White Paper · Lacefield Pedagogical Framework · v1.0
For the mathematical goals most students are pursuing, sustained engagement predicts outcomes more reliably than raw cognitive ability. This paper reviews the evidence, engages the substantial counterevidence honestly, and derives what follows for instruction.
The claim that perseverance predicts mathematics achievement more reliably than IQ, for the goals most students pursue, requires careful scoping. Raw cognitive ability is a documented predictor of mathematical performance, and IQ differences are not educationally irrelevant. The claim is more precisely this: for the mathematical domains that constitute GED, secondary school, and early tertiary mathematics — not research mathematics or theoretical physics — individual differences in sustained engagement, persistence through difficulty, and tolerance for productive struggle are frequently larger predictors of outcomes than IQ differences within the range of students typically enrolled in those programs. Duckworth, Peterson, Matthews, and Kelly (2007) established that grit — perseverance and passion for long-term goals — predicted success incrementally over and beyond IQ in multiple samples. Credé, Tynan, and Harms (2017), in a meta-analysis of the grit literature, found smaller and more variable effects than the original research suggested. We engage both bodies of evidence, specify the scope conditions under which the perseverance claim is most and least defensible, and derive the instructional implication: students who disengage do not disengage because they cannot — they disengage because the conditions for sustained engagement have not been produced. Creating those conditions is a designable instructional problem.
The version of this claim worth defending is not "IQ doesn't matter" — it does, and the evidence is extensive. The version worth defending is narrower: within the range of mathematical goals most students are attempting, and within the range of cognitive ability most students bring to those goals, differences in sustained engagement typically have a larger effect on outcomes than the IQ differences that separate one student from another in the same program.
This matters because the attribution of mathematical failure to fixed cognitive ability — "I'm not a math person," "they're just not mathematically minded" — functions as a terminal explanation that forecloses instructional response. If failure is attributable to fixed capacity, there is nothing the teacher or student can do. If failure is attributable to disengagement that occurred for identifiable and addressable reasons, there is a great deal that can be done. The empirical question is which attribution is more accurate for the population of students who fail secondary and adult education mathematics.
"Students fail not because they lack intelligence. They fail because they become discouraged, lose confidence, stop engaging, and never develop momentum. The disengagement almost always precedes the failure — it is not a response to confirmed incapacity but a preemptive withdrawal from a situation experienced as threatening."
Duckworth, Peterson, Matthews, and Kelly (2007) introduced the construct of grit — defined as perseverance and passion for long-term goals — and tested its predictive validity across multiple samples and outcome measures. Key findings: grit predicted educational attainment in two adult samples (N = 1,545 and N = 690), GPA among Ivy League undergraduates (N = 138), retention across two classes of West Point cadets (N = 1,218 and N = 1,308), and ranking in the National Spelling Bee (N = 175). Grit demonstrated incremental predictive validity over and beyond IQ and conscientiousness across these samples — meaning it accounted for variance in outcomes that IQ alone did not explain.
The effect size is important to state precisely: grit accounted for an average of approximately 4% of the variance in success outcomes. This is a real and replicable effect, but it is modest in absolute terms. The finding is not that grit swamps IQ — it is that grit predicts outcomes independently of IQ, meaning effort and persistence contribute to success in ways that cognitive ability alone does not fully capture.
Grit showed incremental predictive validity over IQ and conscientiousness across 5 samples (total N > 5,000). Effect is real, replicable, and modest: 4% unique variance explained.
88 independent samples. Grit–performance correlation ρ = 0.18 after correcting for unreliability. Substantially smaller than early reports. Largely overlaps with conscientiousness.
n = 11,000+ West Point cadets over a decade. Cognitive ability predicts academic grades most strongly; grit and physical capacity predict military performance and retention. Domain specificity matters.
Credé, Tynan, and Harms (2017) conducted a comprehensive meta-analysis of the grit literature, synthesizing 88 independent samples. Their findings substantially qualify the original Duckworth et al. claims. The corrected grit–performance correlation was ρ = 0.18 — meaningful but considerably smaller than the effect sizes reported in the original research and much of the popular coverage. More significantly, grit showed substantial overlap with conscientiousness (one of the Big Five personality traits), raising the question of whether grit is a meaningfully distinct construct or primarily a repackaging of a well-established personality dimension.
This is genuine counterevidence that must be engaged directly. The Credé et al. meta-analysis does not establish that perseverance is irrelevant — a ρ = 0.18 correlation is a real effect — but it does establish that the effect is smaller and less independent of existing constructs than the original research implied. The honest position is that perseverance contributes to academic outcomes independently of cognitive ability, but by a modest amount, and that the original grit research overstated both the magnitude and the independence of this contribution.
The grit debate — whether it is a distinct construct, how large its effects are, whether it is trainable — is a research question that remains open. The instructionally relevant reframing sidesteps the grit construct entirely and focuses on a more proximal variable: sustained engagement with appropriately calibrated material.
This reframing is supported by a simpler and less contested observation: students who stop engaging with mathematical material stop developing mathematical competence, regardless of their cognitive ability. A student with high cognitive ability who disengages after three sessions learns less than a student with average cognitive ability who engages consistently over thirty sessions. This is not a claim about grit or personality — it is an observation about the relationship between time-on-task and learning, which is one of the most robust findings in educational research.
The instructional implication follows: the question is not whether a student has the cognitive ability to learn a given mathematical domain. For GED, secondary, and most tertiary mathematics, the answer to that question is almost always yes. The question is whether the instructional conditions are in place to sustain engagement long enough for competence to develop. That is a designable problem. The conditions that sustain engagement — appropriate challenge calibration, regular mastery experiences, responsive feedback, evidence of progress — are documented throughout this framework and are the primary targets of instructional design.
IQ predicts more strongly at the extremes of mathematical achievement. Research mathematics, theoretical physics, competitive mathematics olympiad performance — domains that require constructing genuinely novel proofs or developing original mathematical ideas — show much larger IQ effects than the applied mathematics domains most students are working toward. The claim that perseverance matters more than IQ is scoped to the domains where most students fail: GED, algebra, arithmetic, early calculus. It does not extend to the upper tail of mathematical performance.
IQ differences within a typical classroom are smaller than they appear. The Duckworth et al. (2007) finding that grit predicted outcomes at West Point — a highly cognitively selected environment — is particularly informative. When the range of cognitive ability is restricted (as it is in any program with meaningful admissions standards), noncognitive variables predict a larger share of variance. GED students are not a cognitively restricted population, but neither are they spread across the full IQ distribution. The IQ variance in a typical GED classroom may be smaller than it appears, making engagement differences more predictive of who succeeds.
Conscientiousness accounts for most of what grit accounts for. Per Credé et al. (2017), grit's predictive validity is largely shared with conscientiousness, which is a stable personality trait with a genetic component. This does not mean engagement cannot be increased by instructional design — conscientiousness as a trait does not preclude state-level engagement variation — but it does mean that treating perseverance as a purely teachable skill, trainable independently of personality, is not well-supported by the evidence.
Attribute failure to disengagement before attributing it to incapacity. When a student fails to make progress, the first diagnostic question is not "can they do this?" but "are they engaging?" Disengagement — from frustration, from eroded confidence, from material that is miscalibrated to their current level — accounts for a very large proportion of mathematics failure in adult and secondary education. Ruling it out before concluding incapacity is the empirically correct order of diagnosis. It is also worth identifying which specific events initiated the disengagement: a public cold-question that exposed a gap in front of the class, a valid method penalized as wrong, a correct reasoning chain marked as an error. The companion paper on incorrect correction (Lacefield, 2026h) documents the specific mechanisms through which single instructional events can initiate the disengagement cycle that then compounds into the pattern that looks like fixed incapacity.
Build momentum before adding challenge. A student who has disengaged needs early success experiences to reestablish the belief that effort leads somewhere. The instructional sequence for a returning or discouraged student begins below their failure level — where success is assured — and advances upward only as the evidence of competence accumulates. This is not lowering standards; it is rebuilding the engagement foundation without which higher standards cannot be approached.
Make progress visible. One of the most reliable disengagement triggers is the student's inability to perceive their own progress. Regular review of prior material the student has mastered, explicit comparison to earlier performance, and concrete markers of what is now achievable that was not achievable before all serve to make accumulated competence visible. Progress that the student cannot see does not sustain engagement.
Do not accept the fixed-ability attribution. When a student says "I'm just not a math person," the pedagogically and empirically appropriate response is not to argue — it is to provide evidence. The first session where a student solves problems they believed they could not solve is more persuasive than any amount of verbal reassurance. Design for that session to happen early.
The perseverance-over-IQ claim is true in a specific and bounded sense: for the mathematical domains most students are attempting, sustained engagement with appropriately calibrated material predicts outcomes that cognitive ability alone does not fully determine, and the conditions for that engagement are designable. It is not true in the stronger sense that IQ is irrelevant, that grit is a large and independent predictor of performance, or that persistence can substitute for cognitive capacity in domains requiring genuinely novel mathematical construction.
The Credé et al. (2017) meta-analysis is important counterevidence that should be acknowledged directly: the grit effect is smaller and less independent than the original research implied. What survives that counterevidence is the more proximal claim about engagement: students who disengage from mathematical material do not learn it, regardless of cognitive ability, and the conditions that sustain engagement are within the instructor's control. That is the instructionally actionable version of the claim, and it is well-supported.