How the NurseSavvy Math Coach Teaches (and the Research Behind It)

Short answer: one tutor asks one small question at a time, names the exact slip when you make one, and hands over the steps as you show you’ve got them. NurseSavvy has two coaches built on that engine: the Dosage Calc Coach (tutor Tala) for nursing school, and the TEAS Math Coach (tutor Furaha) for the TEAS. This page explains each teaching decision, the research behind it, and how strong that research is.

Two ground rules for reading it. Most of these studies were done with school students or in lab experiments, not adult nursing students, so applying them here is a reasonable extrapolation, not proof. And none of the studies tested NurseSavvy. We don’t yet publish outcome numbers for the coach.

The twelve decisions at a glance

“Design choice” means no study tests that specific thing. We built it to follow a principle and say so.

Each decision, and why

1. One small question at a time

What you see: “Walk me through it” opens a chat. The tutor reads the problem with you first, then asks one plain question per step and waits for your answer. Arithmetic steps get nudges, not answers. The walk-through is scripted from tested recipes, not generated by AI, so it can’t invent clinical math.

Why: VanLehn’s 2011 review found human tutoring at d = 0.79 and intelligent tutoring systems, which give feedback on each step, at d = 0.76, each compared with no tutoring. Step-by-step feedback came out nearly as effective as a human tutor. The limit: a review of tutoring systems in K–12 math found effects near zero (Steenbergen-Hu & Cooper, 2013). Tutoring software is not magic; the design matters.

2. Worked steps that fade

What you see: beginners get the full walk-through. After each right answer, the tutor writes that step out, and the finish screen puts the whole solution back together in your method. The answer box is always there, so if you can do it, you skip straight to the answer.

Why: across math studies from elementary school through college, learning from worked examples beat solving problems cold, g = 0.48 (Barbieri et al., 2023). When help adapts to the learner, it works better still: in a tutoring system, adaptive fading beat fixed fading, which beat plain problem solving (Salden et al., 2010). The full story is in worked examples vs. practice problems.

3. Mistake catchers that name the slip

What you see: a wrong number gets a specific message, like “that’s the mL/hr rate; gravity rates are drops per minute”, then a second try with a hint. The coach knows the common slips: flipping what you have and what you need, pounds used as kilograms, minutes treated as hours, one dose vs. the day’s total. A tenfold miss gets “count the zeros”.

Why: in computer-based learning, a bare “wrong” barely helped (effect 0.05), while an explanation helped far more (0.49) (Van der Kleij et al., 2015). Students’ math errors are often systematic “bugs” that can be named (Brown & Burton, 1978). And feedback aimed at the person instead of the task can backfire: over a third of feedback interventions made performance worse (Kluger & DeNisi, 1996). So the catchers talk about the step, never about you.

4. Three dosage methods

What you see: “Teach it as” in the top bar picks Formula, Ratio & proportion, or Dimensional analysis. Each method asks its own setup questions, your pick sticks, and every method lands on the same answer.

Why: small nursing studies lean toward dimensional analysis, but no review names a best method, and inconsistent methods are linked to weak skills (Rice & Bell, 2005). So the coach teaches the one your program uses, the same way every time. The study-by-study table is in dimensional analysis vs. ratio and proportion vs. formula.

5. Mental math, any valid split

What you see: in the TEAS Math Coach, percent problems can be taught as mental math. You break the percent into friendly pieces (50%, 25%, 10%, 5%, 1%) any way that works, like 35% = 25% + 10% or 10% × 3.5, and the tutor walks the split you picked.

Why: in a study of 132 sixth graders, seeing more than one way to solve equations made students more flexible at choosing one (Star & Rittle-Johnson, 2008). Students’ percent strategies grow from friendly benchmarks like 50% and 10% toward formal methods (Lembke & Reys, 1994). Accepting your own split builds on that instead of drilling one pattern. More in mental math for nurses.

6. Gut check before, sense check after

What you see: one “more or less than ___?” pick on a 5-second bar before the walk-through, with Skip. At the end, the answer is checked against it.

Why: nursing guidance says to know roughly what a sensible answer looks like before you calculate, because it catches the decimal slips behind tenfold errors (Royal College of Nursing). An estimate never replaces the calculation. The 5-second timer is our design choice. More in how to estimate dosage answers.

7. Optional pictures

What you see: one picture per problem type where it helps, behind the chart button so it never clutters the screen: a dimensional-analysis chain with the units struck out, a percent bar, double number lines for proportions, a metric staircase, a balance for two-step equations, and the dose drawn as tablets, a medicine cup or a syringe sized to it. The answer shows as “?” until you find it.

Why: the strongest evidence is for diagrams that show a problem’s structure. In a randomized trial with 1,163 seventh graders, schema-based instruction on proportion and percent problems beat the usual teaching, g = 1.24 (Jitendra et al., 2013). In a randomized study of 120 nursing students, dosage exams with medication and syringe images scored higher, 7.22 vs. 6.04 out of 8 (Weil-Lotan et al., 2026). The warning: a picture helps only if you can connect it to the numbers (Rau & Matthews, 2017), so every picture is labeled with the numbers in the current step. The balance only shows positive amounts, because negative numbers caused errors with that model (Vlassis, 2002). The staircase has no study behind it; it’s a memory aid. The percent pictures are worked through in TEAS percents and proportions, visually.

8. Proportions written stacked, with arrows

What you see: TEAS proportions are written one fraction over another, as on paper. The cross-multiply step draws an X of arrows, with the pair being multiplied highlighted.

Why: researchers have long warned that students use cross-multiplication without knowing why it works (Lesh, Post & Behr, 1988). Showing the structure is our answer to that. No study compares this layout with others, so it’s a design choice.

9. Smart practice

What you see: signed-in students with a few results get Smart practice as the first option. A type you just missed comes back with new numbers until you solve one cleanly. Otherwise you get your weakest type, at the level you’ve earned in it, and every fourth problem mixes in the type you’ve practiced least. Help fades per type: the full walk-through while a type is new or shaky, then the tutor sets it up and hands off, then you work it alone.

Why: mastery learning, not moving on until a skill holds, averaged 0.52 SD across 108 evaluations, but only 0.29 on standardized tests (Kulik, Kulik & Bangert-Drowns, 1990). Tracking each skill separately to decide what to practice next comes from knowledge tracing (Corbett & Anderson, 1995); ours is a simpler running average. The counterweight: being digital isn’t enough. Across 13 nursing schools, a face-to-face lecture improved medication calculation more than an e-learning course (Van Lancker et al., 2016). Our case rests on the design, not the screen. The cut-offs for when help fades are our own choices; no study sets them.

10. Mixed problem types

What you see: the “All” chip mixes every type, and you try the answer before seeing any help.

Why: in a randomized trial in 54 seventh-grade math classes, students who practiced mixed problem types scored 61% on a surprise test a month later, vs. 38% for blocked practice (Rohrer et al., 2020). Across studies the math effect is smaller, g = 0.34 (Brunmair & Richter, 2019). Expect mixed practice to feel worse: in a small study of college students, mixed practice scored lower during practice and much higher a week later (Rohrer & Taylor, 2007).

11. One problem per phone screen

What you see: on a phone, you pick the type, level and method first, then see just the problem, the answer box and a small bottom bar. Tools open one at a time.

Why: working memory is small, so instruction should cut what doesn’t help, and break lessons into learner-paced pieces (Mayer & Moreno, 2003). Reasoning on a small screen was worse than on a full display, most of all for people with low working memory (Sanchez & Branaghan, 2011). The lesson is “fit each step on one screen”, not “phones are bad”.

12. A rounding rule on every dosage answer

What you see: every dosage answer states its rule (a 3 mL syringe measures tenths, drops are whole), and uneven answers get a “how do you round?” question first.

Why: how a dose is written is a safety issue. ISMP (2015) lists the error-prone ways, like trailing zeros, and rounding conventions come from nursing textbooks such as Open RN, not from trials. The table is in dosage calculation rounding rules.

What’s a design choice, and what isn’t proven

We want to be plain about where the research stops.

• Design choices, not tested: the cross-multiply arrows, the metric staircase, the 5-second gut-check timer, the cut-offs for when help fades and levels change, and the 1-minute demo on the coach pages.
• Not proven: which dosage method is best. And we found no trial of retrieval or spaced practice for nursing dosage calculation specifically, so that evidence comes from math and medical education.
• No outcome numbers yet: we don’t publish results for the coach itself. When we do, they’ll go through the same publish rules as the rest of our research page: things like first-try accuracy by problem type over time, and how often a right gut check goes with a right final answer.

Calculation skill is a required, testable safeguard, and we don’t claim more than that: a review of 33 studies found insufficient evidence that medication errors are caused by nurses’ poor calculation skills (Wright, 2010). We teach it because programs and patients both need it done right.

Try both coaches

Dosage Calc Coach: oral and injection doses, reconstitution, IV rates, weight-based doses, titrated drips, heparin and insulin.

Try an oral dose with TalaA fresh problem in place. Answer it, or tap “Walk me through it” for one small step at a time. Free, no account.Start

TEAS Math Coach: percents, ratios and proportions, algebra, measurement, geometry and data, the way the TEAS asks them.

Try a percent problem with FurahaA fresh problem in place. Answer it, or tap “Walk me through it” for one small step at a time. Free, no account.Start

Sources

• VanLehn, K. (2011). The relative effectiveness of human tutoring, intelligent tutoring systems, and other tutoring systems. Educational Psychologist, 46(4), 197–221.
• Steenbergen-Hu, S., & Cooper, H. (2013). A meta-analysis of the effectiveness of intelligent tutoring systems on K–12 students’ mathematical learning. Journal of Educational Psychology, 105(4), 970–987.
• Barbieri, C. A., Miller-Cotto, D., Clerjuste, S. N., & Chawla, K. (2023). A meta-analysis of the worked examples effect on mathematics performance. Educational Psychology Review, 35(1).
• Salden, R. J. C. M., Aleven, V., Schwonke, R., & Renkl, A. (2010). The expertise reversal effect and worked examples in tutored problem solving. Instructional Science, 38(3), 289–307.
• Van der Kleij, F. M., Feskens, R. C. W., & Eggen, T. J. H. M. (2015). Effects of feedback in a computer-based learning environment on students’ learning outcomes: A meta-analysis. Review of Educational Research, 85(4), 475–511.
• Brown, J. S., & Burton, R. R. (1978). Diagnostic models for procedural bugs in basic mathematical skills. Cognitive Science, 2(2), 155–192.
• Kluger, A. N., & DeNisi, A. (1996). The effects of feedback interventions on performance. Psychological Bulletin, 119(2), 254–284.
• Rice, J. N., & Bell, M. L. (2005). Using dimensional analysis to improve drug dosage calculation ability. Journal of Nursing Education, 44(7), 315–318.
• Star, J. R., & Rittle-Johnson, B. (2008). Flexibility in problem solving: The case of equation solving. Learning and Instruction, 18(6), 565–579.
• Lembke, L. O., & Reys, B. J. (1994). The development of, and interaction between, intuitive and school-taught ideas about percent. Journal for Research in Mathematics Education, 25(3), 237–259.
• Royal College of Nursing. (2019). Safety in numbers: Tackling number problems and Estimation [archived web pages]. Internet Archive.
• Jitendra, A. K., Star, J. R., Dupuis, D. N., & Rodriguez, M. C. (2013). Effectiveness of schema-based instruction for improving seventh-grade students’ proportional reasoning: A randomized experiment. Journal of Research on Educational Effectiveness, 6(2), 114–136.
• Weil-Lotan, D., Solnica, A., Benbenishty, J., & Ilsar, T. (2026). Does the addition of images to the medication dosage calculation exam improve metacognitive components and performance? Nurse Educator, 51(5), 231–238.
• Rau, M. A., & Matthews, P. G. (2017). How to make ‘more’ better? Principles for effective use of multiple representations to enhance students’ learning about fractions. ZDM, 49(4), 531–544.
• Vlassis, J. (2002). The balance model: Hindrance or support for the solving of linear equations with one unknown. Educational Studies in Mathematics, 49(3), 341–359.
• Lesh, R., Post, T., & Behr, M. (1988). Proportional reasoning. In J. Hiebert & M. Behr (Eds.), Number concepts and operations in the middle grades (pp. 93–118). Lawrence Erlbaum & NCTM.
• Kulik, C.-L. C., Kulik, J. A., & Bangert-Drowns, R. L. (1990). Effectiveness of mastery learning programs: A meta-analysis. Review of Educational Research, 60(2), 265–299.
• Corbett, A. T., & Anderson, J. R. (1995). Knowledge tracing: Modeling the acquisition of procedural knowledge. User Modeling and User-Adapted Interaction, 4(4), 253–278.
• Van Lancker, A., Baldewijns, K., Verhaeghe, R., Robays, H., Buyle, F., Colman, R., & Van Hecke, A. (2016). The effectiveness of an e-learning course on medication calculation in nursing students: A clustered quasi-experimental study. Journal of Advanced Nursing, 72(9), 2054–2064.
• Rohrer, D., Dedrick, R. F., Hartwig, M. K., & Cheung, C.-N. (2020). A randomized controlled trial of interleaved mathematics practice. Journal of Educational Psychology, 112(1), 40–52.
• Brunmair, M., & Richter, T. (2019). Similarity matters: A meta-analysis of interleaved learning and its moderators. Psychological Bulletin, 145(11), 1029–1052.
• Rohrer, D., & Taylor, K. (2007). The shuffling of mathematics problems improves learning. Instructional Science, 35(6), 481–498.
• Mayer, R. E., & Moreno, R. (2003). Nine ways to reduce cognitive load in multimedia learning. Educational Psychologist, 38(1), 43–52.
• Sanchez, C. A., & Branaghan, R. J. (2011). Turning to learn: Screen orientation and reasoning with small devices. Computers in Human Behavior, 27(2), 793–797.
• Institute for Safe Medication Practices. (2015). ISMP’s list of error-prone abbreviations, symbols, and dose designations.
• Open Resources for Nursing. (n.d.). 5.5 Rounding. In ACC nursing resources. CC BY 4.0.
• Wright, K. (2010). Do calculation errors by nurses cause medication errors in clinical practice? A literature review. Nurse Education Today, 30(1), 85–97.