Grades

We use a unique set of linked administrative data sets to explore the determinants of persistence and academic success in university. The explanatory power of high school grades greatly dominates that of other variables such as university program, gender, and neighbourhood and high school characteristics. Indeed, high school and neighbourhood characteristics, such as average standardized test scores for a high school or average neighbourhood income, have weak links with success in university.

That is the abstract of the paper based on Ontario data "Persistence and Academic Success in University" by Dooley, Payne, and Robb (February 2012). 

The authors offer some policy implications:

First, there is a positive message regarding the educational system in Ontario. Students in our sample from disadvantaged neighborhoods and high schools with weaker performance on standardized tests are as well prepared for university as students with the same individual high school grades but from advantaged neighbourhoods and higher performing high schools. Viewed from the university perspective, the positive message is that these four institutions provide an environment in which students with similar high school grades but otherwise heterogeneous backgrounds have very similar likelihoods of success.

I am not surprised. Probably I expected that gender, "being female," would have a more important explanatory power.  

How generalizable are these results? My hunch is that they are, even in developing countries. Take a student with high high-school-grades from a rural area, take her to one of the top universities in a country. Odds are she will do very well. Of course predicting success at University might be more complex, but high school grades seem to be a reasonably good rule of thumb.