Parents ask us where our methods come from, so we wrote it all down. We teach from published programs and measure against published benchmarks. Our decisions follow rules other teachers have tested for decades. We didn’t invent the methods. Our job is to use them carefully, one child at a time, and to show you whether they’re working.
Below, we go through each part of what we do. For every study, we say what it found and how it shows up in your child’s sessions. Where a study has limits, we say that too. The full list of sources is at the bottom, with links wherever a copy is available.
I. Direct Instruction
When a skill hasn’t been learned yet, we teach it with a published Direct Instruction program. For reading that’s Reading Mastery or Corrective Reading. For math it’s Connecting Math Concepts, Corrective Mathematics or Essentials for Algebra. Each program has its own placement test, so your child starts at the lesson that fits, wherever that is (Engelmann et al., 2008).
The lessons are scripted. We show a skill, we do it together with your child, and then your child does it alone. If an answer is wrong, we correct it right away and come back to it a few minutes later. A skill counts as learned when every answer is correct and quick (Engelmann & Bruner, 2008). The programs are also written so a child gets most new items right on the first try, which keeps lessons moving and keeps kids confident (McGraw-Hill, n.d.).
The best summary of the evidence we’ve found is a 2018 review of 328 studies of these programs. It found positive average effects in every academic subject (Stockard et al., 2018). We want to be fair here. The review’s authors have worked part-time for the National Institute for Direct Instruction, and the big federal study that made Direct Instruction famous, Project Follow Through, was criticized for how its results were labeled (House et al., 1978). The What Works Clearinghouse also reviewed Corrective Reading and found one study that met its standards, with no discernible effects (What Works Clearinghouse, 2010). We’ve read both sides. We use these programs because every step is spelled out and every program comes with its own tests, so we can show you whether it’s working for your child.
II. Precision teaching and the daily chart
Precision teaching was developed by Ogden Lindsley at the University of Kansas Medical Center in Kansas City, Kansas, in the 1960s. The idea is simple. Your child practices one skill for a minute at a time, we count what’s right and what’s wrong, and every timing goes on a chart (Calkin, 2005).
The chart is called the Standard Celeration Chart, and it shows how fast a skill is growing from week to week. We draw a trend line once there are at least seven charted days (Journal of Precision Teaching and Celeration, 2005). We also set a minimum line for how fast the skill should grow, so we can see right away when it falls behind (Lindsley, 1996).
Our decisions follow published rules. In the Great Falls Precision Teaching Project in Montana, schools ran daily one-minute timings and charts. Their rules were plain: if a skill shows no growth for three days, change something, and once a child reaches the goal on two days out of three, move on. The project reported large gains on standardized achievement tests for its elementary students (Beck & Clement, 1991). We use the same rules.
Here’s why we trust this part. A review of 21 controlled studies found that when teachers measured progress often and changed their teaching by set rules, students learned more, with an average effect size of .70 (Fuchs & Fuchs, 1986). Precision teaching’s own research base is smaller, and we won’t oversell it. What we do with the chart is what that review describes: measure often, and act on what you see.
III. Fluency: accurate and quick
We don’t stop when a child gets something right. We keep practicing until it’s right and quick. That’s what fluency means, and it matters because fluent skills last, hold up through longer work and carry over to new tasks (Binder, 1996). A child who has to count on their fingers for 7 × 8 has nothing left for the word problem.
The goals come from how people who already know a skill well perform it. Aims are set from the rates of fluent performers (Koorland et al., 1990). For writing math facts, about 40 digits a minute still falls short. Fluency starts around 80 to 120 digits a minute (Kubina, 2005).
Practice is short. Most timings last one minute. When a skill is new, we do quick 20-second sprints. In one small study, three 20-second tries beat one full minute and gave students about 30% more practice in the same amount of time (Brady & Kubina, 2010).
For reading, the main fluency practice is repeated reading, where your child reads the same short passage a few times with feedback. The National Reading Panel found that guided oral reading, including repeated reading with feedback, improved reading, with an average effect size of 0.41 (National Institute of Child Health and Human Development, 2000). We run it the way Therrien and Kubina (2006) lay it out: short passages, a few rereads, several times a week.
This whole order (teach it, practice it to fluency, then use it) is how Morningside Academy in Seattle has taught for decades (Johnson & Street, 2012). In a randomized trial of Morningside’s math facts curriculum, the students who practiced to fluency did better than students taught the usual way on fluency, endurance and stability (McTiernan et al., 2016). It was a small study of 28 children, so we take it as encouraging and leave it there.
IV. Reading and the published benchmarks
At the free check, your child reads three short grade-level passages aloud for one minute each, and we use the middle score. That number, words correct per minute, is compared with the published benchmark for your child’s grade and the time of year (Hasbrouck & Tindal, 2017). Those benchmarks were compiled from about 6.9 million scores on three widely used reading tests. For seventh and eighth grade, we use the same authors’ earlier table (Hasbrouck & Tindal, 2006).
The time of year matters, because the benchmark rises as the year goes on. For third grade it’s 83 words a minute in the fall, 97 in the winter and 112 in the spring (Hasbrouck & Tindal, 2017). Growth also tends to be faster in the fall than in the spring (Christ et al., 2010).
We’re careful about what this comparison can tell you. The passages are our own, and they haven’t been field-tested yet. That matters, because passages written at the same level can still differ in how hard they read (Christ & Ardoin, 2009). So we give you a benchmark level: well below, needs fluency building, or on track. We don’t turn it into a percentile.
We also don’t judge progress from one or two readings. Single scores bounce around, so we wait for at least eight readings before we call a trend (Fuchs & Fuchs, 2011).
V. Math facts and mastery criteria
For math facts, we use published mastery criteria. These are digits-correct-per-minute levels that show whether a child is still learning a skill or has it down (Burns et al., 2006; VanDerHeyden & Burns, 2009). Experts set these levels. They aren’t norms, so our check doesn’t give a math percentile.
Timed practice is also what the federal practice guides recommend. One guide recommends regular timed activities, with students recording their scores on a chart over time (What Works Clearinghouse, 2021). Another recommends about ten minutes in every session on fact fluency (Gersten et al., 2009). We teach the strategy first, then we time it, and your child charts it.
The standards say when facts should be automatic. Sums should be known from memory by the end of second grade, and products by the end of third grade (National Governors Association Center for Best Practices & Council of Chief State School Officers, 2010).
VI. SAFMEDS
SAFMEDS stands for Say All Fast Minute Every Day Shuffled. Lindsley came up with it in the late 1970s (Quigley, 2014). It’s a deck of cards with a question on the front and the answer on the back. Your child goes through as many cards as they can in one minute, every day, and we chart the count. We use SAFMEDS for things that need to be instant, like letter sounds, sight words and math facts.
One honest note: the basic version doesn’t always build fluency with harder material on its own, and added steps helped in Quigley’s (2014) study. So we watch the chart, and when a deck stalls, we change how your child practices it.
VII. Why we chart every timing and show you
Every timing goes on the chart, and you get the chart every Friday. We do this for two reasons.
First, measuring helps most when someone acts on it. The gains in the review we mentioned above came from teachers who measured often and changed their teaching based on what they saw (Fuchs & Fuchs, 1986). The chart is how we hold ourselves to that.
Second, you should be able to see it too. Within a couple of weeks, you’ll know whether something is working, the same as we do. Your child charts their own timings as well, which the federal math guide recommends (What Works Clearinghouse, 2021).
VIII. Why family tutoring is one-on-one
Families get three one-on-one sessions a week, 45 minutes each, with the same tutor every time. That schedule matches what tutoring research supports. High-impact tutoring is defined as three or more sessions a week with a consistent tutor (National Student Support Accelerator, n.d.). A large review of tutoring experiments found an average effect of 0.37 standard deviations, with bigger effects in the early grades (Nickow et al., 2020).
That same review found one-on-one tutoring had the largest effects, with small groups not far behind. For families, we teach one child at a time, and the reason is the method. Every decision we make comes from one child’s chart: where they’re placed, what their aims are, and when to change course. A child at lesson 16 and a child at lesson 40 need different things in the same minute, and we want every minute of your child’s session to be about your child.
IX. Small groups in schools
In schools we teach small groups, and the research supports that too. In a study of second graders who were behind in reading, one-on-one teaching was not significantly better than groups of three on any outcome. One-on-one did beat groups of ten (Vaughn et al., 2003). The federal reading practice guide gives its strongest evidence rating to intensive, systematic small-group instruction for students who score below benchmark. As the way to carry it out, it suggests groups of three or four, three to five times a week (Gersten et al., 2008).
How often a group meets matters. In a randomized study of second through fourth graders who were behind in math, four sessions a week led to clearer gains than one or two sessions a week, with the same total minutes (Codding et al., 2016). The tutoring review above also found larger effects for programs that met three or more days a week (Nickow et al., 2020). Our school groups meet four days a week.
Direct Instruction programs were written for groups. Each program has a placement test, so we group children who place at about the same lesson. The group answers together on a signal, every child also gets individual turns, and a mastery test every five to ten lessons tells us who needs more time (Silbert, 2005). Precision teaching stays individual inside the group. Each child works on their own sheet during the same timing and moves on when they reach their own aim (McTiernan et al., 2016).
There’s one tradeoff we take seriously. In a large review of programs for struggling readers, one-on-one tutoring showed larger effects than small-group tutoring, though both were positive (Neitzel et al., 2022). In a randomized study of first graders who were behind in reading, one-on-one teaching beat groups of two, three and five on eight of nine measures (Schwartz et al., 2012). So our groups stay small, and they’re smallest for the youngest children and for the children who are furthest behind. When a child’s chart stays flat, we recommend one-on-one tutoring. More about the School Program is on our For Schools page.
X. Why the group size depends on the grade
Three kindergartners and three eighth graders need different things, so we size our school groups by grade. Groups in kindergarten and first grade have two children, three at most, and meet for 30 minutes. Groups in grades 2–5 and 6–8 have three students, four at most, and meet for 45 minutes. Every group is one grade band and one program, with students from the same grade or two neighboring grades.
The large tutoring review above shows why the youngest children get the smallest groups. In preschool and kindergarten, one-on-one programs averaged an effect of 0.51, pairs 0.47, and groups of three or more 0.32. First grade looked the same: 0.44, 0.41 and 0.32. In grades 2–5 the pattern turned around: groups of three or more averaged 0.46, and one-on-one and pairs each averaged 0.25. These are averages across different programs, and the authors didn’t test the differences, so we read them as a pattern. In the authors’ words, groups of three or more “seem to perform especially well in grades 2-5” (Nickow et al., 2020).
Session length follows the same logic. The federal reading guide says a child’s grade should set the length, from 20 to 30 minutes in kindergarten up to 40 to 50 minutes in second grade (Gersten et al., 2008). The National Student Support Accelerator sets 30 minutes as the floor for high-impact tutoring and notes that later grades can benefit from 45 to 60 minutes (National Student Support Accelerator, n.d.). So our kindergarten and first-grade sessions are 30 minutes, and sessions from second grade up are 45.
One honest note about early math. In two randomized studies, groups of two did about as well as groups of five, in kindergarten and in first grade (Clarke et al., 2017; Doabler et al., 2024). We still keep K–1 groups at two or three. A smaller group gives each young child more turns in a short lesson, and the Reading Mastery kindergarten teacher’s guide says to make the lowest-performing group the smallest (Engelmann & Bruner, 2008).
For grades 6–8, we lead with the program and the time. Older students who are behind need a program written for them, so we teach from Corrective Reading, which its publisher writes for grades 3 through 12 (Engelmann et al., 2008), and from Corrective Mathematics and Essentials for Algebra. For grades 4–9, the federal reading guide gives its strongest rating to three practices: teaching students to decode long words, purposeful fluency-building activities, and routines that build comprehension (Vaughn et al., 2022). Every one of our reading sessions for grades 6–8 covers all three. Sessions run 45 minutes, which fits the tutoring guidelines for later grades (National Student Support Accelerator, n.d.), and every student keeps their own chart.
Group size matters less at this age. In a year-long study of seventh and eighth graders who were behind in reading, groups of three to five did about the same as groups of ten to fifteen (Vaughn et al., 2010). We keep groups at three or four anyway, so every student gets individual turns and we can follow every chart. A widely cited plan for tutoring in public schools pairs three middle schoolers with each college tutor (Kraft & Falken, 2021). And as we said above, the What Works Clearinghouse’s one qualifying study of Corrective Reading showed no discernible effects, so when an older student’s chart stays flat, we change course quickly.
Who keeps the chart changes with age as well. In kindergarten through second grade, we run the timings, keep each child’s chart and show it to them every day. From third grade up, students mark their own charts and we check every entry, as the federal math guide recommends for timed practice (What Works Clearinghouse, 2021). In grades 6–8, students also set their own aim for the week. At every age, feedback stays private: we never read a score aloud or compare one student with another.
Our praise is specific, too: we name exactly what a child did well. In a review of six studies, praise like that increased students’ time on task, though the evidence base is small (Royer et al., 2019).
XI. Where a percentile comes from
A percentile tells you how your child compares with a large group of students in the same grade. A real percentile can only come from a test that was normed on a group like that. Our own passages and probes don’t produce one.
When a percentile is part of the check, it comes from easyCBM. Its 2025 norms are based on 2024–25 scores from school districts in four regions of the U.S. (Nese & Wallin, 2025). An easyCBM percentile is the share of students in that group who scored below your child (Riverside Assessments, 2025). Later, we plan to add aimswebPlus, a nationally normed screening system from Pearson (Pearson, 2026). Whenever we report a percentile, we’ll tell you which test it came from.
XII. Practice at home
Between sessions, your child practices about five minutes a day at home: a one-minute timing on a skill that’s already accurate in sessions, and a quick mark on the chart. We’ll show you exactly what to do.
A review of family reading programs found the strongest results when parents were shown how to practice specific skills with their child, like letter sounds and sight words (Sénéchal & Young, 2008). Short practice spread across days, with a little self-testing, also came out on top in a large review of study methods (Dunlosky et al., 2013). Five minutes is enough, as long as it happens most days.
If you’d like to read any of these yourself, the links are below. And if you ever wonder why we’re doing something in a session, ask us. We’ll show you where it comes from.