All a GPA indicates is the average grades that person got. But that is not the same thing as ability, skills, motivation, passion, etc. Most classes aren't really that similar to the actual work in the industry either, so it's not even a good indicator of probability of work success.
If this is for a hiring type decision, I think what projects a person has done is more important than their GPA.
GPA is useful when curated with the 'reputation' of a particular college or university. For example, "Person P scored 3.25 from XYZ college, which is more impressive than person Q who got 3.75 from ABC college". This reputation information is usually known to those who have interviewed several years worth of interns and entry-level hires.
It usually means that P from XYZ has the discipline to undergo the grind, so at the minimum one can expect some business value from that discipline. It doesn't say whether Q is incapable of the same discipline, just that P is a safer bet. Outlier E&OE.
You have no evidence for that, because there is no evidence for that.
Google tried to find indicators for success at Google. One quote from a NYT interview:
> One of the things we’ve seen from all our data crunching is that G.P.A.’s are worthless as a criteria for hiring, and test scores are worthless — no correlation at all except for brand-new college grads, where there’s a slight correlation. Google famously used to ask everyone for a transcript and G.P.A.’s and test scores, but we don’t anymore, unless you’re just a few years out of school. We found that they don’t predict anything.
> What’s interesting is the proportion of people without any college education at Google has increased over time as well. So we have teams where you have 14 percent of the team made up of people who’ve never gone to college.
>After two or three years, your ability to perform at Google is completely unrelated to how you performed when you were in school, because the skills you required in college are very different. You’re also fundamentally a different person. You learn and grow, you think about things differently.
> Another reason is that I think academic environments are artificial environments. People who succeed there are sort of finely trained, they’re conditioned to succeed in that environment. One of my own frustrations when I was in college and grad school is that you knew the professor was looking for a specific answer. You could figure that out, but it’s much more interesting to solve problems where there isn’t an obvious answer. You want people who like figuring out stuff where there is no obvious answer.
Even in your own scenario, would you prefer someone who graduated with a 3.25 and triple majored in physics, business, and history, or would you prefer someone with a 3.75 who majored in business?
The person who went to a 'high reputation' college is more likely to have had better peers, and so more likely to have learned more by thermalization.
Also, software jobs are unique - one can be effective in a software job without knowing much about computer science. Jobs such as designing transistors or aircraft require strong basics, of the type that is usually acquired by undergraduate rigor. I doubt if a situation would exist where 14% of aero engineers at Boeing or Airbus never went to college.
Where is your evidence that "a 3.25 GPA from XYZ college, is more impressive than person Q who got 3.75 from ABC college"?
As we all know now from Moneyball, long-held views can be wildly ungrounded in reality, so why should I believe "those who have interviewed several years worth of interns and entry-level hires" are any better at judging competence than, say, professional baseball scouts in the pre-Moneyball era choosing someone based on how sexy his girlfriend looked?
Or any better than orchestras before the introduction of blind auditions in the 1970s/1980s. Before then, female musicians were 5% of the top orchestras. By 1997 (see 'Orchestrating Impartiality: The Impact of "Blind" Auditions on Female Musicians') they were 25%.
Or any better than the hiring practices where the only difference in a resume was a "white" name vs. a "black" name. The resumes with a black name required more attempts to get a callback (see http://www.nber.org/digest/sep03/w9873.html )
Unfounded unconscious bias pervades everything. Why should I assume that there's no unfounded bias in your observation, given that all attempts at quantifying what you believe to be true have failed to find a long-term predictor of success?
Also, there are plenty of jobs where one doesn't need to know much about the underlying science in order to be successful. Programming is not unique in that regard. One can be an excellent chef without knowing "rigorous" food chemistry, an excellent athlete without a degree in sports medicine or biology, an excellent writer without a degree in English, an excellent musician without knowing much about pressure waves, an excellent CEO without a business degree or training in economics.
In a more civilized age, an analog hiring manager chose students based on some mysterious Force, mentored them - and they worked in that company for a long time, often on several generations of the same chip.
The scenario today is different. Hiring managers designing a chip product are in a prisoner's dilemma situation - they follow the 'hire from XYZ school' formula precisely because they don't want the risk of a "contrarian hire" to be found useless for the current chip tapeout.
The XYZ universities that have higher 'analog chip design reputation' also offer more immersion: star faculty, elective courses, exposure to tools and labs, sending project chip designs for fab in a foundry shuttle. An XYZ hire - even with lower GPA - can be put to productive use immediately, no learning curve with tools etc. The learning happened in school, talking with other students while doing projects.
Contrast with a student from ABC school that offers only one proforma analog course, has better grades but needs at least a quarter to ramp up. The short-term costs in this field of statistically proving the lack of a long-term correlation may be too high - no matter the quantitative evidence from other fields such as software.
I think I read a different interpretation of your original post than you suggested, but I think you have the wrong emphasis, to the point of distraction. You talked about the "'reputation' of a particular college or university" when you probably meant "'reputation' of a specific degree program for producing graduates with the training needed for a specialized field."
This is a much more constrained and much less actionable qualifier.
To express it in more words, I thought in your original statement you were saying something like "Harvard has a good overall reputation so a 3.25 from there is better than a 3.75 from a school with a (sufficiently) lower reputation, all else being equal."
Now you're saying something like "Missouri University of Science and Technology's EE program has a good reputation in vacuum power engineering, so someone with a 3.25 from that program is better than a 3.75 from the EE program at Harvard, all else being equal, if you need someone with vacuum power engineering skills."
I assumed you implied an "all else being equal", because it's meaningless to say that someone with a music degree from Stanford is a great hire when you're looking for someone to work with x-ray crystallography, just because Stanford has a great biophysics program with lots of world-famous x-ray crystallography work being done there.
No, I think you should be saying "if you have two candidates, both with EE degrees and similar coursework, then the reputation of their corresponding EE programs should also be factor, and not just GPA."
But your scenario just now has the student from XYZ (the better school) also taking more courses in the specific field of interest, while the person from ABC only took a single pro-forma course.
So I'll invert it. Let's say that student X from college ABC, which has a second quartile ranking as an EE program, graduates with triple major in EE (with a focus in analog chip design), physics, and English, with a 3.75 GPA.
While student A from college XYZ (which has the best analog chip design training reputation in the world), graduates with an English degree with a 3.25 GPA, and a EE minor, with a single analog circuits class. (Student A came in thinking to be a EE grad, then in sophomore year switched to English.)
Who would you rather hire - the student with a higher GPA and with more training and interest, but from a lesser reputation program, or the student with a lower GPA and not much interest or training, but from the school with the higher reputation program?
BTW, the hiring manager is not in a 'prisoner's dilemma situation'. That requires two people who can make a choice, where the payoff matrix depends on both choices, but there's only one here. The scenario you describe is much closer to the FUD of "nobody ever got fired for buying IBM equipment."
10 comments
[ 3.5 ms ] story [ 34.9 ms ] threadIf this is for a hiring type decision, I think what projects a person has done is more important than their GPA.
It usually means that P from XYZ has the discipline to undergo the grind, so at the minimum one can expect some business value from that discipline. It doesn't say whether Q is incapable of the same discipline, just that P is a safer bet. Outlier E&OE.
Google tried to find indicators for success at Google. One quote from a NYT interview:
> One of the things we’ve seen from all our data crunching is that G.P.A.’s are worthless as a criteria for hiring, and test scores are worthless — no correlation at all except for brand-new college grads, where there’s a slight correlation. Google famously used to ask everyone for a transcript and G.P.A.’s and test scores, but we don’t anymore, unless you’re just a few years out of school. We found that they don’t predict anything.
> What’s interesting is the proportion of people without any college education at Google has increased over time as well. So we have teams where you have 14 percent of the team made up of people who’ve never gone to college.
>After two or three years, your ability to perform at Google is completely unrelated to how you performed when you were in school, because the skills you required in college are very different. You’re also fundamentally a different person. You learn and grow, you think about things differently.
> Another reason is that I think academic environments are artificial environments. People who succeed there are sort of finely trained, they’re conditioned to succeed in that environment. One of my own frustrations when I was in college and grad school is that you knew the professor was looking for a specific answer. You could figure that out, but it’s much more interesting to solve problems where there isn’t an obvious answer. You want people who like figuring out stuff where there is no obvious answer.
Even in your own scenario, would you prefer someone who graduated with a 3.25 and triple majored in physics, business, and history, or would you prefer someone with a 3.75 who majored in business?
Also, software jobs are unique - one can be effective in a software job without knowing much about computer science. Jobs such as designing transistors or aircraft require strong basics, of the type that is usually acquired by undergraduate rigor. I doubt if a situation would exist where 14% of aero engineers at Boeing or Airbus never went to college.
Where is your evidence that "a 3.25 GPA from XYZ college, is more impressive than person Q who got 3.75 from ABC college"?
As we all know now from Moneyball, long-held views can be wildly ungrounded in reality, so why should I believe "those who have interviewed several years worth of interns and entry-level hires" are any better at judging competence than, say, professional baseball scouts in the pre-Moneyball era choosing someone based on how sexy his girlfriend looked?
Or any better than orchestras before the introduction of blind auditions in the 1970s/1980s. Before then, female musicians were 5% of the top orchestras. By 1997 (see 'Orchestrating Impartiality: The Impact of "Blind" Auditions on Female Musicians') they were 25%.
Or any better than the hiring practices where the only difference in a resume was a "white" name vs. a "black" name. The resumes with a black name required more attempts to get a callback (see http://www.nber.org/digest/sep03/w9873.html )
Unfounded unconscious bias pervades everything. Why should I assume that there's no unfounded bias in your observation, given that all attempts at quantifying what you believe to be true have failed to find a long-term predictor of success?
Also, there are plenty of jobs where one doesn't need to know much about the underlying science in order to be successful. Programming is not unique in that regard. One can be an excellent chef without knowing "rigorous" food chemistry, an excellent athlete without a degree in sports medicine or biology, an excellent writer without a degree in English, an excellent musician without knowing much about pressure waves, an excellent CEO without a business degree or training in economics.
In a more civilized age, an analog hiring manager chose students based on some mysterious Force, mentored them - and they worked in that company for a long time, often on several generations of the same chip.
The scenario today is different. Hiring managers designing a chip product are in a prisoner's dilemma situation - they follow the 'hire from XYZ school' formula precisely because they don't want the risk of a "contrarian hire" to be found useless for the current chip tapeout.
The XYZ universities that have higher 'analog chip design reputation' also offer more immersion: star faculty, elective courses, exposure to tools and labs, sending project chip designs for fab in a foundry shuttle. An XYZ hire - even with lower GPA - can be put to productive use immediately, no learning curve with tools etc. The learning happened in school, talking with other students while doing projects.
Contrast with a student from ABC school that offers only one proforma analog course, has better grades but needs at least a quarter to ramp up. The short-term costs in this field of statistically proving the lack of a long-term correlation may be too high - no matter the quantitative evidence from other fields such as software.
This is a much more constrained and much less actionable qualifier.
To express it in more words, I thought in your original statement you were saying something like "Harvard has a good overall reputation so a 3.25 from there is better than a 3.75 from a school with a (sufficiently) lower reputation, all else being equal."
Now you're saying something like "Missouri University of Science and Technology's EE program has a good reputation in vacuum power engineering, so someone with a 3.25 from that program is better than a 3.75 from the EE program at Harvard, all else being equal, if you need someone with vacuum power engineering skills."
I assumed you implied an "all else being equal", because it's meaningless to say that someone with a music degree from Stanford is a great hire when you're looking for someone to work with x-ray crystallography, just because Stanford has a great biophysics program with lots of world-famous x-ray crystallography work being done there.
No, I think you should be saying "if you have two candidates, both with EE degrees and similar coursework, then the reputation of their corresponding EE programs should also be factor, and not just GPA."
But your scenario just now has the student from XYZ (the better school) also taking more courses in the specific field of interest, while the person from ABC only took a single pro-forma course.
So I'll invert it. Let's say that student X from college ABC, which has a second quartile ranking as an EE program, graduates with triple major in EE (with a focus in analog chip design), physics, and English, with a 3.75 GPA.
While student A from college XYZ (which has the best analog chip design training reputation in the world), graduates with an English degree with a 3.25 GPA, and a EE minor, with a single analog circuits class. (Student A came in thinking to be a EE grad, then in sophomore year switched to English.)
Who would you rather hire - the student with a higher GPA and with more training and interest, but from a lesser reputation program, or the student with a lower GPA and not much interest or training, but from the school with the higher reputation program?
BTW, the hiring manager is not in a 'prisoner's dilemma situation'. That requires two people who can make a choice, where the payoff matrix depends on both choices, but there's only one here. The scenario you describe is much closer to the FUD of "nobody ever got fired for buying IBM equipment."
And you're right - the hiring manager isn't a case of prisoner's dilemma. Thanks for calling that out.
I've learnt several new things from this discussion, especially the references to scientific studies of selection bias. Thank you.