Blast from the Past: This blog first posted on November 3, 2018, and was reposted in revised form on March 14, 2026. The original title was, “The Whys and Hows of Research and the Teaching of Reading.” As you can see, I’ve retitled it. The term “science of reading” is both widely used and widely misunderstood today, so I have provided considerable revision here – though the overall points are the same. If you want to see the original, it is linked at the bottom (along with the more than 50 comments it elicited).
I talk a lot about research in this space.
I argue for research-based instruction and policy.
I point out a dearth of empirical evidence behind some instructional schemes, and champion others that have been validated or verified to my satisfaction.
Some readers are happy to find out what is “known,” and others see me as a killjoy because the research findings don’t match well with what they claim to “know.”
Members of this latter group are often horrified by my conclusions. They often are certain that I’m wrong because they read a book for teachers that had lots of impressive citations contradicting my claims. Sometimes these days, it doesn’t even take that. A few nasty assertions on X can do the trick.
I’m often challenged with statements like, “I thought that was part of the science of reading.”
What is clear from these exchanges is that many educators, legislators, journalists, and parents don’t know what research is, why we should rely on it, or how to interpret research findings. That, perhaps, is not surprising but it does make these very important people susceptible to groundless claims and false promises.
Research is used to try to answer a question, solve a problem, or figure something out. It requires the systematic and formal collection and analysis of empirical data – using methods able to provide the kind of answer being sought. That means not all research can be used to provide a valid answer to all questions. Even with that, research can never prove something with 100 percent certainty. It is useful for reducing uncertainty, for increasing the likelihood that we will get something right – it cannot guarantee it.
“Systematic and formal” means that there are rules or conventions for how data in a research study should be handled; the rigor of these methods is what makes the data trustworthy and allows research to reduce uncertainty. If a researchers want to compare the effectiveness of two instructional approaches, they must make sure the groups to be taught are equivalent at the beginning. Without that we can’t be sure which method did best. Likewise, we are more likely to trust a survey that defines its terms, or an anthropological study that immerses the observer in the environment for a long period of time.
Research reports do much more than just report the results or outcomes of a study, but they explain—preferably in detail—how those results were obtained. Most people may find such detailed description of methods to be mind-numbingly boring, but it is that detail that allows us to determine what a study really means and how much weight to place on it.
Here are some simple guidelines to remember.
1. Just because something is written, doesn’t make it research.
Many practitioners think that if an idea is in a book or magazine (or even just comments on social media) that it is research. Some even think this blog is research. It is not, and neither is the typical Reading Teacher article or Heinemann book.
That’s not a comment on their quality or value, but a recognition of what such writing provides. In some cases, as with my blog, there is a serious effort to summarize research findings accurately, and to distinguish opinions (what someone believes) from research findings (what we know).
Many publications for teachers are no more than compendia of opinions or personal experiences, which is fine. However, these suffer all the limits of that kind of thing. Just because someone likes what they’re doing (e.g., teaching, investing, cooking) and then writes about how well they’ve done it… doesn’t necessarily mean it’s great. That’s why 82% of people believe they’re in the top 30% of drivers; something that obviously can’t be true. (Often teachers will tell me they are getting great results doing something that contradicts my research summaries. I always request supporting test data that demonstrate the claim. Not surprisingly, they almost never write back and when they do it is not to offer any proof.)
As human beings we all fall prey to overconfidence, selective memory, and just a plain lack of systematicity in how we evaluate what we are doing and its impact.
Often teachers tell me that kids now love reading because of the teaching they have provided them. I ask how they know. What evidence do they have? Usually the answer is something like, “A parent told me that their child now likes to read.” Of course, that doesn’t reveal how the other 25 kids are doing, or whether that parent is a good observer of such things, or even what the motivation may have been for the praise.
Even when you’re correct about things improving, it’s impossible—from personal experience alone—to know the source of the success. It could be the teaching method, or maybe just the force of your personality. If another teacher adopted your methods, things might not be so magical.
And, then there is opportunity cost. We all struggle with this one. No matter how good an outcome, I can’t possibly know how well things might have gone had I done it differently. The roads not traveled may have gotten me someplace less positive—but not necessarily. You simply can’t know.
That’s where research comes in… it allows us to avoid overconfidence, selective memory, lack of systematicity, lack of reliable evidence, incorrect causal attribution, and the narrowness of individual experience.
When I first published this in 2018, my concern was that many teachers were placing too much confidence in first-hand accounts of other teachers and were too likely to buy into junk from Teachers Pay Teachers. It was not that I didn’t believe teachers could gain worthwhile ideas from such sources. My concern was that teachers were often placing these sources above sound research results.
If an article recommends a technique that the authors claim worked for them, I might want to give it a spin in my own classes to see what I think. But I’d have much more confidence in this if someone had objectively evaluated it.
How have things changed over the past several years?
I’m noticing that teachers are getting a lot more advice about how to teach from journalists, social media influencers, and scientists who may study reading but not reading instruction. These sources usually have little or no experience teaching reading and their knowledge of the relevant research and their ability to evaluate it are questionable at best. I don’t mind when these folks make recommendations as to what they want kids to learn (everybody has a vested interest in such curricular issues), but the best ways to teach something? Nah!
My 2026 take on this is: For the best advice on how to teach reading, turn to studies that explore the effects of reading instruction and what works best. Teachers’ advice based on their own experiences in delivering such instruction can’t replace the research studies but can provide valuable practical information that would never make its way into a research study. The value of advice on how best to teach reading from folks who don’t teach reading and who aren’t relying on instructional research? Not so much.
2. Research should not be used selectively.
Many educators use research the same way advertisers and politicians do—selectively, to support their beliefs or claims—rather than trying to figure out how things work or how they could be made to work better.
I wish I had a doughnut for every time a school official has asked me to identify research that could be used to support their new policy! They know what they want to do and seek research that will allow them to say their idea is a good one. What they should be doing is turning to the research first and then using that to determine the best way forward.
Cherry-picking an aberrant study that matches one’s claims or ignoring a rigorously designed study in favor of one with a preferred outcome may be acceptable debater’s tricks but bad science. That can only lead to bad instructional practice.
When it comes to determining what research means, you must pay attention not just to results that you like. Research is at its best when it challenges us to see things differently.
I vividly remember early in my career when Scott Paris challenged our colleagues to wonder why DISTAR, a scripted teaching approach was so effective, even though most of us despised it. Clearly, we were missing something. Our theories were so strong that they were blinding us to the fact that the approach to teaching that we didn’t like was providing something beneficial for kids—at least for some kids or under some conditions (the kinds of things that personal experience can’t reveal).
Anything to add to this in 2026?
These days my concern is not just with cherry-picked evidence but with evidence inappropriate to the question. Recently, I was speaking with legislators who were telling me about the importance of phonics instruction because of the studies of how the brain works.
Those studies are fascinating and have value, but their value is not that they reveal how we should teach phonics, how much phonics to teach, which programs to use, which curricular sequence to focus on, what activities to use and so on.
These studies provide us with a possible explanation for why phonics instruction has been successful in so many instructional studies. Who knows, someday this brain work might even suggest new ideas for designing or redesigning some of that instruction, but we won’t know the benefits of such invention until someone tries it out in classrooms in the context of a rigorous instructional study. All those things mentioned in the previous paragraph about how to teach decoding is something that those neurological studies can’t tell us. Instructional studies need to be the source of the determinative evidence as to how instruction should proceed – if we are really trying to reduce our uncertainty about how best to increase reading achievement. Legislating instructional practice based on brain scans is dopey.
3. Research, and the interpretation of research, require consistency.
Admittedly, interpreting research studies is as much an art as a science. During the nearly 60 years of my professional career, the interpretation of pedagogical research has changed dramatically.
It used to be entirely up to the discretion of each individual researcher as to which studies they’d include in a review and what criteria they’d use to weigh these studies.
That led to some pretty funky science: research syntheses that identified only studies that supported a particular teaching method or inconsistent criteria for impeaching studies. What that would mean is that a study would be ignored because it suffered from a serious design flaw unless its findings matched with what the researcher wanted to find – then that design flaw was no longer a problem.
Since I wrote this, research has found long-term negative consequences for Reading Recovery (May, Blakeney, Shrestha, Mazal, & Kennedy, 2024). Studies have long reported short-term gains (e.g, May, Gray, Sirinides, Goldsworthy, Armijo, Sam, Gillespie, & Tognatta, 2015; Shanahan & Barr, 1995), but the long-term harm of the approach is greatly concerning.
When this blog first posted, those long-term consequences were unknown. At that time, it was common to hear the short-term positive results of Reading Recovery dismissed because it was taught one-to-one (“anything works if it is taught one-on-one,” critics would say).
That dismissal disturbed me, not because I was a big Reading Recovery fan (I was always skeptical), but because the folks who were pooh-poohing Reading Recovery in that way, were simultaneously embracing the phonics results of the National Reading Panel (NICHD, 2000). The problem with this? Many of the studies that reported strong positive results had evaluated one-on-one phonics instruction, which didn’t seem to bother these critics at all.
I’ve retained this paragraph from the 2018 version: “I’m persuaded that both phonics and Reading Recovery work (because they both have multiple studies of sufficient quality showing their effectiveness). That doesn’t mean I think they work equally well, or that they are equally efficient, or that they even accomplish the same things for students.”
That paragraph still makes the important original point that we need to evaluate all research and research results using the same standards and criteria. But it now also reveals the fact that what we think we know from research may change over time. New studies may overturn previous research-based practices. That is not a problem, that is a strength of science.
In the arguments over the “science of reading,” phonics opponents often have argued that it is inappropriate to require phonics instruction since there is no such thing as “settled science“ (e.g., Tierney & Pearson, 2024). They’re not wrong about the nature of science – new better research can always alter our perspectives and practices and in that sense science can never be settled.
Nevertheless, that does not mean we cannot proceed to apply science to our lives. No physician would withhold today’s most effective cancer treatments because there is no such thing as “settled science.” We apply the best knowledge that we have and continue through science – not rhetoric – to keep adding to what we know, even if it means that in the future we will need to alter our course (which is a good reason to resist overly prescriptive legislation even if it appears to be consistent with today’s scientific findings, since up the road science may make such legislation problematic).
If we want to do better in reading education, we’re going to need to rely on science. That means valuing science over loud opinionizing, evaluating scientific information consistently, and rejecting arguments that there is never enough research to dictate our actions.
Let’s do better.
References
May, H., Blakeney, A., Shrestha, P., Mazal, M., & Kennedy, N. (2024). Long-term impacts of Reading Recovery through 3rd and 4th Grade: A regression discontinuity study. Journal of Research on Educational Effectiveness, 17(3), 433–458. https://doi.org/10.1080/19345747.2023.2209092
May, H., Gray, A., Sirinides, P., Goldsworthy, H., Armijo, M., Sam, C., Gillespie, J. N., & Tognatta, N. (2015). Year one results from the multisite randomized evaluation of the i3 scale-up of Reading Recovery. American Educational Research Journal, 53(3), 547-581.
National Institute of Child Health and Human Development. (2000). Teaching Children to Read: An Evidence-Based Assessment of the Scientific Research Literature on Reading and Its Implications for Reading Instruction: Reports of the Subgroups. Bethesda, MD: National Institute of Child Health and Human Development, National Institutes of Health.
Shanahan, T., & Barr, R. (1995). Reading recovery: An independent evaluation of the effects of an early instructional intervention for at-risk learners. Reading Research Quarterly, 30(4), 958–996. https://doi.org/10.2307/748206
Tierney, R. J., & Pearson, P. D. (2024). Fact-checking the science of reading: Opening up the conversation. Literacy Research Commons. https://literacyresearchcommons.org
Comments
See what others have to say about this topic.