Report Cards vs Test Scores, Phone Pouches and AI Tutors

A national report on Canadian learning loss, a large study of locked phone pouches, a UK trial of AI and human tutors, and a Stanford look at why math stays hard

By
Connect Education
November 18, 2026
Report Cards vs Test Scores, Phone Pouches and AI Tutors

Quick take

  • A national report covering every province and territory except Manitoba and Nunavut found learning loss in all of them, while teacher-assigned grades went up.
  • A study of more than 3 million US students found locked phone pouches added about one to two weeks of learning a year, a small gain other researchers still question.
  • In a UK trial, AI and human tutors fixed mistakes equally well in the moment, but neither moved scores on a broader math test after 10 weeks.
  • Stanford researchers found children with persistent math difficulties are slower to switch problem-solving strategies, which points to how math is taught, not just how much.

Across Canada, teacher-assigned grades went up after the pandemic while standardized test results went down. A national report released on November 13, 2026 says that gap may have hidden how much ground students really lost (University of Ottawa, 2026).

This digest covers four pieces of research released between late October and mid-November 2026. One is Canadian: the most detailed look yet at pandemic learning loss across the country. The other three come from the United States, the United Kingdom and Stanford University, and each speaks to a choice Canadian families are weighing right now: whether phone rules help, whether an AI tutor can do a teacher's job, and why some children stay stuck in math no matter how much they practise. Read side by side, they share one lesson about measurement. The quick signal, a good grade, a right answer in the moment, extra practice time, does not always match what a deeper check shows.

15
studies combined in Canada's national learning loss report (University of Ottawa, 2026)
3M+
US students in the new phone pouch study (NPR, 2026)
1–2 wks
of learning a year gained in schools that locked phones away (NPR, 2026)
68
children aged 8 to 10 in Stanford's math strategy study (ScienceDaily, 2026)

Did Canadian Students Recover From Pandemic Learning Loss? A National Report Says Report Cards Can Hide It

The report, Pandemic Learning Loss in Canada: Student Performance Before, During, and After COVID-19, was released by the University of Ottawa and published with the Royal Society of Canada. It combines 15 studies covering every province and territory except Manitoba and Nunavut, and it draws on ministry of education data from the 2014-15 school year through 2024-25. The panel was chaired by Tracy Vaillancourt, Canada Research Chair in Youth Mental Health and Violence Prevention at the University of Ottawa, with researchers from universities in nearly every province (University of Ottawa, 2026).

Every jurisdiction studied recorded declines in at least some standardized assessment results. The size and staying power of those declines were not uniform: they differed by subject, grade, gender and region. Some rural students were hit harder than others, and in many provinces the math gap between boys and girls widened in favour of boys. "The clearest message from this report is that learning loss occurred across Canada, but it did not follow a single pattern," Vaillancourt said in the release (University of Ottawa, 2026).

The finding most parents will want to sit with is the second one. While standardized tests showed losses, teacher-assigned grades showed improvement over the same years. "Grades rose even as test performance fell," Vaillancourt said. "And if we use report cards only, we risk believing that learning loss never took place or has been recovered when it hasn't" (University of Ottawa, 2026). The report recommends tracking achievement consistently over time and reading teacher grades alongside standardized results, and it calls for evidence-based support aimed first at the students who need it most. The news release does not give a single national figure for the size of the losses; the province-by-province detail sits in the full report.

For families, this does not mean a report card is wrong. Teachers grade what they see across many weeks, and a test captures one day. It means a strong mark is a good sign worth checking against other signals: provincial test results where your province has them, how homework actually goes, and the comments your child's teacher writes. Our guide on what your child's report card really means walks through what to look for, province by province. Many parents only find a gap when a report card and a test disagree, which is exactly the moment to ask the teacher where the difference comes from.

Do School Phone Bans Raise Test Scores? A Study of 3 Million Students Says a Little

On November 17, 2026, researchers from Harvard's Center for Education Policy Research, Dartmouth College and the testing organization NWEA released a study of schools that make students lock their phones in pouches for the whole school day. As NPR reported, the team analyzed student-level math and reading scores for more than 3 million elementary, middle and high school students over five school years, comparing 844 schools that adopted the pouches with similar schools that did not (NPR, 2026).

Harvard's Tom Kane, a co-author, said US students have been losing two to three weeks of learning each year since about 2013, and that strict phone bans "seem to plug about half of that leak," or roughly one to two weeks of learning a year. The gains did not fade after the first year; they appeared to continue for up to three years after schools adopted the pouches. Kane put the cost at roughly $25 US per student (NPR, 2026).

The caveats are real. The study has not yet been peer reviewed, and a separate, peer-reviewed study from April 2026 pooled school-level scores from several thousand pouch schools and found close to zero overall effect: some gains in high school math, offset by losses in middle school. One of that study's authors, Stanford's Tom Dee, called the early evidence on phone bans "largely disappointing" and said more study is needed (NPR, 2026). In Canada, Quebec has gone furthest: since September 18, 2025, students may not use a phone, headphones or other personal device anywhere on school property from the start of classes to the end, with exceptions teachers can allow (Éducaloi, 2025). For parents, the honest reading is modest. Fewer distractions help a little, and phones out of reach during homework at home costs nothing. A phone rule will not, on its own, close a specific gap in fractions or reading comprehension.

Is an AI Tutor as Good as a Human Tutor? A 1,179-Student Trial Says It Depends What You Measure

On October 29, 2026, The 74 reported on a randomized trial by the London ed-tech company Eedi and Google DeepMind, run in 10 public schools in the United Kingdom from April to July. Secondary students spent one class period a week, about 30 to 35 minutes, practising math on Eedi's platform with a teacher in the room. Whenever a student answered a question wrong, they got one of four kinds of help, assigned at random and kept for the whole study: the app's standard explanation, a live text chat with an expert human tutor, a text chat with an AI tutor whose every message a human approved, or the same AI tutor given a summary of the student's past mistakes (The 74, 2026).

In the moment, tutoring of either kind worked. Students who got AI or human tutoring answered similar follow-up questions correctly 56 to 58 per cent of the time, compared with 43 per cent after the standard explanation. That is a change from Eedi's smaller study a year earlier, where the supervised AI tutor came out slightly ahead of humans, 66.2 to 60.7 per cent. But on Star Maths, a broad standardized test students took before and after the 10 weeks, the results were flat. Among the 916 students who sat both tests, those tutored by AI scored about the same as those who only practised, and those with human tutors scored slightly lower, likely because they covered less material in more depth. Giving the AI each student's history did not help much, and the subgroup results raised a new question: tutored girls tended to do slightly better than girls who only practised, while tutored boys did worse (The 74, 2026).

Follow-up math questions answered correctly after a wrong answer, Eedi and Google DeepMind trial, 2026

Standard explanation in the app43%
AI tutor or expert human tutor (range across groups)56–58%

On the broader Star Maths test after 10 weeks, neither kind of tutoring beat practice alone. Source: The 74, October 29, 2026

Keep the setting in mind before drawing big conclusions. This was short text-chat help inside a math app, triggered by a wrong answer, for 10 weeks, not regular one-on-one lessons with a teacher who gets to know the child. The results were also released by the company that makes the tool, and Eedi says a larger trial with more than 15,000 students is already under way. The useful takeaway is the difference between fixing a mistake and building a skill that shows up on a later test. Whatever help your child gets, ask how progress will be checked a month from now, not just today. Our post on why one-on-one tutoring works looks at what a steady match between child and tutor adds.

Why Do Some Kids Keep Struggling With Math? Stanford Points to Switching Strategies

A study published in JNeurosci and described by the Society for Neuroscience on November 12, 2026 looked at children with developmental dyscalculia, a learning disability that causes ongoing difficulty with math problem solving. Researchers at Stanford University, led by Oliver Lasnick, studied 68 children aged 8 to 10, looking at how they chose and used strategies on math problems and how that related to their brain activity (ScienceDaily, 2026).

Children with dyscalculia were generally less effective at using different strategies than their peers. They took longer to shift from one method to another, for example from counting each number to using a fact they already knew, and they responded less when a problem got harder. Over time, other children got better at picking a good approach; children with dyscalculia did not show the same improvement. Activity in one brain system differed between the groups and predicted how well children counted and switched strategies. The authors connect this to executive functions such as attention, working memory and cognitive flexibility, and Lasnick stressed that dyscalculia looks different from one child to the next (ScienceDaily, 2026).

This is a small study, and the brain differences it found are linked to the difficulties, not shown to cause them. It is not a test parents can use at home. What it adds is a practical clue. A child who still counts on fingers for problems classmates solve from memory may not need more of the same drills; they may need someone to teach, out loud and step by step, when to switch to a better method and why. Our guide on how to help a child who hates math has ideas for home, and our math tutoring page explains how families find a tutor for a specific gap.

What This Means for Families and Teachers

Each of these studies is, in its own way, about telling the difference between looking fine and being fine. Canadian report cards improved while test scores slipped. Phone pouches nudged scores up a little, but not enough to fix a missed skill. An AI tutor corrected mistakes as well as a person did, yet neither showed up on a broader test after 10 weeks. And children who struggle with math may be practising hard with the wrong strategy. The signal that matters is steady progress on the specific thing a child finds hard, checked over weeks, not a single good moment.

If a report card and a test are telling you two different stories, start by asking your child's teacher what the gap is. If your child needs more help than the classroom can give right now, a tutor can work on that one skill and show you how it is moving. Every tutor on Connect Education is a certified teacher, and lessons happen online, at your home or at a public library. You can browse certified tutors or tell us about your child to get matched.

Resources

Sources and further reading

  1. University of Ottawa (2026), Landmark report finds consistent pandemic learning loss and gaps between grades and test scores across Canada (November 13, 2026), https://www.uottawa.ca/en/news-all/landmark-report-finds-consistent-pandemic-learning-loss-gaps-between-grades-test-scores
  2. NPR, via MPR News (2026), Does locking up a student's cellphone improve test scores? A new study suggests so (November 18, 2026), https://www.mprnews.org/story/2026/11/18/npr-school-cellphone-ban-learning-loss-yondr-research
  3. Éducaloi (2025), No More Cellphones in Schools: What You Need to Know (September 4, 2025), https://educaloi.qc.ca/en/legal-news/no-more-cellphones-in-schools-what-you-need-to-know/
  4. The 74 (2026), Five Things You Should Know About New AI Tutor Study (October 29, 2026), https://www.the74million.org/article/five-things-you-should-know-about-new-ai-tutor-study/
  5. Society for Neuroscience, via ScienceDaily (2026), A surprising brain difference may explain why some kids struggle with math (November 12, 2026), https://www.sciencedaily.com/releases/2026/11/261112025542.htm
Related topics
Parenting and school
Smiling certified teacher holding a tablet
Real answers from real people
Certified teachers
Online, home or library
FAQ

Frequently asked questions

What are the early signs of dyscalculia in a child?

Common early signs include still counting on fingers for simple facts long after classmates have stopped, trouble judging which of two numbers is bigger, mixing up which operation a word problem needs, and strong worry about math. These signs alone do not mean a child has dyscalculia. A diagnosis comes from a qualified professional, usually a psychologist, so start by asking your child's teacher whether the school can arrange an assessment.

Should my child use an AI tool for math homework?

An AI tool can be useful for a quick explanation or extra practice questions, especially when a parent is unsure how a topic is taught now. It works best as a helper, not the only source of help: it may not notice why your child keeps making the same mistake, and answers can be wrong. Ask your child to explain the step back to you in their own words before moving on.

Connect Education

Get your child the right support

Certified teachers, matched to your child's goals and how they learn. Start with a free 15-minute assessment.

A one-on-one plan built around your child
Online, in person or at a public library
Background-checked, certified teachers
Progress updates after every session