Hacker News Reader

Comment by timmg | Hacker News Reader

I, personally, grew up in a small, (lower-ish) middle-class mid-western town with below-average (I think!) schools. I did not get good grades (I just didn't care that much). But I got really good test scores. So I got into a pretty decent (state) school with a legit engineering program.

One could argue that my lack of grades is a good reason for me to not get into a good school. But without the tests, I definitely wouldn't have. I shutter to think how my career/life would be different had I not gotten into a decent school.

So, at least for me, I have a real affection for these kinds of tests. And, as far as I know, I think the US is like the only place that doesn't use tests as the primary decider for higher education.

It seems like the move away from testing was "not based on science". Though it was sometimes couched that way. I'm glad to see sanity (and science) prevail. But it all seems very fragile to me right now.

View on HN

I think standardized tests correlate very highly (perhaps highest of any measure) to performance. I regularly see schools in the US with a 10% math proficiency and 40% literacy proficiency with a 99% graduation rate. What's up with that?

Standardized tests are the most fair and simplest way to evaluate students. Plus gaming the test is what exactly? Hiring a tutor and spending hours studying, much of it alone? And it's not a shock that success on studying and doing well on exams correlates to college where you're evaluated on how well you study and do on exams

It reminds me of that Key and Peele sketch where they're planning a long con bank heist but by the end it's obvious they're just working a job and saving for retirement

https://youtu.be/jgYYOUC10aM?si=uiceINWHRRf3brZ5

You probably want to do some test prep but you can do that with a test prep book at your local library.

Years ago, in a grad course of some sort, we looked at some papers of predicting performance in school. Basically, even at the grad level, GMATs were far and away the best predictor of academic performance, followed by undergrad grades. Things like letters of recommendation were pretty worthless.

The argument is that some people have access to private tutors who "teach the test" and others don't, not "some people study harder than others". And there is absolutely truth to that. I had to work after school, to make ends meet. Ain't got no time for extra studying.

That doesn't mean we should do away with test scores altogether, but we do need to realize the limitations. These scores come with error bars and uncertainty, and it matters how you use them. "Everyone with a score higher than the requirement ends up in a lottery" tends to be reasonably fair, but "we take the top-200 scores" tends to be biased to the well-off (sometimes very biased).

We should advocate for social programs that allow high school kids “make ends meet” without impacting everyone’s academic acceptance criteria.

Harvard isn't some lottery that just pays out. It's a school. If you have to work after school and didn't have time to study for an exam, maybe you should go to a less competitive school. If you're not prepared enough, getting into a very challenging school isn't good. Its like I don't know how to read because I was taking care of my dying mother but now I want to get into a PhD program for literature. The PhD program wouldnt be doing me any favors by letting me in. Instead i would be frustrated, waste a lot of time and money and drop out eventually

I agree that there is likely no difference in performance between a 1500 and 1600 SAT so a lottery is reasonable. Although I will say it's pretty strongly correlated even at the extremes. But why should a school solely focus on maximizing gpa? It should minimize failure but choosing between a student who will get a 3.5 or a 3.75 is pointless

https://zeescorrelationstudy.weebly.com/

> If you have to work after school and didn't have time to study for an exam, maybe you should go to a less competitive school. If you're not prepared enough, getting into a very challenging school isn't good.

Getting extra tutoring, focused on improving your test results, does not make you more prepared for college.

It's not like your analogy.

> some people have access to private tutors who "teach the test" and others don't, not "some people study harder than others". And there is absolutely truth to that.

There is truth to the fact that some students have better access to study the test than others, but in my recollection, it really doesn't help those kids very much. I'm sorry I don't have any sources off hand, but I remember it being something like that all the extra prep will add maybe 10-30 points to the SAT.

To put another way: a dumb rich kid with private tutors will still score lower than a bright poor kid who takes the test with no prep.

I agree with your concern, but the problem is that Harvard will have to choose somehow, and most other ways of choosing are even more biased toward rich parents: some high school kids are writing conference papers and starting a venture, just imagine what kind of social background would make these kids even consider such an opportunity.

I don't know if there's any good answer.

> I think standardized tests correlate very highly (perhaps highest of any measure) to performance

This depends on what kind of performance you're trying to measure. There will be some people who test very highly but are much less inclined to persevere and follow-through. Grades can be a better measure of this (though with rampant grade inflation, their utility is undercut).

This is a tautological argument. Test scores correlate with performance as measured via tests. Many of the best computer programmers I know have zero formal eduction in the field. They would fail the tests beloved to Silicon Valley and yet have proven their worth time after time.

> Plus gaming the test is what exactly? Hiring a tutor and spending hours studying, much of it alone?

Only if you aren't in Asia. Cram schools are a thing in India, China, Japan, Korea...

Of course, willingness and perseverance to go an extreme in work ethic has to be related indicator of future success. But that's your competition if you are just hiring a tutor and spending some hours studying alone.

The only thing standardized tests measure very well in aggregate are parental income and educations level

Not requiring testing seems late stage ideological-cultural destruction plan to displace America, the values and guide rails that allowed the West to prosper, attempting to displace merit.

I did as little work as possible in high school. I was depressed due to unknown reasons at the time, diet playing a large role in that - as well as being high functioning autistic, Asperger's, and so in classes I cared about then I was passionate in - getting the only or one of the only 101% in a computer class that that high school had given anyone due to my quickly doing my assignments and then helping anyone else who needed help.

The teachers who saw me excel in their classes fought at the end of high school, my final year, when deciding what final grades to give - pitted against the teachers who saw the absolute least motivated and incompetent versions of me; they settled on giving me 80% honours - which allowed me better placement at top tier universities, technically 2 points above where my grades would have averaged otherwise - but where they had discretion.

It's time the pendulum begins to swing the other way again - and then for proper education to begin so that it limits or prevents it from swinging again as far as it has recently.

The most extreme other end of the spectrum is the South Korea model. I hope we find something that's equitable and fair.

I think a key will be merit-based resource distribution - which means not having a centralized pool funded by tax payers that then is a single point of failure, nor a single point of capture for wannabe tyrants, whereby still collect taxes for the purpose of education - however distributing it based on the parents' selecting what lineage of knowledge-skillsets they want their children exposed to or indoctrinated in.

Perhaps the greatest difficulty with this, for policy makers, is it's hard to quantify or account for externalized benefits. However the other end of the spectrum with the student loan scandal, giving loans to people for "education" that - has already occurred and so it's time to go back towards merit-based acceptance and distribution of resources - tied to an actual path with an ROI, and then where "liberal arts" can thrive once enough parents have enough to pay for their children to go into those programs - rather than society funding them, like we poorly-irrationally do in Canada.

The policy politicians will push for will depend on if their goal is to maximize centralized control and dependance on government or not, where if more and more "students" require state-funds for "acceptable" state accreddited education - where they must comply and be "good citizens" and follow your orders, mandates, if they want to continue to be able to suck on the teet of the centralized state control - while that same state may be (and is what's happening in Canada) has policy in place to dramatically increase and continue to increase the cost of living, reducing the quality of life - so where more and more people are going to become forced to depend on the government aka a communist state; or if the state as part of their strategy is also to accept foreign students, flooding the institutions with foreign students for their revenues, also as a way to bring in revenues from external sources, for better or worse.

“Give me just one generation of youth, and I'll transform the whole world.” ― Vladimir Lenin

"nor a single point of capture for wannabe tyrants"

This is a very good point. Americans rarely realize that single points of capture are magnets for wannabe tyrants. If you can't be a full dictator of the entire country for life, you may settle for being a full dictator of some massive institution that meddles in everyone's life.

To be clear, I don't think that Americans are stupid for not realizing this. They just don't have enough experience with tyranny.

I have a related story. I was admitted to the CS department in the late 90s before it got super popular and they had to enact stricter admissions requirements. In hindsight I likely wouldn't have made the cut if those standards were applied to me. Fast forward and I've been a successful software engineer for the past 20 years.

I've never even been asked in 10+ jobs over a 20+ year career where I went to school, and mine had a great engineering program (which I had terrible grades in). They might have noticed it at the bottom of my resume but never asked.

Some of the best programmers I've had the chance to work with had no degree, though I know that's an HR bar in itself to leap over.

It's pretty normal not to be asked but many people do look--not so much for negative signal but they may notice. I'm pretty sure it didn't make the difference with my last job but probably sealed the deal as my manager went to the same school.

It makes a big difference in your first job, but then only if it is a name brand school. My company regularly solicits internships from certain top schools. Its hard to imagine CS at my school had something similar (I was not a CS major).

I went to Chico State, and HP took interns and new hires out of there in droves.

That was rather a long time ago (like when HP existed), so I don't know what's happening now.

Maybe they didn't ask since it was on your resume.

Lol yeah it's one of the first things you read on a resume.

This has to vary. It was the last thing I read for anyone not just starting in the field. I wanted to know what they'd done, both in and out of work. I wanted to know if they were a driven, problem-solving go-getter. Degree and alma mater is a low-value data point in that regard.

I guess so, I like to start from the beginning and go through the timeline.

Isn't your personal experience just anecdotal, which could be the opposite of statistical evidence.

Your school and GPA are relevant for a vanishingly short time on a resume. Maybe 2 years post graduation at best. After that, it's all about the work experience. If you have work experience before graduation, then school and GPA are almost completely immaterial.

HS GPA factors into getting into college, and is irrelevant afterwards. College GPA factors into getting internships and the first job placement and is irrelevant afterwards. If an employer actually cared in any meaningful way about my uni or grades after 5 years of work experience, I would take it as a red flag during hiring.

I might be wrong, but: What has science got to say about who should be admitted to schools, or more broadly what the priority of civilization should be? I think you're trying to universalize some of y(our) internalized beliefs into axioms.

I agree, my test score enabled me to get into a much better school than expected. All I did was borrow a test practice book from the library, take a few practice exams improving my score each time, and getting an even better score on the real one.

Test scores are an excellent way to distinguish moderate students from strong ones. In my experience, when you're distinguishing top applicants from each other, much of the difference will come down to the intensity of test preparation. Of course, for these Ivy League schools, the overwhelming majority of applicants fall into this category. The desire to outperform drives a huge investment in test prep, and selects for students who have put all their time into that activity -- rather than actually being curious about the world.

Per a link re:MIT in these comments, at that point, the best "solution" (such as it is) is to say they have a good enough SAT and move on to other metrics that may not be as accurate but are perhaps more illuminating. Does it really matter if someone has a Math SAT of 750 vs. 800 or whatever the numbers are these days?

Indeed, at some point the metric becomes very subject to stochastic uncertainty.

IQ tests administered by psychologists admit that there is a cutoff beyond which they fail, and so presumably the same situation exists with standardized tests.

> rather than actually being curious about the world

Great phrase that allows anyone to make any kind of shit up to evaluate someone.

We can't just make the tests hard enough so preparations will have diminishing returns? When I look at the exams in elite universities like Tokyo University, JEEs, and etc, I'm not sure how preparation can help the students in different talent pools. An extreme example would be those selective STEM summer camps. For instance, students in SUMaC will pick a subject in number theory, abstract algebra, or algebraic topology to study for three weeks. I don't think everyone can even understand the topics no matter how hard they try.

It'd be incredible for "test prep" to have to mean "learn things" to result in better scores.

The simpler a rule is, the less likely it gets gamed. Simple as that. As for the standard tests, it baffles me why the US won't take a more rigorous approach like other countries do. At least add word problems, proofs, or in general problems that require some writing.

> And, as far as I know, I think the US is like the only place that doesn't use tests as the primary decider for higher education.

Yeah, it baffles me too. It's also hard for someone outside of the US to understand how even raising such question will be political and worse, racial. For god's sake, we don't need national or even state exams. Universities can simply design their own tests. Britain does that. Japan does that. Germany does that. East European countries do that. India does that. China does that (in addition to their national entrance exams).

How is a simple measure so divisive in the US? Oh well, what is not divisive in the US?

> The simpler a rule is, the less likely it gets gamed.

Why?

As a counterexample, measuring a programmer's productivity by lines of code written is almost as simple as a rule possibly can be, and is extraordinarily easy to game.

By and large, elite US universities are not interested in using testing to pick and choose among the most academically qualified applicants. Instead, they use testing primarily as a cutoff mechanism to distinguish applicants who can probably do the work and applicants that probably can't.

You may not agree, but the system seems to work pretty well.

Reply on news.ycombinator.com