Evidence-Based Multiplication Mastery for a 9-Year-Old Homeschooler: A Science-of-Learning Synthesis

August 26, 2026

TL;DR

The strongest evidence supports a two-stage approach: first build understanding of what multiplication is (equal groups, arrays, area models) and teach reasoning strategies anchored on 2s/5s/10s/squares; then drive remaining facts to automatic recall via short, daily, spaced retrieval practice (flashcards, incremental rehearsal, simple games), with low-stakes fluency checks. This is the consensus across the IES “Assisting Students Struggling with Mathematics” practice guide (2021), NCTM, cognitive scientists like Willingham and Dehaene, and math educators Bay-Williams & Kling.
The “fluency vs. understanding” war is a false dichotomy. Pure rote drill without conceptual grounding produces brittle knowledge; pure conceptual work without retrieval practice fails to build the automaticity that frees working memory for later math (fractions, long multiplication, algebra). Roughly 10–15 minutes a day, 4–5 days a week, for 8–16 weeks is the realistic dose for most 9-year-olds.
For a homeschool parent, the most defensible plan is: (1) teach strategy-based fact derivation using a curriculum like Singapore Math, RightStart, Math with Confidence, or Bay-Williams & Kling’s Math Fact Fluency; (2) practice via Leitner-box flashcards or incremental rehearsal, plus a few well-chosen dice/card games (no high-pressure timed tests); (3) optionally add a research-validated app such as Reflex; (4) monitor for math anxiety and signs of dyscalculia, but expect that most children get there with consistency and patience, not novelty.

Key Findings

1. What the science of learning actually says about arithmetic facts

Automaticity is not optional — it’s the mechanism by which higher math becomes possible. Working-memory capacity is severely limited (about 4 chunks). Cognitive load theory (Sweller) and decades of mathematical cognition research (Geary, Ashcraft, Willingham) converge on the point that if a child has to calculate 7×8 in the middle of a long-multiplication or fraction problem, the working-memory cost is enormous and accuracy plummets. The 2008 National Math Advisory Panel concluded that “the achievement of automaticity, that is, the fast, implicit, and automatic retrieval of a fact or a procedure from long-term memory” is essential. Children who remain on counting/derived-strategy phases on multidigit problems show error rates driven by fact inaccuracy, not algorithm failure (Cumming & Elkins; Geary).

Retrieval practice beats restudy. A 2023 classroom experiment (Ophuis-Cox et al., Applied Cognitive Psychology) directly compared flashcard retrieval against chanting tables aloud in second-grade classrooms. Flashcards produced significantly better short-term and one-week-delayed recall — the act of trying to remember an answer is what builds the long-term memory trace (the “testing effect,” Roediger & Karpicke). A large meta-analytic literature (Adesope et al. 2017; Agarwal et al. 2021; Carpenter, Pan & Butler 2022 in Nature Reviews Psychology) confirms this is one of the most robust findings in cognitive science.

Spacing trumps massing. Rohrer & Taylor’s mathematics studies, and Cepeda et al.’s broader meta-analyses, show that the same total practice produces dramatically better long-term retention when distributed across days rather than crammed. For multiplication facts, an expanding-interval schedule (review today, then in 2 days, then 5 days, then 2 weeks) is supported as well as or better than fixed intervals.

Interleaving helps for problem-type discrimination. Rohrer, Dedrick & Stershic (2015) and Rohrer’s 2020 cluster-randomized trial in Journal of Educational Psychology showed that mixing problem types (e.g., ×, ÷, +) within a worksheet produced roughly double or triple the test scores of blocked practice on delayed tests. For learning a single fact set (e.g., the 4s), some blocking is necessary; for consolidating once facts are introduced, mixing helps.

Siegler’s overlapping waves model is empirically supported for multiplication. Van der Ven, Boom, Kroesbergen & Leseman (2012, Journal of Experimental Child Psychology) tracked 98 children weekly through multiplication acquisition and confirmed Siegler’s prediction that children use multiple strategies simultaneously — counting, repeated addition, derived facts, retrieval — with the share of retrieval growing gradually over weeks. Children with weaker working memory progressed more slowly because they couldn’t form the necessary fact associations. The implication: expect a messy, overlapping progression rather than a clean “stage 1 → stage 2 → stage 3.”

Math anxiety is real and consequential — but the timed-test claim is contested. Ashcraft’s program of research (e.g., Ashcraft & Krause 2007, Psychonomic Bulletin & Review) shows that math anxiety hijacks working-memory resources and harms timed performance specifically. Jo Boaler (“Fluency Without Fear,” 2015; Teaching Children Mathematics 2014) has popularized the claim that timed tests cause math anxiety for ~one-third of students. Critics, notably Greg Ashman, have shown that the original references for that specific quantitative claim are weak — Boaler’s cited sources are largely her own classroom observations, not controlled studies. The defensible synthesis: (a) math anxiety does interfere with performance via working memory; (b) high-stakes, public, speed-based testing is plausibly a contributor for some children; (c) for many children, brief untimed retrieval practice or low-stakes self-timing for personal goal-setting is fine and even motivating; (d) the homeschool environment is naturally low-stakes, so the worst conditions almost never apply.

2. The developmental and pedagogical sequence

There is strong consensus across NCTM (Bay-Williams & Kling 2014, 2019), the IES practice guide, and curricula like Singapore Math, RightStart, and Math with Confidence on the order of acquisition. Kling & Bay-Williams’s three-phase model (counting → reasoning strategies → mastery) is the dominant framework:

Phase 1 — Foundational facts (build understanding + memorize):

2s (anchored in doubles facts from addition)
10s (anchored in place value)
5s (anchored in skip counting and clock face)
0s and 1s (anchored in identity and zero properties)
Squares (2×2, 3×3, … 10×10) — these become powerful anchors for derived strategies and are often easier to remember than expected

Together these account for roughly half of the 100 single-digit multiplication facts.

Phase 2 — Derived-fact strategies: Use known foundational facts to reason out the harder ones. The key strategies (NCTM, May 2015):

Adding/subtracting a group: 6×7 = 5×7 + 7 = 42; 9×6 = 10×6 − 6 = 54
Doubling: 4s as double-2s; 6s as double-3s; 8s as double-4s
Halving and doubling: 6×8 → 3×16 = 48
Near-squares: 7×8 = 7×7 + 7 = 56; 6×7 = 6×6 + 6 = 42

The remaining “hard facts” (4×6, 4×7, 4×8, 6×7, 6×8, 7×8, 7×9) are precisely where derived strategies pay the most. Burns, Ysseldyke, Nelson & Kanive (2015, School Psychology Quarterly) analyzed data from 15,402 third- through fifth-graders and found empirically that facts involving 4s, 5s, 6s, and 7s required significantly more repetitions to master than 2s and 3s.

Phase 3 — Mastery / automaticity: Once a child can produce a fact in ≤3 seconds reliably, either by direct recall or a very efficient derived strategy, it counts as fluent. Bay-Williams and Kling explicitly define fluency as accuracy + efficiency + flexibility, not raw speed.

Concrete → Representational → Abstract (CRA). A 2025 meta-analysis of 30 studies (Ebner et al., Remedial and Special Education) found a very large overall effect (Tau-BC ≈ 0.99) for CRA instruction in math, including multiplication. Milton et al. (2019) demonstrated specifically that alternating CRA instruction improved both fluency and conceptual understanding of multiplication facts in students with learning difficulties. The practical sequence: use physical objects (counters, base-ten blocks, arrays of cubes) → pictures (dot arrays, area models, number lines) → symbols. Many homeschool curricula (Singapore, RightStart, Math-U-See, Math with Confidence) build CRA directly into their lessons.

Number sense matters. Stanislas Dehaene’s The Number Sense argues — and brain-imaging evidence supports — that humans do not have a hard-wired multiplication module; multiplication tables are stored as a linguistic, rote-learned set associated with language areas of the left hemisphere. But strong number sense (an intuitive grasp of magnitudes, number relationships, and place value) is the foundation on which derived-fact strategies and error-checking rest. This is why all major homeschool programs invest substantial time in number sense before pushing multiplication to automaticity.

3. Specific evidence-based techniques

Flashcards / Leitner-box retrieval practice. Strongest single technique by current evidence. The Leitner system (multiple boxes; cards move forward if recalled correctly, back if not) is a paper-based spaced-repetition algorithm. Apps like Anki implement the same idea digitally. For a 9-year-old, Leitner with physical cards is usually more engaging, less screen-heavy, and equally effective.

Incremental rehearsal (Burns 2005 and replications). A specific, well-validated drill structure: introduce one unknown fact embedded among ~9 known facts; rehearse the sequence; when the new fact is mastered, drop the oldest “known” and introduce another unknown. A meta-analysis (Burns, Zaslofsky, Kanive & Parker 2012) and replication studies (Codding et al. 2010; Burns et al. 2019) report 2–3× higher retention than traditional drill. It is highly feasible at home: requires only a deck of index cards and 5–10 minutes a day.

Timed drills and fluency-building activities. The IES 2021 practice guide explicitly recommends “regularly including timed activities as one way to build fluency in mathematics” — a position notably at odds with the strong anti-timed-test rhetoric from some math-education writers. The reconciling view: time pressure helps build automaticity (because it pushes children off slow counting onto direct retrieval), but the framing matters enormously. Use timed activities for self-improvement and personal-best benchmarks, not for comparison, ranking, or grades. Codding, Burns & Lukito’s 2011 meta-analysis found that drill-and-practice with modeling produced the largest effect sizes among basic-fact interventions.

Games (board, dice, card). A favorite of math-education researchers (Bay-Williams & Kling devote ~60 games to this in Math Fact Fluency). Evidence base is moderate rather than overwhelming — most studies are small or quasi-experimental — but games provide naturally spaced and interleaved retrieval practice, reduce anxiety, and sustain engagement. Examples that double as practice: Multiplication War (deck of cards, each flips two cards, higher product wins both), Multiplication Bingo, Salute! (three players; two see each other’s cards and shout the product, the third deduces their own card), Pig with multiplication (roll three dice, multiply two and add the third), Damult Dice.

Songs, chants, mnemonics. Some evidence of benefit, more for engagement and engagement-driven persistence than for direct memory advantage. Mark Ellis (Cal State Fullerton) and Mastropieri & Scruggs’s mnemonic instruction studies report meaningful gains, particularly for students with learning disabilities. But the 2023 Dutch flashcard-vs-chanting study found chanting clearly inferior to retrieval practice. The reasonable role for songs: as an entry point and skip-counting backup, not as the main vehicle to automaticity. Times Tales is a specific mnemonic system that converts each “hard fact” into a tiny story (e.g., 7×8 → Mrs. Week and Mrs. Snowman). Marketing claims “research-proven” but the company’s “statistical analysis report” is internal and small-scale; independent peer-reviewed evidence is sparse. The broader mnemonic literature (Greene 1999 in LDR&P; Nelson, Burns, Kanive & Ysseldyke 2013) does support mnemonic instruction for struggling learners, though typically slightly less than well-implemented rehearsal-based methods.

Visual/spatial methods (arrays, area models, dot patterns). Strongly supported as part of the conceptual phase. Boaler’s “Fluency Without Fear” emphasizes that connecting visual and symbolic representations activates multiple brain networks and deepens learning — a claim consistent with Dehaene’s neuroscience. Arrays of dots or square tiles make properties like commutativity (3×4 = 4×3) and distributivity (6×7 = 5×7 + 7) literally visible.

Strategy-based instruction. One of the more striking results in the literature: studies (Carr, Taasoobshirazi, Stroud & Royer 2011 in Contemporary Educational Psychology; Brendefur et al. 2015) report that students taught derived-fact strategies outperformed pure drill-practice peers, sometimes substantially — one cited study reported roughly 7× more facts gained per minute of strategy-based intervention versus pure repetition. Bay-Williams and Kling argue (with broad NCTM support) that skipping the strategy phase produces brittle, easily-forgotten memorization.

Apps and digital tools.

Reflex Math (ExploreLearning) has the strongest independent evidence base among consumer apps. Multiple randomized and quasi-experimental studies and an ESSA Tier 3 evidence certification from Digital Promise support meaningful fluency gains. Fact-family approach, adaptive, engaging — moderate cost.
Anki (free, customizable) implements optimal spaced repetition but requires parent setup. Better for older or self-motivated children.
XtraMath (free) — simple, evidence is modest but generally positive for fact recall in short daily doses.
Imagine Math Facts / Times Attack — a more game-like product; Sheridan et al. (2018) found it effective for third-grade multiplication fact fluency.
Multiplication.com / Times Tales–style mnemonic apps — entertaining and may help individual children, but evidence is mostly testimonial.

The 2024 Burns et al. meta-analysis of technology-based math-fact practice found generally positive effects, though smaller than well-implemented in-person rehearsal methods.

Physical/embodied learning. Some evidence (especially from researchers like Manches and from work in embodied cognition) suggests that movement-based and manipulative-based practice supports early acquisition, particularly for the foundational facts and conceptual phase. Finger gnosia — children’s awareness and use of their fingers — is linked in some neuroscience studies to mathematical development. But by age 9, persistent reliance on finger-counting for multiplication facts is a warning sign that automaticity has not yet developed (see Caveats).

4. Major curricula and approaches — what the evidence says

No major homeschool math curriculum has been rigorously tested against alternatives in randomized trials with large samples. Curriculum reviews are largely based on alignment with the practices that research does support. Here is the consensus view among informed homeschool educators (Cathy Duffy, Kate Snow, Well-Trained Mind) and how each aligns with the research base:

Program	Multiplication approach	Research alignment	Best for
Singapore Math (Primary Mathematics, Dimensions Math)	Concrete-Pictorial-Abstract (CPA, the same as CRA); bar modeling; emphasizes mental math; multiplication concept introduced early but mastery built gradually	Excellent — CPA is the most strongly-supported instructional sequence; emphasizes derived strategies and number sense	Conceptually-oriented kids; parents willing to teach actively; needs supplementation with explicit fact practice
RightStart Math	Abacus-based (AL Abacus); strong visualization of quantities; many games (over 200); derives multiplication from doubling/grouping	Excellent — explicitly games-based and conceptual, builds strong number sense	Hands-on learners; younger kids especially; parent-intensive (a real cost)
Beast Academy (Art of Problem Solving)	Comic-book format; problem-solving emphasis; assumes underlying fluency rather than building it explicitly	Good for problem-solving and conceptual depth, weaker for fact-fluency building per se	Mathematically-strong, motivated kids who already have basic facts; needs supplementary fact practice
Math-U-See	Mastery approach (one operation per year-level); Demme manipulative blocks; video lessons; whole year on multiplication at “Gamma” level	Solid CRA foundation; some critics note insufficient mixed/spiral practice	Visual/kinesthetic learners; parents who want video instruction; kids who need one thing at a time
Saxon Math	Spiral/incremental; daily fact practice built in; lots of repetition; less conceptual depth in early grades	Good fit with spaced-practice principles and the IES recommendation for systematic instruction with cumulative review; weaker on conceptual depth	Kids who thrive on routine; parents who want hands-off; less ideal if conceptual understanding is the gap
Math with Confidence (Kate Snow)	CPA-based; scripted; integrates strategy-based fact teaching; well-paced spiral review	Strong alignment with current research; closely modeled on Singapore + Bay-Williams approach	Parents who want a research-aligned, easy-to-teach option
Math Facts That Stick / Multiplication Facts That Stick (Kate Snow)	10-week strategy-then-practice supplement; games and short worksheets; designed to plug into any curriculum	Excellent — implements derived-strategy phase + spaced retrieval	Parents who want a focused, parent-friendly supplement to whatever main curriculum they use
Times Tales	Mnemonic stories for hard facts only	Limited independent evidence; mnemonic instruction does have a research base for struggling learners and students with LDs	Children who haven’t responded to other approaches; particularly children with dyslexia or weak rote-memory
Touch Math	Touch-points on numerals; sequential counting	Evidence base is concentrated in students with special needs (autism, intellectual disabilities) for addition; weaker for multiplication and not really intended to build multiplication-fact automaticity	Children with significant learning disabilities; not a recommended general approach
Reflex Math	Adaptive fact-family software; games-based; 15–20 min daily	Strongest independent evidence among apps (ESSA Tier 3)	Supplement, especially for children who like screens; about $35/year for home use
Rocket Math, XtraMath, Reflex, Anki	Various spaced/timed retrieval approaches	Generally evidence-supported, especially for the spacing and retrieval mechanisms themselves	Different child temperaments — pick by fit
Multiplication.com / similar mnemonic sites	Picture-mnemonic associations	Some support from the broader mnemonic literature; not rigorously tested as products	Children who haven’t gelled with other methods

There is no curriculum that has been shown to be uniquely best. The research-supported principles (CRA, derived strategies, spaced retrieval practice, low-pressure fluency building, mixed/interleaved review) matter more than the brand name.

5. Practical implementation for a homeschool parent

How much daily practice? The research and clinical recommendations cluster tightly: 10–15 minutes a day of dedicated fact practice, 4–5 days per week, is optimal. Longer sessions show sharply diminishing returns; shorter ones often work too. Distributed practice (every day) is dramatically more efficient than massed (twice a week for an hour). Burns et al. (2015) found that low-skill learners need 4–5 repetitions per fact to retain; higher-skill learners need only 2–3. Most 9-year-olds will reach full multiplication-fact mastery in 8–16 weeks of consistent daily practice if the strategy phase is done well first.

Timeline expectation. Conceptual introduction and 2s/5s/10s: roughly 2–4 weeks. Derived-strategy phase (3s, 4s, 6s, 9s, squares): 3–6 weeks. Hard-fact consolidation (7s, 8s, 6×7, 7×8, etc.): 3–6 weeks. Maintenance and mixed review: ongoing, ideally for the rest of the school year.

Detecting and avoiding math anxiety. Watch for: physical signs (clenched body, withdrawn affect, complaints of stomach ache before math); avoidance behaviors; sudden drop in apparent ability under speed pressure when ability is fine in untimed work; negative self-talk (“I’m dumb at math”). The single most effective preventive: emphasize that struggle and mistakes are how the brain grows (growth mindset framing, supported by Blackwell, Trzesniewski & Dweck 2007 — though it’s worth noting that recent meta-analyses such as Sisk et al. 2018 and Macnamara & Burgoyne 2023 show much weaker average effects for growth-mindset interventions than the popular literature suggests; the effect appears strongest for struggling and low-SES students). Avoid framing speed as the marker of math ability. Use stopwatches as personal-best tools, not comparison tools. If a timed activity provokes distress, drop it and use untimed retrieval instead.

Signs of fluency vs. still-counting strategies.

Fluent: answer within ~3 seconds, no visible finger movement, no whispered counting; child can answer in different orders, can articulate the inverse (24÷6=4), can use the fact inside larger problems.
Counting still: lip movement, finger movement, eyes drift up-and-to-the-side as if visualizing a number line, latency >5 seconds, can do the fact but can’t reverse it, errors of “off by one” (the signature of skip counting).

Formative assessment ideas (from Kling & Bay-Williams 2014, Teaching Children Mathematics):

Interview a fact: ask the child to solve and explain their thinking on a single problem.
Three-card sort: child sorts a small set of fact cards into “I just know this,” “I can figure it out fast,” and “I have to count” piles. Excellent diagnostic, reveals progress over weeks.
Untimed accuracy check: 20 problems mixed across all tables; look at both accuracy and time, but track them separately.
Application probes: does fluency transfer to area problems, word problems, multi-digit calculations?

Common homeschool pitfalls:

Skipping the strategy phase and going straight to rote drill. Produces brittle knowledge that erodes over the summer.
Treating all 100 facts as equally hard. They aren’t. Front-load the easy ones; use them as anchors.
Trying to memorize 4×6 the same way as 3×4. By 4×6, derived strategies are usually more efficient than rote.
Switching curricula in panic. Most failures come from inconsistent implementation, not from the wrong program. Pick one, stick with it for at least a full year.
Using only one method. The research supports a combination: conceptual modeling (arrays, area models), strategy teaching, spaced retrieval, brief fluency timing, and games.
Too much volume, too little spacing. Forty problems on one Saturday is much worse than 8 problems daily.
Confusing “memorization” with “automaticity.” You want the latter: a fast, virtually attention-free retrieval, ideally undergirded by understanding that lets the child reconstruct or check the answer if memory fails.
Letting math become identity-laden. Avoid “you’re so smart at math” praise; instead praise effort, strategy choice, and the act of trying hard problems (Dweck-style process praise).

6. Individual differences

Dyscalculia affects roughly 3–7% of children, comparable to dyslexia in prevalence but far less recognized. By age 9, watch for: persistent finger-counting for simple multiplication facts that don’t stick after substantial practice; weak number sense (can’t compare quantities; can’t immediately recognize that 8 is bigger than 5); inability to subitize small quantities; confusion about the meaning of operations; reversal of digits well past first grade; trouble with telling time, money, and estimation; and disproportionate gap between math and other subjects. If these are present despite consistent, well-structured practice, seek an evaluation by an educational psychologist who specifically screens for math learning disabilities (some screening tools: Number Sense Screener, TEMA-3, KeyMath-3). Early identification matters because targeted intervention works far better than waiting (Cleveland Clinic; Child Mind Institute; Understood.org).

ADHD complicates fact mastery primarily because attention to repetitive practice is hard. Solutions: shorter sessions (5 minutes × 3 a day rather than 15 in one block); games rather than worksheets; movement-based practice; clear timers and reward structures. Many children with ADHD do master facts with appropriate accommodations.

Math anxiety (Ashcraft 2002; Ashcraft & Krause 2007) is distinct from a learning disability — it’s a chronic stress response that compromises working memory specifically during math tasks. It often coexists with otherwise-normal math ability. Treat it like any anxiety: gradual exposure, low-stakes practice, control over pace, positive framing, and if severe, professional support.

Growth mindset and motivation. Dweck’s mindset work, popularized for math by Boaler, holds that children who believe ability can grow with effort persist longer and learn more. The strongest evidence is for targeted interventions with struggling students; broad universal mindset interventions show smaller effects than originally claimed (Sisk et al. 2018 meta-analysis). For homeschool: praise process and strategy (“you tried three different ways to figure that out”), normalize confusion and mistakes (“your brain is growing when this feels hard”), and avoid identity statements (“you’re just not a math person”). Boaler’s Stanford “Youcubed” site has free family-friendly materials grounded in this perspective.

Details: A Concrete Plan and Actionable Recommendations for a 9-Year-Old

Stage 1 — Concept solidified (1–2 weeks). Before any fact memorization, confirm the child can:

Show 6×4 as 6 equal groups of 4, as a 6×4 rectangular array, and as 4+4+4+4+4+4.
Read a 12×12 multiplication chart, find a fact in two directions (commutativity), and explain why 3×4 = 4×3 using an array.
Skip-count by 2s, 5s, and 10s smoothly to at least 100.
Use base-ten blocks or square tiles to build small products physically. Use Singapore Primary Math 2B/3A, Math with Confidence 3, RightStart C/D, or Math-U-See Gamma’s early lessons here.

Stage 2 — Foundational facts (2–4 weeks).

Teach 2s (link to addition doubles), 10s (place value), 5s (clock face + skip counting), then 0s and 1s.
Add squares (2×2, 3×3, … 10×10) as a high-leverage set of anchors.
10–15 minutes a day: ~5 minutes of game-based or array-based introduction, ~5 minutes Leitner-box flashcard retrieval, ~5 minutes mixed review of previously learned facts.

Stage 3 — Derived facts (3–6 weeks).

Teach 3s as “double plus one more group” (3×7 = 2×7 + 7). Teach 4s as “double the doubles” (4×7 = double 2×7). Teach 6s as “5s plus one more group” (6×7 = 5×7 + 7). Teach 9s as “10s minus one group” (9×7 = 10×7 − 7).
Explicitly script the child’s strategy aloud at first, then encourage internalization.
Keep the Leitner flashcard practice rolling; cards with derived facts initially get the “I figure this out fast” pile.

Stage 4 — Hard facts and consolidation (3–6 weeks).

Focus on 7×8, 6×7, 6×8, 7×9, 8×8, 8×9 — typically the last sticking points.
Use a combination of: derived strategies (7×8 = 7×7 + 7), a mnemonic for facts that resist (Times Tales is reasonable here), incremental rehearsal with cards, and games like Multiplication War limited to face cards 6–10.
Begin interleaved review: each session mixes facts from all phases.
Optional: 10 minutes/day on Reflex or Anki to consolidate.

Stage 5 — Maintenance (ongoing through the year).

5–10 minutes of mixed-fact retrieval, 3 times a week.
Apply facts in real contexts: long multiplication, area problems, fractions (which will start appearing immediately), money, cooking measurements.

Tools to gather:

A multiplication chart (printable, on the wall).
~120 index cards for a Leitner system (or pre-printed flashcards).
Square tiles or base-ten blocks for arrays.
Two ordinary six-sided dice, a 0–9 ten-sided die if possible, and a standard playing-card deck.
One scripted resource — Multiplication Facts That Stick by Kate Snow, or Bay-Williams & Kling’s Math Fact Fluency (60+ classroom-tested games adaptable to home), or Singapore’s Extra Practice books.
Optionally a subscription to Reflex Math, or free Anki, or XtraMath.

Weekly rhythm example:

Mon/Wed/Fri: 10 minutes of strategy work (introduce/review a derived-fact strategy with arrays) + 5 minutes of Leitner flashcards.
Tue/Thu: 10 minutes of a multiplication game + 5 minutes of mixed flashcards.
Once a week: 5-minute untimed “personal best” probe — 20 mixed facts, child checks their own work. Frame as “let’s see how much your brain has grown.”

Red flags to act on: if after 12–16 weeks of consistent, well-structured practice the child still cannot retrieve 2s/5s/10s reliably, or shows persistent finger counting on the easiest facts, or appears distressed disproportionate to the work, consult a learning specialist for screening.

Caveats

The “best curriculum” question is undersettled. No homeschool math curriculum has been compared in a large randomized trial against alternatives. Recommendations rest on alignment with research-supported principles, plus anecdotal homeschool community experience, not on direct head-to-head efficacy data.
The Jo Boaler / Greg Ashman debate is genuinely contested. Boaler’s claim that timed tests cause math anxiety in one-third of students is widely repeated but is not as solidly evidenced as her writing implies; Ashman and others have shown the cited references do not support that specific quantitative claim. At the same time, Ashcraft’s underlying research showing that math anxiety harms working memory during timed math is robust. The takeaway is to be skeptical of strong proclamations from either side and to choose moderate practices (some timing for personal goal-setting, no high-stakes timed tests).
Growth mindset effects are smaller than popularly claimed. Sisk et al. (2018) and Macnamara & Burgoyne (2023) meta-analyses find that universal growth-mindset interventions show small average effects, with the largest gains for students already struggling. The basic insight — praise effort and strategy, not innate ability — is sound and worth doing, but don’t expect mindset alone to compensate for inadequate instruction or practice.
Many vendor claims about fluency apps and mnemonic systems are based on internal studies. Reflex Math has the strongest independent evidence base (ESSA Tier 3). Times Tales, Multiplication.com, Imagine Math Facts, and most other commercial products have weaker independent evidence; they may work very well for individual children, but the marketing language should not be mistaken for randomized-trial results.
The CRA meta-analysis (Ebner et al. 2025) reports unusually large effect sizes (Tau-BC ≈ 0.99), driven largely by small single-case design studies which tend to overstate effects relative to RCTs. The directional finding (CRA helps) is well-supported, but the practical magnitude in a typical homeschool setting will be smaller.
Individual variation is large. Burns et al. (2015) showed that some children master a multiplication fact in 2 repetitions and others need 5+. Pace matters less than consistency and method.
This synthesis is biased toward English-language, U.S. and U.K.-centered research, which is most of the rigorous literature on math fact fluency. East Asian (particularly Singaporean, Japanese, and Chinese) approaches may emphasize different sequences and oral-rote elements that work well in those contexts but transfer imperfectly.
Finally, “mastery” is not “done forever.” Without maintenance review, multiplication facts decay over months. Plan for ongoing low-frequency mixed retrieval through fourth and fifth grade, especially over summer breaks.