If your child's school has changed reading programs in the past few years, you may have heard "structured literacy" and "balanced literacy" used as if they were opposite ends of a dial. A 2025 review in Discover Education (volume 4, article 488) tried to put numbers behind that dial — and, just as usefully, to say where the numbers stop being decisive.
What the authors did
The team calls the work an exploratory quantitative analysis. Rather than searching by keyword, they started from commercially popular reading programs and picked them by how often they appear in public discussion. To count as structured literacy (SL), a program had to be commonly called SL and to meet the 2000 National Reading Panel's definition of systematic phonics: a planned, sequential set of sound-letter elements, taught explicitly. Thirteen qualified, among them Jolly Phonics, Wilson, Spire, Lexia, Open Court and Reading Mastery. To count as balanced literacy (BL), a program had to be commonly described that way and match the panel's description of unsystematic phonics. Four did: Fountas and Pinnell Core Instruction, Levelled Literacy Instruction, Reading Recovery and Units of Study.
The authors then pooled the experimental studies attached to those programs: 40 for SL, 28 for BL.
The numbers
SL studies had a mean unweighted effect size of .43 (95% CI .28 to .57) and a fixed weighted mean of .46. BL studies came in at .19 (95% CI −.02 to .41), with a fixed weighted mean of .29.
Two details matter more than the headline. First, the SL interval sits comfortably above zero; the BL interval does not, because its lower bound dips below zero. Read plainly, the BL evidence as a group is compatible with no effect at all — my reading of the interval, not a sentence from the paper. Second, the SL figure is essentially the same as the 0.44 the National Reading Panel reported in 2000 across 36 phonics studies, a stability the authors highlight.
The gap also widened with time: in long-term studies, SL effects were larger, with a mean difference in effect sizes of .13.
For context, the 2000 panel found systematic phonics ahead of literature-based instruction containing only incidental phonics by g = 0.31, and ahead of whole-word instruction with no systematic phonics by g = 0.51. The less phonics the comparison group received, the bigger the advantage. Later work by Stuebing and colleagues found the same pattern, which matters whenever two programs are compared.
The definition problem
Balanced literacy has no universally accepted definition. Many BL authors point to Pressley and colleagues' paper "Balanced Literacy Instruction," which loosely describes balancing skills instruction with authentic reading and writing. The distinctions the review lists: no specified phonics scope and sequence, reliance on undefined "context clues," no recommendation for decodable texts, less skill-and-drill, more emphasis on motivation and real reading.
The authors then make an argument rather than report a finding. They contend that the Pressley definition is consistent with the National Reading Panel's definition of whole language; that Reading Recovery, often labelled BL, was listed as a whole-language program in that 2000 report; and therefore that BL is a rebranding of whole language and SL a rebranding of systematic phonics, with SL adding comprehension, fluency and syntax instruction. That is a claim about definitions, and the programs' own authors might dispute the labels. It is a clarifying reframing, not a verdict.
What it cannot settle
The authors are unusually direct about their main limitation: programs were chosen by media prominence, and the initial inclusion was, in their words, "less than fully systematic." That is why the paper is exploratory and why they describe the findings as hypothesis-generating rather than confirmatory. They also note it may be the most comprehensive synthesis to date by number of experimental papers.
Four BL programs is a thin base for a category whose definition is contested. And the critique from Bowers (2020) — that phonics meta-analyses have not shown long-term gains in reading comprehension and may be affected by weak study design and publication bias — is probed with moderator analyses and funnel plots but not dissolved. Effect-size conventions (.20 small, .50 moderate) are rules of thumb, not laws; John Hattie's 0.40 benchmark for education is influential and contested. So: a promising advantage for structured literacy, not a guarantee for every child, and no final word on long-term comprehension.
What to ask instead of which label
Look past the brochure. Is the phonics teaching explicit — does the teacher explain and model — and systematic, following a sequence where each skill builds on the last? Are early texts chosen to match the sounds a child has learned? When a child stumbles, does the adult prompt decoding, or guessing from the picture and the first letter? Is comprehension taught alongside word reading? Is there still time for real books, and for children who want to read them?
Those questions travel better than labels. Balanced literacy's emphasis on motivation is not nothing, and the review's own account of the approach includes it. The reasonable position for now is unglamorous: explicit, sequential teaching of decoding has the better evidence, especially over time, and any program still deserves watching in your own child's case.