Why marks don't predict success (and what the evidence does support)

The student who topped Class 12 at your school ten years ago. Where are they now? Most schools do not know, which is itself the interesting fact: the institution that measured them most carefully stopped measuring at the exact point the measurement would have been testable.

The claim that marks do not predict success is usually made badly, in a motivational register, by people who did not do well in exams. It can be made precisely instead, and the precise version is more useful because it says what marks do predict and for how long.

What the evidence actually says

The standard reference is a 1996 meta-analysis by Roth, BeVier, Switzer and Schippmann in the Journal of Applied Psychology. It exists because employers and academics had been disagreeing about this for decades on the strength of small studies.

Two things follow, and they point in opposite directions, which is why this gets argued about.

First: grades are not noise. A corrected correlation in the .30s is a real relationship, and anyone telling a fifteen-year-old that marks are meaningless is misinforming them.

Second: the relationship decays, and it decays fast. By the six-year mark the observed association is close to nothing. Marks are a good predictor of the next stage and a poor predictor of the stage after that.

That pattern makes sense once you look at what a mark measures. Performance on structured, timed, externally specified tasks with a known syllabus. The first job after graduation still looks a little like that. The eighth year of a career does not.

The Google anecdote, correctly attributed

The story everyone repeats is real but is routinely inflated into a study. What happened is that Laszlo Bock, then running people operations at Google, told the New York Times in 2013 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,” and that Google “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.”

The second half of that first sentence is usually cut off in the retelling, which is a pity, because the exception Bock makes is the same one the research finds: the signal is real for people fresh out of study and gone shortly after.

That is one large employer's internal analysis, described in an interview, about a specific and unusual applicant pool. It is consistent with Roth et al. and it is not independent evidence of the same weight. Worth quoting; not worth dressing up as research.

Measurable things that are not marks

If the strongest academic signal explains a modest share of later performance and fades, the obvious question is what else is measurable at school age.

Conscientiousness. The most durable finding in this literature. Barrick and Mount's 1991 meta-analysis of the Big Five found conscientiousness to be the one dimension showing consistent relations with all job performance criteria across all occupational groups they examined. Other dimensions predicted well in particular occupations (extraversion for roles built on social interaction, for example) but conscientiousness held everywhere.

Marks capture some of this, which is part of why they predict anything at all. But a 90% earned by three weeks of cramming and a 90% earned by nine months of steady work are identical on the marksheet and different underneath. The first student meets trouble in any setting where the content cannot be compressed into a fortnight.

Interest fit. Not “I find science interesting” but a measured profile across interest dimensions. Interest is a weaker predictor of raw performance than ability, and a better one of persistence and satisfaction. Someone can finish an engineering degree on discipline alone with their interests pointing elsewhere. Thirty years of that is competence without contentment, which is a poor thing to design a school system around.

Specific cognitive strengths, not a single number. Verbal, numerical, spatial and abstract reasoning are separable, and the marksheet blends them. Two students on 82% can have opposite profiles, which matters enormously at stream selection and is completely invisible in the document the family is using to decide. That decision, in detail.

Where marks genuinely earn their place

This is not an anti-marks argument, and the anti-marks version of it is lazy.

Marks give a common standard across schools that have every reason to flatter their own students. They give families a rough progress signal. In a country where the number of candidates vastly exceeds the number of seats, they provide a sorting mechanism that scales, and no one arguing against them has proposed a workable replacement at that scale.

The problem is not that marks exist. It is that marks are the only thing measured, and a single measurement then gets treated as a verdict on a person.

Judge a cricketer on batting average alone and you have a real number that tells you nothing about fielding, temperament, or what they do when the team is losing. An average of 45 is excellent and is not a career.

What a school can actually do

Board exams are not going away and arguing with them is unproductive. What a school can do is add measurements rather than remove one.

1. Measure more than the subject total. A percentage is a blend. Per-concept results from the same papers you already set tell you which parts of the blend are strong. How that works.
2. Run a real psychometric assessment, not a quiz. The difference is reliability, and it is not a matter of taste. What separates the two.
3. Report both to families together. Marks show academic performance; the profile shows where the strengths point. Presented separately they compete. Presented together they are a decision-making basis.
4. Be honest about the strength of the evidence. Including with parents. A corrected correlation in the .30s is a real finding and not a promise about their child. Anyone selling certainty about a fifteen-year-old's future is selling something.

The goal is not to choose between marks and everything else. Marks open the next door, and the rest of the profile helps decide which door is worth opening. Most Indian students currently get one of the two.