"Which is more important for reading comprehension, knowledge or strategies?"
... does this question make sense without more context?
“Which is more important for reading comprehension, knowledge or strategies?”
Many readers of this post are likely aware of the controversy surrounding this question. For those unfamiliar, be assured it has caused some discord.
But a question such as this, or a statement that answers it, needs a shared context for it to make sense. Unfortunately the “knowledge vs. strategies” debate is often held in the absence of a shared context, and therefore the question verges on being nonsensical.
For example, are we talking about comprehension of informational text or narrative? If informational, is the paper highly technical? If narrative, is it placed in a familiar modern-day setting or is it historical fiction? If the latter, is an understanding of the historical setting critical to following the story?
Is the text high- or low-cohesion? Are we talking about ‘world knowledge,’ or any knowledge at all? If the former, what counts as ‘world knowledge’? Does vocabulary count as knowledge? How are we defining and measuring comprehension? And what are we even counting as a ‘strategy’?
To discuss the relative contributions of knowledge and strategies to reading comprehension, at least some of the items above need to be identified. This article fleshes out the various contexts in which reading comprehension occurs and how these contexts alters the relative contributions of knowledge and strategies to understanding a text.
Additionally, this article posits that knowledge and strategies should not be pitted against each other, but rather that the focus should be on examining how they work together to elevate learning.
To dive into this, the remainder of this article is outlined as follows:
Important Considerations
Part 1: What does ‘knowledge’ mean within this conversation?
Part 2: What are reading comprehension strategies?
Part 3:
What is more important, knowledge or strategies?This question needs context!Context 1: Narrative or expository text?
Context 2: Where’s your knowledge baseline?
Context 3, 4, & 5: How are we measuring comprehension? What about text coherence? Does the text length matter?
Context 6: What do we mean by all these terms, and do we actually disagree?
Part 4: Knowledge and strategies need each other.
Final Thoughts.
Important considerations
This article focuses on knowledge and skills because that is where the debate has been centered. However, the ability to understand a text emerges from many processes including a student’s ability to decode, understand syntax, and read with fluency. Vocabulary knowledge is also critical, and while this is technically ‘knowledge’ it tends to be discussed separately from content knowledge (see more below).
Some views on reading comprehension and its instruction can be found here, here, here, and here.
Part 1: What is knowledge?
When referencing knowledge-building, the word ‘knowledge’ is typically used as a shorthand for ‘academic’ or ‘world knowledge.’ The term often, but not always, focuses on the content knowledge students would learn in science or social studies. It takes a broad view of these subjects – art history and Greek myths, for example, could be considered under this umbrella.
‘Knowledge’ encompasses academic information a child could learn from a parent, through a camp or after-school program, from a trip to a museum, from books read at home, or from a variety of other sources.
Knowing that plants use sunlight to create their own food or understanding the tensions that led to World War I would be examples of ‘knowledge.’ Understanding the properties of different paints or what makes clay superior to dirt for making pottery would also be in this category.
Within the context of knowledge-building, I argue that ‘knowledge’ typically does not include what we would consider ‘common knowledge’ for a child. Examples of common childhood knowledge would be knowing what a playground is or knowing that dogs are kept as pets.
Clearly this issue can get muddy very quickly. For example, I grew up in New England and all kids had experience with snow. But in San Diego, most kids have not. Knowing the properties of snow may be considered ‘common knowledge’ in one location, but not another. Similarly, knowledge that is ‘common’ within a cultural or religious group is not necessarily well-known by those outside the community.
Do we need to determine the boundaries of what is and isn’t common knowledge? Well, if you are trying to determine the relative importance of ‘knowledge’ to reading comprehension, the definition does become important.
While most studies are clear about what type of knowledge they are measuring, the word is not defined in most discussions on the topic. Without a shared definition of what ‘knowledge’ means, conversations on this issue will be muddled.
Importantly, despite frequent mischaracterizations, knowledge does not mean a “list of facts” (Kirschner, 2025). Gaining basic factual information is a first step, but the ultimate goal of knowledge-building is deep understanding.
Reviews and overviews of knowledge-building can be found here and here.
What about vocabulary?
Vocabulary is often treated as separate from “knowledge.” However, I question this distinction. We literally talk about “vocabulary knowledge,” and what is knowing what a word means if not a form of … knowledge?
I especially question attempts to separate vocabulary knowledge from content knowledge when it comes to science text. ‘Depth’ of vocabulary, not just breadth is critically important (Qian, 2008). Deep understanding of many science words requires conceptual understanding. How do we separate teaching the meaning of ‘magnetic’ from teaching which materials are magnetic? Or for social studies, how do you separate teaching the definition of ‘democracy’ from teaching about democracies?
Nonetheless, since vocabulary is typically treated as distinct from content knowledge, I (mostly) follow that separation here.
Part 2: What are reading comprehension strategies?
I am just starting to truly dig into the research on reading comprehension strategies and it is a bit overwhelming. Please note that I am far from an expert.
My best understanding of what falls under the banner of ‘reading comprehension strategy’ would be an action or process that a student can choose to do when they encounter a text to enhance their comprehension.
I’ve seen pro-strategy arguments include discussion about things like fluency, but this is more of a teaching strategy. While poor fluency will impede reading, a student cannot sit down with a text and say, “I’m going to read more fluently so I can understand this text.”
A student can however, sit down with a passage and say, “I’m going to identify what type of text this is. It’s informational. Now I’m going to think about why the author wrote this article.”
Some students may employ strategies automatically, particularly for text that is accessible to them. However, other students won’t. And there is plenty of research showing positive effects from strategy instruction. While aspects of this research are under debate, even those calling for substantially less strategy instruction agree that some is effective (e.g. Willingham 2006).
It’s sometimes said that strategy instruction should not start until upper elementary, but programs like Story Champs have well-documented success teaching story and text structure (a strategy) to early elementary and even preK students.
Similar to how we cannot view knowledge-building as teaching “a list of facts,” it’s unfair to view strategies only within the context of “strategy of the week” instruction. The most impactful modes of strategy instruction may look quite different from what many are used to (for example, the “PACT” approach, see Vaughn & Wanzek, 2023). And of course, we need to move away from instruction that uses scattered expository (i.e. informational) texts.
Some or many commonly taught strategies are also likely effective. Literacy has been called a curricular “bully,” and strategies have been the biggest bully of all. It’s natural to want to banish a bully, but we should instead look to find the highest-impact strategies and balance instructional time to adequately address all components of reading comprehension.
Part 3: What is more effective, knowledge or strategies? This question needs context!
Any discussion on the relative contributions of knowledge and strategies to reading comprehension requires a shared context. Otherwise the question becomes meaningless. In this section, multiple important contexts are discussed.
A note that the studies below are looking at reading skill, which will include aspects of reading that cannot be taught, like working memory.
Context 1: Narrative or expository text?
At the very least, the text in question needs to be identified as expository, narrative, or narrative with a high degree of domain-specific knowledge. Expository texts are generally considered more difficult to comprehend and more reliant on background knowledge (e.g. Clinton et al, 2020; Mar et al., 2021).
Domain-specific knowledge is predictive of performance on domain-specific texts. Numerous studies support the idea that knowledge is one of the biggest, if not the biggest, contributors to comprehension of expository text (e.g. Marr & Gormley, 1982; Recht & Leslie, 1988; Best et al., 2008; Ozuru et. al, 2009; Härtig et al., 2022).
You may be familiar with “The Baseball Study” (Recht & Leslie, 1988). Here, middle schoolers read a text that heavily referenced baseball knowledge. (Note this was a narrative story, but with a heavy domain-specific component). Each student was evaluated on two features: whether they had “high” or “low” knowledge about baseball, and whether they were a “high-ability” or “low-ability” reader.
The researchers found that reading skill barely affected comprehension of the text. Students in the high-knowledge group had the highest scores regardless of their reading ability. Likewise, students with low baseball knowledge did poorly, even if they were high-ability readers.
This study is a useful exemplar for introducing the idea that knowledge is important for reading comprehension. However, the results also suggest that knowledge is the only important factor.
While a vast number of studies confirm that knowledge is critical component of comprehension, nearly all other studies show that reading skill is also an important component.
For example, Adams et al. (1995) did a similar study using narrative passages with heavy football references. They split students into the same four groups (here called high-skill, high-knowledge (HS-HK); high-skill, low-knowledge (HS-LK); low-skill, high-knowledge (LS-HK) and low-skill, low-knowledge (LS-LK)).
As might be predicted, the “high-skill, high-knowledge” group did the best on comprehension measures. However, in contrast to the baseball study, the scores of the remaining groups were roughly indistinguishable from each other. Unlike the baseball study, “high-knowledge” did not fully make up for “low-skill.”
Also notable, the LS-LK group had the slowest reading speed, while the other three groups were roughly equal, again showing that both knowledge and skill contributed to reading success.
What about “domain-general” narrative text?
There’s less research examining the effect of knowledge on “domain-general” narrative text for obvious reasons. What knowledge are you going to measure if the text isn’t focused on any particular domain? However, a few studies offer insight.
The Adams et al. (1995) “football study” had a control group that read a story about someone being rescued from a fire. Here, as you might predict, knowledge of football had no correlation with reading comprehension of this text and the two high-skill groups scored the best.
In another study (Best et al., 2008), third graders’ “world knowledge” was more predictive of comprehension on two expository texts (titled “Heat” and “Needs of plants”) than decoding skills. In contrast, decoding ability was found to be the main predictor for comprehension of the narrative text.
The take-away from this body of research is that “world knowledge” is very important for expository text, but far from the only contributor to comprehension. In contrast, for narrative text set in familiar settings and with plots about familiar issues, “world knowledge” will correlate much less with successful comprehension.
Another note, as papers get more technical, logically, the relative contribution of background knowledge will increase. If you have no background in engineering, you won’t be able to engage with a technical engineering paper no matter how skilled you are.
What about vocabulary knowledge?
Vocabulary knowledge is important for all text types. If you include vocabulary knowledge in your definition of ‘knowledge,’ then ‘knowledge’ will one of the main predictors for comprehension of all texts, even domain-general narrative ones.
Conclusions for Context 1
We can conclude that the relative importance of knowledge to reading comprehension will vary based on whether the text is narrative or expository, how technical the text is, and whether you include vocabulary knowledge in your definition of “knowledge.”
Context 2: Where’s your knowledge baseline?
In the Best et al. (2008) study described above, the researchers concluded that third graders’ “world knowledge” was more predictive of comprehension on expository texts than their decoding skills.
What does this mean exactly? Would it be fair to take these results and conclude “knowledge is more important than decoding for comprehending expository text?”
No, that would be ridiculous. Clearly, students in this study had a baseline decoding ability.
We can say this study provides evidence that, once children reach a certain proficiency with decoding, “world knowledge” will have a bigger impact on comprehension of expository text than further decoding practice.
Below some decoding baseline, however, it obviously doesn’t matter how much knowledge students have. If they can’t read the words in a text, they will not be able to comprehend the text.
There’s a parallel here to knowledge. There is a baseline level of knowledge needed to read texts. For simple narrative texts, the necessary baseline level may simply be common childhood knowledge (and vocabulary). For these simple texts, students are on an “even knowledge playing field,” and their understanding of the text will be driven by differences in other areas like comprehension skills, fluency, and working memory.
However, for expository texts and narrative texts with a heavy knowledge component, the baseline level of knowledge needed for comprehension is higher. The more technical the text, the more background knowledge you need.
If a seventh-grade student is sitting in history class reading about World War I, but they only vaguely understand that Europe is a different continent filled with many different countries, and they are learning about most of those countries for the first time, they are going to have a hard time with it. If they are being taught about the bombing of Pearl Harbor but don’t know that Hawaii is both a U.S. state and an island located far from the continental U.S., their working memory is going to be taxed and their comprehension reduced.
I’m curious if you, the reader, think these examples are hyperbolic. I’m truly not sure what people think is a common amount of knowledge for a middle-school student in the U.S..
Far too many students receive little cohesive content
Certainly, there are many strong schools with students of all demographics that teach rich content. However, there are far too many schools that teach students nearly no cohesive content. For example, in a 2017 survey, 29% of elementary principals said they had adopted no social studies curriculum (Diliberti et al., 2023) and in another, albeit older, survey, only 11% of California elementary principals said it was likely students would receive high-quality science education at their school (Dorph et al., 2011). The science instruction is so paltry that no one even attempts to lie about it.
Unfortunately, there’s a correlation with higher poverty and less content (Dorph et al., 2011; Gonring et al., 2021; Smith et al., 2021; Diliberti et al., 2023; personal observations from providing science lessons at numerous local schools). I’ve personally seen many early elementary classrooms with effectively no social studies or science curriculum. Most of the day at many of these schools is spent on the Benchmark ELA curriculum. Students receive only the scattered content within this curriculum and whatever each individual teacher chooses to supplement with. (This is changing slightly, but not enough).
In an informal survey of some early fourth graders we work with, about half of them could not identify California as a state on a multiple-choice quiz. And we live in California. I had several third graders in multiple classrooms who did not know what a “magnet” was prior to instruction. It is not the students’ fault that they are not being taught content. We cannot assume students will learn content by osmosis.
I suspect that those who are knowledge-hesitant see the baseline of “common childhood knowledge” as much higher than it is. They might work in better-resourced schools or live outside the US. Or maybe they only interact with students for literacy research studies and don’t give much thought to what else the students know. Maybe they have a niece or grandchild in upper elementary who knows a lot about the world and subconsciously serves as their reference. If you assume students already possess a lot of “world knowledge,” the urgency of knowledge-building programs might be unclear.
The conclusion of the Best et al. 2008 study described above - that knowledge is more predictive of comprehension than decoding skill - only makes sense if we all understand that there’s a baseline of decoding ability that students in that study possess.
Likewise, it’s hard to have meaningful conversations about the importance of knowledge for reading comprehension, (and thus the importance of high-quality knowledge-building programs), if we have a differing understanding of how low students’ “knowledge baseline” can be.
So, how important is knowledge compared to strategies? Well, if you have almost no academic knowledge and you are expected to read academic texts in school, then lack of knowledge probably does become “the most important” contributor to poor reading comprehension.
Making this point one more time
This is one of the most important points and one I’ve struggled to articulate. So let’s put it this way: We know one’s hips, knees, and ankles are all important for running. You can argue which body part is “more important” for movement, but if you break your ankle, it’s the ankle that prevents the running. For many students, knowledge is their “broken ankle.” And you heal the bone that’s broken.
Or perhaps an even better analogy is to imagine you have injuries and pain in all your joints and muscles, and you give all the body parts attention except the ankle. Then someone says, “Yes, we know the ankle is so important for locomotion. And we have data that treatment X is very helpful for walking and probably jogging, but the data is weak that treatment X helps with sprinting. Therefore, there’s no data that treatment X helps with ‘general locomotion.’ And yes, we could spend our time and influence advocating for making treatment X better, but instead we’re going to say that the people focusing on treatment X trust themselves too much.”
Conclusion for Context 2:
In discussions about the importance of background knowledge for reading comprehension, we must have a shared understanding of a “knowledge baseline.”
Context 3, 4, & 5: How are we measuring comprehension? What about text cohesion? Does the text length matter?
Comprehension can be measured in a variety of ways, and the ways in which it is measured are likely to affect the “relative importance” of knowledge and strategies.
For example, Kraal et al. (2018) used different types of reading comprehension questions with second grade students. The students did better on the “text-connecting” and “gap-filling” questions in the narrative text than in the expository text. However, there was no difference in their ability to answer “literal” questions between the two genres.
Ozuru et al. (2009) found complex relationships between reading skill, domain-specific background knowledge, text cohesion, and question type in a study with college students reading biology passages. (Text cohesion can be thought of as how well the text helps guide you through the material. In a high-cohesion text, you will need to make fewer inferences).
The high-knowledge groups scored the best for every question type, with reading skill also clearly playing a role in comprehension. However, for the “global-bridging questions,” the effect of reading skill was smaller (and non-significant) than the effect of skill on “text-based” and “local-bridging” questions. Furthermore, while they found a small benefit of high-cohesion for skilled readers generally, the high-cohesion text actually had a negative effect on low-skill but high-knowledge readers. (This latter result is likely because high-cohesion can make the text longer and more linguistically complex, and may also be related to the expertise reversal effect).
Another note about the Ozuru et al. (2009) paper is that the texts and questions were relatively technical, which of course also influences their results. (One sample question: “How do ectothermic animals secure heat?”)
Text length may also affect relative contributions to reading comprehension. Härtig et al. (2022) found that the influence of vocabulary knowledge on reading comprehension varied based on passage length. They said that possibly “students can handle comprehension of shorter text with general skills, but it becomes harder with longer text.”
In contrast, revising texts to make them higher-cohesion (and less dependent on background knowledge) makes text longer (Beck et al., 1991). So the effects of text-length will likely vary depending on the conceptual density of the text.
There are more studies with similar types of questions and results. The point of including this information is not the details per se, but to demonstrate the complexity of the relative contributions of knowledge and skills for comprehension.
Conclusion for Context 3, 4, & 5
Differences in text features like text-cohesion and text-length, as well the mode of comprehension measurement will impact the relative contributions of knowledge and skills to reading comprehension.
Context 6: What do we mean by all these terms, and do we actually disagree?
The terms ‘skills’ and ‘strategies’ are often used interchangeably but one can easily find a difference between teaching “reading skills” and “reading strategies.”
Even taking away the knowledge aspect, there are still many factors that contribute to reading skill. We have word recognition, fluency, language or syntax comprehension, and working memory in addition to the strategies that students can employ.
Working memory, as far as we know, is fixed. There no known way to improve this.
There is widespread agreement that it’s important to work on decoding, fluency, and syntax and language comprehension. (The latter often suffers more from neglect than disagreement). I’ve yet to see a knowledge proponent who doesn’t believe these aspects of reading are important.
This article hasn’t even touched on the impact of writing on reading comprehension, but again everyone agrees structured writing is important.
My takeaway from knowledge proponents has always been that the main target is the “skill of the week” approach that still dominates elementary instruction. And strategy proponents also don’t seem thrilled with how strategy instruction actually plays out in most elementary schools.
Meanwhile, there are a non-trivial number of public-facing literacy experts who have described knowledge-building in ways that make it clear that they don’t have a strong grasp of the concept. In fact, more than once, advice has been given on how knowledge-building could be better, only for the advice to describe how knowledge-building already works.
All these things put together make me think that the substantive disagreements on teaching practices are actually relatively small. There are, however, large disagreements on messaging, and messaging is important. Still, we should be careful that we are not arguing over things we actually agree on.
Conclusions for Context 6:
We probably agree on more than we think.
Part 4: Knowledge and strategies need each other
While it’s a mistake to endlessly practice discrete reading comprehension skills on disconnected texts, it’s also a mistake to discuss reading strategies as though they exist in isolation and are irrelevant to learning content.
Personally, I started looking into comprehension strategies to better understand how to teach and design elementary science text.
Content and ELA cannot truly be separated. In order to most effectively use text as part of our content teaching, we have to understand how to best support our students in reading, understanding, and learning from complex text. This probably looks different than the strategy instruction seen in a typical elementary school. But it does include strategies.
A 2013 study by Vaughn et al. illustrates this concept well. In it, the authors compared a “content-focused” approach to text to a “strategy-focused” approach. They built on the work McKeown et al. (2009) who wrote that “comprehension enhancement might derive from a focus on continually striving for meaning as reading of the text moves along rather than considering when and how to call up specific routines to deal with new information.”
Despite the names, both of these approaches can fall under the ‘strategies’ umbrella. After all, “continually striving for meaning as you read” is a strategy a student can employ to make meaning of the text.
In the Vaughn et al. study, researchers used a content-focused text-based approach to social studies instruction called PACT (Promoting Adolescent Comprehension of Text) in a middle school social studies classroom. While the approach included measures to learn content, the research team hypothesized that the PACT group and a control group would learn similar amounts of social students knowledge, as both received instruction on the same topics.
Vaughn et al. instead hypothesized that the PACT group would increase in their comprehension of domain-specific text compared to a control group.
However, the treatment group improved in both measures. What I find so interesting about this is that the research group was ostensibly interested in finding out how to improve reading comprehension without hurting content instruction, but ended up lifting both. The research group has continued to publish on this promising approach (e.g. Vaughn et al., 2024, 2025).
Final thoughts
While criticism of the “strategy-of-the-week” approach is valid, the “knowledge vs. strategies” debate has outlived its usefulness.
Both knowledge and reading skill contribute to reading comprehension, with their relative contributions varying depending on the type of text, text coherence, how you define knowledge, how you measure comprehension, and more. Moreover, to have a meaningful discussion of the importance of knowledge on reading comprehension, a shared understanding of the extent of the content crisis is needed.
Instead of knowledge and strategies being pitted against each other, we should better appreciate how knowledge and strategies enhance each other to promote rich academic experiences for our students.
Thanks for this analysis. The only point I have reservations about is the conclusion for "Context 6"--i.e., that we probably agree on more than we think.
That may be true about some things, but there are reading experts out there who say that, sure, building knowledge is fine, but there's no evidence that it improves reading comprehension. So it's okay to build knowledge in social studies or science, but it has no place in the two-hour literacy block. (This is an argument that, for example, Tim Shanahan has made.)
In practice, this often means that kids don't get much access to knowledge at all, because the reading block takes up so much of the school day -- sometimes more than 2 hours, especially in schools where test scores are low (which are also the very schools serving kids who need in-school access to knowledge the most).
The answer to this question is yes.