that reported their scores that I had to aggregate them-- these programs are in "psychology: other". Once again this data is not parametric because it lacks independence of scores and possibly normalcy. A candidate may be in the rejected or accepted sample multiple times given they reported their scores for multiple schools, so be cautious when coming to conclusions. Also we have no clue how accurate reported scores are; candidates may have forgotten their score or flipped their quant and verbal (these are things I cannot account for).

Conclusion: I had to actually go back and look at the raw data because the results weren't as I expected. There are many occasions where those rejected had a higher average than those accepted. I will work on further validating my numbers but I can say that the average score between those rejected and those accepted (based on the Grad Cafe sample) don't seem to consistently differ. This trend is not consistent across programs.

Social Psych applicants seem like the smarties among the top program-types.

Another conclusion is that it seems the grad cafe sample is higher on these metrics than an average applicant (I need to do more research) but given the average GRE scores are definitely above the 50th percentile, GPA is also high too.

The data is on the last few pages. Those of you that haven't seen the other graphics please feel free. You can see it HERE.

I'm also going to attach the file for those of you that don't trust clicking links (though I promise it's safe).

 

 

Grad Cafe Decisions.pdf

[Clinapp2017]

Very interesting! I’d say your main conclusion makes a lot of sense given these metrics are likely just used to sceeen candidates out (or weed out those with poor marks and GREs). The difference between GRE 310/315+ is relatively neglible in terms of diminishing returns from what I understand from admissions. 

 

I’m curious about selection biases that drive people to disclose people to share their stats on grad cafe. That never has made sense to me since it isn’t that hard (in theory) to work backwards as a faulty monitoring this page to see an applicant disclosing the interview invites lol. 

[lewin]

I would be very hesitant to draw any conclusions about the relationship between GPA/GRE and admissions based on self-reported data.  I don't doubt the analysis, but selection bias is so impactful that response bias in reporting socially desirable traits is one of the opening examples from the classic book How to Lie With Statistics. It's also possible that gradcafe members differ from the applicant population as a whole in meaningful ways, or that we have a restricted range of scores. The studies I'm aware of (e.g., Posselt's book on gatekeeping in admissions) find that GRE/GPA are critically important when screening the first wave of applicants. For top programs, they sometimes do their first cuts using spreadsheet with just GPA/GRE/institution.

"Another conclusion is that it seems the grad cafe sample is higher on these metrics than an average applicant (I need to do more research) but given the average GRE scores are definitely above the 50th percentile, GPA is also high too.

I'm not sure that "average applicant" and "average GRE test-taker" are the same thing. e.g., If the bottom quartile of test-takers gives up on graduate school but the remaining 75% moves on to apply somewhere, and gradcafe's GRE scores are ~50th percentile overall, this would mean they're below the score of average applicant.