reading comprehension research paper

Precursors of reading text comprehension from paper and screen in first graders: a longitudinal study

• Open access
• Published: 08 October 2022
• Volume 36 , pages 1821–1843, ( 2023 )

Cite this article

You have full access to this open access article

• Elena Florit   ORCID: orcid.org/0000-0002-9130-459X 1 ,
• Pietro De Carli   ORCID: orcid.org/0000-0001-6538-2175 2 ,
• Antonio Rodà 3 ,
• Samantha Domenicale 4 &
• Lucia Mason   ORCID: orcid.org/0000-0001-7134-0510 4  

3238 Accesses

3 Citations

Explore all metrics

Research on text comprehension in relation to the reading medium (paper or screen) has mainly involved undergraduate or high school students. To advance current knowledge on the effects of reading medium, this longitudinal study focused on beginner readers, specifically, the role of precursors in first graders’ comprehension of narrative and expository linear texts from reading on paper and computer screen. Working memory and inference skills as cognitive precursors and basic digital skills were measured at the end of preschool (T1); reading text comprehension and word reading, as a control variable, were measured at the end of the first grade (T2). Sixty-three children participated in total. The first graders read four texts, one narrative and one expository, on both paper and computer screen, in a counterbalanced order. Results showed no main effects of the reading medium or text genre, but the interactive effect of these variables was significant. At T2, the children had higher comprehension scores after reading narrative than descriptive texts from paper. In addition, reading from the screen was preferred at post-test, after all texts were presented. As precursors, working memory and inference skills predicted both printed and digital text comprehension. In contrast, basic digital skills predicted only digital text comprehension after controlling for medium, text genre, and word reading.

Similar content being viewed by others

Twice Upon a Mind: Preschoolers’ Narrative Processing of Electronic and Printed Stories

Dilek Altun

The Contribution of Reading Abilities to the Writing Quality of Expository Text Structure in Hebrew Speaking Elementary School Children

Reading digital- versus print-easy texts: a study with university students who prefer digital sources

Noemí Bresó-Grancha, María José Jorques-Infante & Carmen Moret-Tatay

Avoid common mistakes on your manuscript.

Children born in the digital age, known as digital natives (Prensky, 2001 ), are growing up in an environment saturated by digital media, in which they interact with technologies (Hisrich & Blanchard, 2009 ). These technologies expose them to a variety of digital texts compared to more traditional texts on paper. The use of digital texts during storybook reading is constantly increasing from preschool onwards (Furenes et al., 2021 ; Lauricella et al., 2014 ). When children enter primary school, they also start using digital devices for learning activities, including learning to read. The present study focuses on text comprehension from paper and computer screen in first graders who can read independently.

When considering text comprehension in relation to reading medium, recent meta-analytic research has shown that, at different educational levels, students comprehend better when reading on paper than on screen, although they may prefer digital reading (Clinton, 2019 ; Delgado et al., 2018 ; Kong e al., 2018 ). In addition, meta-analysis has shown that children’s comprehension of printed narrative books read by an adult exceeds their comprehension of digital narrative books read to the child by a voiceover (Furenes et al., 2021 ). However, meta-analyses on the relationship between reading comprehension and reading medium are based on only a very few studies with primary school students. Therefore, we know little about the youngest readers and the precursors that can sustain reading text comprehension from paper and screen. There is no cohesive theoretical framework that concurrently describes the various precursors involved in reading from different mediums (Dahan Golan et al., 2018 ). However, it is likely that, in reading comprehension from screens, it is not only the well-known cognitive factors that are involved but also skills related to the use of digital devices.

This longitudinal study, part of a wider research project on traditional and digital reading, had two goals. First, we sought to determine whether, in beginner readers at the end of the first grade, text comprehension differentiates across printed and digital mediums when considering both narrative and expository texts, and whether children prefer reading on paper or screen. Second, we sought to understand whether precursors assessed at the end of preschool—specifically traditional literacy skills (i.e., cognitive abilities: working memory and inference skills) and digital skills (i.e., basic digital skills such as navigation skills)—sustain text comprehension from paper and screen, and account for any differences in text comprehension across mediums, after controlling for word reading skills.

Reading from paper or screen?

Whether reading on paper or screen, text comprehension is a coherent mental representation of the meaning of a text. According to the well-known construction-integration model of text comprehension (Kintsch, 1998 ), such a mental representation is formed when textual propositions (the textbase representation or a more superficial representation) are integrated with the reader’s background knowledge to construct a global and deeper representation of the meaning of the text. Text comprehension, therefore, is the outcome of an active, complex process in which both information explicitly presented in the text and implicit information inferred by the reader must be comprehended.

Today, both printed and digital texts are read increasingly frequently. Therefore, a crucial question arises: are reading from paper and reading from screen equal for text comprehension? In recent decades research has flourished on meaning construction from text in relation to reading medium. Four meta-analyses have been published in the last four years, synthesizing accumulated evidence. Three of these works (Clinton, 2019 ; Delgado et al., 2018 ; Kong et al., 2018 ) involved independent readers at all educational levels, mainly university students. Kong and colleagues  ( 2018 ) reported that comprehension is superior when reading from paper. Delgado and colleagues ( 2018 ) confirmed the inferiority of comprehension when reading expository texts digitally, in particular under a limited time frame. Clinton ( 2019 ) again documented superior comprehension for expository texts on paper, for both literal and inferential reading performance. No differences were revealed for narrative texts. The most recent meta-analysis (Furenes et al., 2021 ) analyzed whether digital books with different types of enhancements can support the story comprehension of children who cannot read independently (mainly aged 4–5 years) without reliance on adults. They showed that children’s comprehension of stories was better in printed books than in digital books when the latter included only the voiceover or highlighted print as additional features, and not enhancements that target story content. No differences emerged for book genre (fiction or non-fiction), which did not moderate the effect of medium.

Why does the inferiority of screen emerge as a shared issue across meta-analyses? Several hypotheses have been proposed, two of which are relevant for the present work. The first, offered by studies of independent readers, is the ‘shallowing hypothesis’ (Annisette & Lafreniere, 2017 ). It posits that readers process digital text in a superficial and incomplete way, due to their frequent use of digital devices based on rapid and fragmented interactions, for purposes other than learning. The more students are accustomed to quick interactions with digital devices, the less they can use them in challenging tasks that require focused and sustained attention, such as comprehension of expository texts (Delgado et al., 2018 ). That is, acting as a contextual cue, a digital medium may prime a reader to process the text as if for leisure; in contrast, paper may act as a contextual cue to process a text as if for study and learning (Clinton, 2019 ).

The second hypothesis posits that digital reading may be affected by the higher cognitive load imposed by limited familiarity with devices and reading on screen, while readers have consolidated experience with traditional reading and reading strategies on paper (Ackerman & Lauterman, 2012 ). To explain the results with non-independent readers, Furenes and colleagues ( 2021 ) also proposed that children’s handling of digital devices (e.g., point, click, and swipe) to read enhanced digital books may require additional resources that, given the limited capacity of human information processing system (Mayer, 2009 ), cannot be allocated to the construction of the meaning of the story.

Note that the three meta-analyses on readers who can read independently are based mainly on studies that involved university or high school students with much accumulated experience of reading on paper. In the very few studies reviewed in the three meta-analyses that involved younger readers, those students were from the higher end of primary school. We have also identified a few other more recent investigations not included in the meta-analyses. All these studies used comparable printed and digital texts. They differed to some extent in the technical requirements imposed on the digital reading tasks, but they did confirm the screen inferiority effect in fifth graders (Halamish & Elbaz, 2020 ; Støle et al., 2020 ) and in fifth and sixth graders (Dahan Golan et al., 2018 ). Text comprehension on paper outperformed text comprehension on screen even for children who at pre- and post-test reported that they preferred reading digitally or had no clear medium preference (Dahan Golan et al., 2018 ; Halamish & Elbaz, 2020 ). These findings in children around the end of primary school might, arguably, be related to their lack of digital skills and experience. But Støle et al. ( 2020 ) pointed out that today’s 10-year-olds have fundamental digital skills and a lot of experience with digital devices that they also use for learning activities. According to Støle et al. ( 2020 ), a more plausible interpretation is that reading disruption caused by scrolling is likely to draw on the limited working memory capacity needed for comprehension (Sanchez &Wiley, 2009 ).

A recent study also involved first graders with considerable experience with technology at school (Florit et al., 2021 ). The screen inferiority effect did not emerge for main idea and literal comprehension. It was, though, present for inferential comprehension in children with low word reading skills when considering a composite score for narrative and expository texts. Participants in this study did not show a preference for reading on screen, probably because they did not predominantly use the digital medium for leisure and the traditional medium for learning, and so did not perceive reading on screen as more attractive than reading on paper.

Therefore, it remains an open question worth investigating whether the screen inferiority effect manifests even in beginner readers who can read independently and use digital devices (tablets or computers) not only for leisure but also at school for learning-related activities, and whether they prefer reading on paper or screen.

Individual factors in text comprehension of beginner readers

Traditional literacy skills include individual factors from preschool, before children learn to read, that sustain later reading comprehension on paper (e.g., Kim, 2017 ; Language and Reading Research Consortium & Chiu, 2018 ). These traditional literacy skills may be hypothesized to be also involved in reading comprehension on screen (Reading Study Group, 2002 ). Verbal working memory is a foundational cognitive skill. It is involved in the construction of the textbase level of comprehension, as readers must temporarily store linguistic information while they process and integrate it with new linguistic input. Working memory is also involved in constructing the situation model. The reader needs to maintain and manipulate information when she is integrating sentences and establishing coherence both locally and globally (Kim, 2017 ). Most studies, indeed, find a relation between working memory and reading comprehension (Ariasi & Mason, 2014 ; Cain et al., 2004 ; Carretti et al., 2014 ; Language and Reading Research Consortium & Logan, 2017 ; Oakhill & Cain, 2012 ; Seigneuric & Ehrlich, 2005 ). However, verbal working memory uniquely accounts for reading comprehension over and above the contribution of higher-level cognitive skills, such as inference skills, in some cross-sectional studies (Cain et al., 2004 ) but not in others (Language and Reading Research Consortium & Logan, 2017 ). Finally, working memory is not a unique longitudinal predictor of reading comprehension once the role of higher-level cognitive skills is considered (Oakhill & Cain, 2012 ). These mixed results may be explained by the fact that working memory supports the integrative processes involved in higher-level cognitive skills, such as inference skills (Currie & Cain, 2015 ).

When constructing a coherent mental model of a text, the child is required to generate two types of fundamental inferences. These are text-connecting inferences, which link propositions at a local level, and integrative inferences, which integrate information about the text with prior knowledge of the world (for a review, see Clinton et al., 2020 ). Research has documented that inference skills between 4 and 6 years predict subsequent reading comprehension (Lepola et al., 2016 ; Silva & Cain, 2015 ). There is also evidence that text comprehension can be promoted via training in preschool and primary school that targets inference skills (Bos et al., 2016 ; Dicataldo et al., 2020 ). Most of these studies considered narrative texts; inferences generation is also necessary, although more difficult, for expository texts comprehension (Clinton et al., 2020 ). To the best of our knowledge, studies comparing comprehension of printed and digital texts have considered the role of attention (Delgado & Salmerón, 2021 ), another foundational cognitive skill, but not verbal working memory and inference skills.

Our study also considered children’s basic digital skills, particularly behavioral skills rather than explicit factual knowledge about technology. In mature readers, basic digital skills are defined as the skills that enable them to understand and use the medium where the text is presented, such as a computer or tablet. In addition, while advanced readers are supposed to possess general knowledge about the structures and functionalities of technology, in younger readers or pre-readers, digital skills are more dependent on the medium used (e.g., computer or tablet). Digital natives do not necessarily possess these skills or spontaneously acquire them (Fajardo et al., 2016 ). Like traditional literacy skills, they develop from the preschool years (Neumann et al., 2016 ).

Studies on preschool children have documented possible advantages of being exposed to digital texts (e.g., digital storybooks) in addition to traditional texts in promoting the development of some literacy skills in printed text comprehension (Lauricella et al., 2014 ; Takacs et al., 2015 ). Yet, when considering digital texts, basic digital skills are crucial, for example, the ability to use a mouse, to press buttons to move back and forward, to click, and to scroll up and down. With readers in the fifth year of primary school, Fajardo and colleagues ( 2016 ) have shown that these basic skills play an important role in the comprehension of digital texts, independently of the readers’ comprehension of printed texts.

Finally, given our interest in beginner readers, we examined word reading skills as a control variable that is fundamental for printed and digital text comprehension when children are at the end of the first grade and have just learned to read (Florit et al., 2021 ; Kim, 2017 ; Tobia & Bonifacci, 2015 ).

The study: research questions and hypotheses

This longitudinal study aimed to extend current research on children’s reading from paper and screen in the crucial transition from preschool to the first grade. Specifically, relevant precursors such as working memory, inference skills, and basic digital skills were measured at the end of preschool (T1); digital and printed text comprehension and reading preferences, and word reading skills as a control variable were measured at the end of the first grade (T2). Comparable digital and printed texts included narrative and expository (i.e., descriptive) texts typically used from the first grade. The study was guided by two main research questions (RQ):

(1) Do reading medium (paper vs. screen) and text genre (narrative vs. descriptive) differentiate text comprehension in children at the end of the first year of primary school (T2)? Do beginner readers prefer reading on paper or on a computer screen?

(2) Do working memory, inference skills, and basic digital skills at the end of preschool (T1) predict text comprehension at the end of the first year of primary school (T2), considering both text genre and reading medium, and controlling for word reading skills? If differences emerge at T2 for comprehension in relation to reading medium and text genre, are they predicted by the precursors measured at T1?

For RQ1 we hypothesized a difference in text comprehension related to reading medium. That is, we expected lower comprehension for reading from the computer screen than from paper, since a higher cognitive load is required of children for digital behaviors, such as scrolling for reading text and questions and pressing buttons to move back and forward across pages for answering questions (Neumann et al., 2016 ; Støle et al., 2020 ). Moreover, concerning text genre, we hypothesized greater comprehension of narrative than descriptive texts because they are more familiar to beginner readers who have experienced them in preschool and at home during storybook reading (Furenes et al., 2021 ). Furthermore, based on recent meta-analyses, we hypothesized an interaction effect of reading medium and text genre. Specifically, we expected superior text comprehension when children read descriptive texts, but not narrative texts, on paper, as expository texts require a deeper level of content processing than narrative texts (Clinton, 2019 ; Delgado et al., 2018 ). We expected beginner readers to report a clear preference for reading on screen over reading on paper (Dahan Golan et al., 2018 ). As digital natives, indeed, their experience with digital devices and on-screen activities must be more pleasant than the one with printed materials.

For RQ2, we hypothesized that comprehension of texts presented on paper and screen at T2 would be predicted by inference skills at T1, after controlling for word reading skills. Predictions were less clear for working memory since mixed findings have been reported about its contribution to text comprehension on paper. However, we did not expect working memory at Time 1 to predict comprehension of texts presented on paper and screen at T2, because—as in a previous longitudinal study (Oakhill & Cain, 2012 ) — we included higher-level cognitive skills (i.e., inference skills) among predictors. Finally, we expected that basic digital skills would contribute to the comprehension of texts presented on screen but not on paper (Fajardo et al., 2016 ). Finally, we hypothesized that basic digital skills, but not working memory and inference skills, would predict the expected difference in text comprehension, due to reading medium, after controlling for word reading skills.

Design and participants

At T1, 72 Italian children participated, and one year later, at T2, 63 out of the 72 participants were involved. The participants in this study were included in a larger project on medium effects on text comprehension in beginner readers. However, the hypotheses, measures, analyses, and results presented in this paper are unique to this study. The children participated with their own verbal assent and their parents’ written consent. The mean age of the 63 children (F = 35) was 5.8 years ( SD  = 0.3) at T1 and 6.8 ( SD  = 0.3) years at T2. The 12.5% attrition between T1 and T2 was due to their families having relocated, absence on the day of the assessment session, and lack of written consent from parents. Of the participating children, 70% had Italian parents, 8% had one Italian parent, and 22% had immigrant parents. According to the reports of preschool and primary school teachers, none of the 63 participants had been referred to the National Health Services for cognitive impairments or language difficulties in Italian. Average performance scores on a standardized measure of receptive language, the Italian version of the Peabody Picture Vocabulary Test (PPVT; Stella et al., 2000 ), collected at T1 ( M  = 88.08, DS  = 13.64) and nine months later during the first year of school ( M  = 95.86, DS  = 12.39), also showed that the average performance was within the age-appropriate range according to available norms ( M  = 100, SD  = 15). At the same two time points, a measure of general executive functioning was collected using the Inhibition subtest from the Italian version of the standardized assessment NEPSY-II (Urgesi et al., 2011 ). Performance in the Inhibition subtest, both at T1 ( M  = 8.63, DS  = 2.49) and nine months later ( M  = 10.47, DS  = 2.49), was within the age-appropriate range according to available norms ( M  = 10, SD  = 3).

At T1, parents completed questionnaires on demographic information and data on the home environment. The average educational level of mothers and fathers was high school ( M  = 2.28, SD  = 0.99, range 1–5 where 0 = primary school and 5 = master level/PhD). The annual income of the families was near the average Italian family income ( M  = 1.68, SD  = 1.01, range 0–4 where 0 = far below the mean and 4 = far above the mean; ISTAT, 2009 ). Taken together, these results suggest that all children shared a homogeneous middle/working-class social background. In addition, concerning the availability of digital devices and traditional and digital books, parents reported that 64–65% of the children used smartphones and tablets at home, and 40% of children used laptops or computers. Families had on average 21–40 traditional books for children (M = 2.44, SD = 1.32, range 1–5 where 0 = none and 5 = more than 80 books). Most of the families have few or no digital books for children (M = 0.18, SD = 0.38, range 0–1 where 0 = none and 5 = more than 80 books).

According to their teachers’ reports, the children attended 11 classes from 6 schools in the same school district, one where technology was used for educational activities 2/4 hours per month. In Italy, the national curriculum for teaching digital skills in public primary schools requires children to spend 1 h per week using computers.

Four texts about animals were used, two narrative and two descriptive (expository). The four texts were used in a previous study (Florit et al., 2021 ) and were partly modified to be used in a wider project on text comprehension on paper, computer, and tablet. Narrative and descriptive texts were devised as representative of the text genres used in first grade, and therefore appropriate for identifying possible differences across mediums (Delgado et al., 2018 ). The two narrative texts introduced a short story about an animal (a teddy bear and a hedgehog). The two descriptive texts provided information about an animal’s physical characteristics and behavior (tortoise and parrot). Texts were 101–106 words long, and their readability (Gulpease Index, Lucisano 1992 ) ranged from 81 to 85, showing that the texts were appropriate for primary school children (Piemontese, 1996 ). Examples of the texts are reported in the online supplementary information.

Measures at T1 (end of preschool)

Working memory . This was measured using a backward word span task (Florit et al., 2014 ). The task consisted of six series of word lists that had no semantic or phonological relations, and with an increasing number of items (list length ranged from 2 to 7 words; four series for each list length). Children were presented with the lists of words and asked to repeat them in reverse order. The test stopped when the child failed to repeat three of four series of the same length. One point was credited for each series correctly repeated (possible range 0–24). The reliability for the task was 0.68 (McDonald’s omega).

Inference skills . An inferential task, which did not involve complex discourse comprehension, was used (Florit et al., 2014 ; Dicataldo et al., 2020 ). The task consisted of ten episodes, each comprising three short sentences describing common and familiar events, and two inferential questions (one text-connecting inference and one integrative inference). Answers were evaluated on a 0–2 point scale for a maximum score of 40. The reliability for the inferential task was 0.92 (McDonald’s omega).

Basic digital skills . Following a checklist inspired by the work of Javorsky ( 2014 ), the children were asked 29 questions, mainly performance, to measure basic digital skills. The skills were related to the use of computers and laptops, the digital devices usually available in Italian primary schools. The checklist was piloted on a small group of 5 preschoolers who were not involved in the present investigation in order to confirm that the requests were appropriate to the technical experience of very young children and to refine the scoring system. Pilot work led to minor modifications. In the final version, the questions asked included: (a) mechanical skills concerning the use of the various components of the device (e.g., “Can you show me how you make the computer work?”), (b) navigation skills regarding access and navigation within and across pages of digital environments (e.g., “How can we go back to the first page”?), and (c) symbolic skills related to the knowledge of the main symbols of digital environments (e.g., “Show me the icon to play with letters”). In order to assess basic digital skills, a laptop with a touchpad and a mouse was used, and a Google-like environment was developed as a situated assessment for the present study. Children were first asked performance questions concerning their mechanical skills. Later, the navigation and symbolic skills were assessed by asking children to click on Internet Icons available on the screen to find web pages with stories and pictures. Specifically, children were required to click on one of four common Internet icons (i.e., Internet Explorer, Chrome, Firefox, and Safari) to follow an established path in accessing six web pages. These web pages constituted the Google-like environment. The icons and webpages were in fact offline and stored locally. Navigation skills were assessed by asking children to move within and across the web pages, and symbolic skills were evaluated by considering symbols presented on the web pages. At the end of the task, the child received a coloring page (e.g., a drawing of an animal/object that he/she had previously seen on the web pages) as a gift. Answers were coded 0–2 according to accuracy and completeness. A total score was computed. The reliability for this task was 0.72 (McDonald’s omega). For the analysis, 21 of the 29 items were retained since eight items of the symbolic skills subset tap-on skills were also involved in the linear printed text comprehension (e.g., "Show me where to start to read the text").

Measures at T2 (end of the first grade)

Text comprehension. Based on studies involving both older (Singer & Alexander, 2017 ) and same-age students (Florit et al., 2021 ), text comprehension was measured using seven multiple-choice questions with four alternative answers for each text. These were: one question on the main idea, three questions on literal comprehension, and three questions on inferential comprehension. Examples of questions are reported in the online supplementary information. The questions were devised by the first author and had been used in a previous study (Florit et al.,  2021 ). We relied only on verbal questions instead of verbal and pictorial questions for two main reasons: (a) to maximize the likelihood of comparable presentations of the comprehension questions across mediums and (b) because, according to the language arts teachers, verbal questions were also commonly used in first grade for the assessment of text comprehension. Each correct answer was awarded 1 point, and a total score was computed (the maximum score for each text was 7). The reliability of texts ranged from 0.68 to 0.71 (McDonald’s omega).

Medium preference questions (screen text vs. printed text) . Before and after reading all four texts, participants completed a preference question asking whether they preferred reading a text on paper or reading it on screen.

Word reading skills (control variable) . The word reading and non-word reading tasks used came from the Test Battery for the Evaluation of Developmental Dyslexia and Dysorthography (Sartori et al., 2007 ). The children were asked to read 112 words and 48 non-words as accurately and quickly as possible. Word reading and non-word reading accuracy (number of correct answers/number of items) and fluency (number of syllables read per second) were computed. Z-scores were derived for accuracy and fluency, and a composite score of word reading was calculated and used in the analyses. The reliability of the accuracy for words and non-words in the present sample was 0.93 and 0.85, respectively (McDonald’s omega). Test-retest reliability reported in the manual for reading fluency is 0.80.

The measures were administered in one session held at the end of the last year of preschool (during May and June) at T1 and five sessions, approximately one week apart, one year later at T2, as part of the broader project. Each session was 40–45 min long. T1 measures were administered individually in a quiet room of the school and in a counterbalanced order. In the first session at T2, in a quiet room of the primary school, students were individually asked the preference question at pre-test and then given the word and non-word reading tasks. At T2, in sessions two to five, the reading comprehension tasks were group-administered in the classroom and the computer lab. Each student read four texts, two on paper (one narrative and one descriptive) and two on a computer screen (one narrative and one descriptive), and responded to four sets of comprehension questions, two on paper and two on a computer screen. The order of text presentation was randomized by medium and genre. Each text was presented on a single page, and comprehension questions were asked and answered in the same medium as the one in which the text was presented; that is, the children had to tick or click the correct answer when reading on paper or computer screen, respectively. The screen resolution chosen required the children to use arrows to go down on the page to read the last sentence of each text. They also had to click on the arrow to go to the next page and to read the last comprehension question. All the children used the mouse autonomously to perform the digital text comprehension tasks. The children could access texts while answering questions in both mediums. They read the four texts at their own pace since the task had no time limit. The font size of all texts and questions was 16-point Cambria, the same as in textbooks with uppercase block letters that teachers used. Double spaced A4 sheets were used to present the printed texts. Digital texts appeared on a 17″ computer screen by means of the open-source software LimeSurvey; the resolution was 1280 × 1024 pixels. At the end of the fifth session, the children answered the medium preference question at post-test.

Data analysis

First, descriptive statistics were computed in terms of means, standard deviations, skewness and kurtosis, and correlations were computed for all continuous variables in the study. For reading medium preference, mean percentages were computed at pre-and post-test. Second, to address RQ1, linear mixed models were used to test the effects of medium (paper vs. digital), text genre (narrative vs. descriptive), and their interaction on text comprehension. If a significant interaction term was found, a simple slope analysis was performed, and data were plotted. A binomial test was also computed to test differences in reading medium preference at pre- and post-test. Third, to address RQ2, linear mixed models were run to test for (a) the contribution of working memory, inference skills, and basic digital skills at T1 to text comprehension at T2, and (b) the interaction effect between basic digital skills and medium to text comprehension at T2. These linear mixed models controlled for word reading skills at T2, medium, and text genre. Mixed models were used to take into account the nested structure of the data, since observations were nested in individuals (i.e., repeated measures of text comprehension across mediums and text genre) who were nested in classes. Estimation problems prevented the fit of planned models with both random intercepts and slopes. As suggested by Barr et al. ( 2013 ), non-converging models were dealt with by progressively simplifying the random effects structure until convergence was reached, resulting in our case in a random-intercept-only model.

Analyses were performed with R (R Development Core Team, 2021 ) using the lme4 package (Bates et al., 2015 ) to test generalized and linear mixed models and the lmerTest (Kuznetsova et al., 2014 ) to obtain standard errors for linear mixed models. The percentage of variance explained by each model was computed following the approach of Nakagawa and colleagues ( 2017 ). It included both marginal R 2 and conditional R 2 , which refer, respectively, to the variance explained by the fixed effects and the global variance of random and fixed effects. Plots were built with the ggplot2 package (Wickham, 2009 ).

Descriptive statistics of the continuous variables and correlations among them are presented in Tables  1 and 2 , respectively.

The data did not deviate substantially from normality for skewness and kurtosis, which were within the acceptable range (Tabachnick & Fidell, 2013 ). Performances in the working memory task were in line with previous studies (Florit et al., 2014 ). Performances on the tests of inference skills, basic digital skills, and text comprehension covered almost the whole range of possible scores. They did not suffer from either floor or ceiling effects. Table  1 reports the composite scores (Z scores of accuracy and fluency) of word reading skills. The average performances for measures of reading accuracy ( M  = 0.83, DS  = 0.08; M  = 0.91, DS =  0.14 for words and non-words) and fluency ( M  = 0.65, DS  = 0.24; M  = 0.70, DS =  0.31 for words and non-words) were within the age-appropriate range according to available norms (Sartori et al., 2007 ). Zero-order correlations, which range from low to large, were reported for descriptive purposes only since Pearson’s correlations do not consider the nested nature of the data.

At pre-test, 48% of children reported that they preferred reading on paper and 52% on screen. At post-test, 22% of children said they preferred reading on paper and 78% on screen.

RQ1: comprehension of narrative and descriptive texts from paper and screen and reading medium preferences

The first linear mixed model found the non-significant main effects of both medium (b = 0.01, 95% CI -0.11–0.14, SE = 0.06, t (245) = 0.19, p  = .85) and text genre (b = -0.10, 95% CI -0.23 – 0.03 SE = 0.06, t (245) = -1.54, p  = .12), but their significant interaction on text comprehension (b = 0.23, 95% CI 0.10–0.35, SE = 0.06, t (245) = 3.52, p  < .001). Figure  1 presents the effects of the interaction effect, showing that narrative texts are associated with better text comprehension than descriptive texts in the case of paper (b = -0.65, 95% CI -1.04 – -0.26, SE = 0.20, t (121) = 3.30, p  = .001) but not in the case of the digital medium (b = 0.25, 95% CI -0.10–0.60, SE = 0.18, t (121) = 1.43, p  = .15).

Interaction of medium and text genre on text comprehension

The Binomial tests showed that the children had a clear preference for reading on screen at post-test ( p  = .001) but no clear medium preference at pre-test ( p  = .801).

RQ2: predictors of narrative and descriptive text comprehension on paper and screen

Table  3 presents the results of the second linear mixed model.

Working memory and inference skills at T1 were significant unique and positive predictors of text comprehension at T2, after controlling for medium, text genre, and word reading skills at T2. The latter, considered as a control variable, was also a unique significant predictor of text comprehension at T2. The interaction between basic digital skills and medium, as hypothesized, was also significant: digital skills contributed to the comprehension of texts presented on screen but not on paper. Figure  2 shows the simple slope analysis to interpret the moderation effect.

Simple slope analysis of the interaction between medium and basic digital skills on text comprehension

Basic digital skills showed a positive, marginally significant unique longitudinal effect on digital text comprehension (b = 0.04, SE = 0.02, t  = 1.77, p  = .08), but not on printed text comprehension (b = 0.00, SE = 0.02, t  = 0.01, p  = 1.00).

This study sought to extend current knowledge about text comprehension in relation to reading medium and text genre. To our knowledge, it is the first study with a longitudinal design involving children in the crucial transition from preschool to primary school. We focused on the comprehension of comparable linear texts in first graders who have just learned to read independently. We followed these children from the end of preschool as we were interested in fundamental and high-level cognitive precursors of text comprehension, which have long been the subject of research on the comprehension of linear printed texts (e.g., Ariasi & Mason 2014 ; Kim, 2017 ; Oakhill & Cain, 2012 ). We also considered the precursor of basic digital skills, as it has been shown that it is required to effectively read from the screen of an electronic device (Fajardo et al., 2016 ).

Our first research question asked whether reading medium (paper vs. screen) and text genre (narrative vs. descriptive) differentiate text comprehension in beginner readers at the end of the first grade. Our findings did not confirm our hypotheses, as the two main effects of reading medium and text genre were not significant. In contrast, a hypothesized interaction between reading medium and text genre did emerge; the narrative text was comprehended better than the descriptive text when reading on paper. The lack of the main effect of the reading medium does not accord with the outcomes of the most recent meta-analyses (Clinton, 2019 ; Delgado et al., 2018 ; Kong et al., 2018 ), which supported a screen inferiority effect in independent readers. In addition, our results for beginner readers cannot be explained by reference to the shallowing hypothesis (Annisette & Lafreniere, 2017 ). This hypothesis, which was adopted to explain the results of previous meta-analytic research, proposes that the medium acts as a contextual cue for text processing due to the different uses of paper and digital mediums.

Previous comparative studies on reading medium have involved students at the secondary level (Clinton, 2019 ; Delgado et al., 2018 ) and, even when primary school students were involved in recent studies (Dahan Golan et al., 2018 ; Halamish & Elbaz, 2020 ; Støle et al., 2020 ), they were in the higher grades. Therefore, recent work relied on older independent readers with much accumulated experience of reading on paper mainly for study and learning and of using the digital medium for leisure. According to our data, participants in this study used technology at home (i.e., they mainly use smartphones and tablets) probably for leisure and entertainment (e.g., for listening to music, watching videos, and playing with videogames; Rideout & Robb, 2020 ). They were also exposed to pleasant activities involving the paper medium, such as shared book reading. However, they had much less accumulated experience both of reading on paper mainly for study and learning and of using the digital medium for leisure. These differences in the use of the two mediums between older and younger students may limit the medium effect. That is differences in exposure both prevent the paper medium acting as a contextual cue to processing a text for study and learning, and the digital medium acting as a contextual cue to processing a text for leisure.

A second hypothesis to explain the disadvantage of text comprehension on screen considers the higher cognitive load imposed by the students’ lower level of familiarity with reading on screen than on paper and with using digital devices (Ackerman & Lauterman, 2012 ; Furenes et al., 2021 ). We found few differences in the children’s reading strategies between reading on paper and screen. As discussed above, this can be explained by their limited experience as independent readers compared with older students who have learned how to read from paper. A few of the children returned to the text while responding to the comprehension questions when reading texts on paper but not on screen. This strategy seems to be related to the children’s literacy experience at school rather than to basic digital skills (i.e., navigation skills). The teachers habitually advised children to go back to the text to find information when answering comprehension questions, rather than relying on memory. When doing so, the teachers were mainly using printed texts. Therefore, some children adopted this strategy, at least to some extent, while reading on paper but not on screen. The use of this strategy is linked to the use of paper mainly for study and learning.

Our results do not align with existing evidence of a disadvantage for the comprehension of enhanced digital books vs. printed books when the former included few enhancements and required additional resources to point, click, and swipe (Furenes et al., 2021 ). It is difficult to compare the results of the present study with those of Furenes and colleagues ( 2021 ), who focused on the comprehension of books in non-independent readers. Nevertheless, we may speculate that the reason a disadvantage for digital texts did not emerge in our study is that the children faced less cognitively demanding digital texts than they would meet with enhanced digital books. Indeed, results from the longitudinal analysis undertaken in the present study support the critical role of digital skills.

The results on the interactive effect of reading medium and text genre showed that text genre influences text comprehension only when the traditional reading medium is considered. According to parents’ reports, children have limited experience with technology at school and digital books at home, but much accumulated experience of storybook reading at home. We can also speculate that this outcome is mainly due to the children’s greater familiarity with independent reading of narrative text from paper at school, which is what they are mainly asked to do in the first grade as soon as they have learned to read (Diakidoy et al., 2005 ). Furenes and colleagues ( 2021 ) found no differences when considering book genre (fictional or non-fictional) as a moderator of medium effect. This different result may be due to the fact that the most common book genre in the study of Furenes and colleagues ( 2021 ) was fiction for a younger age group.

Concerning reading preferences, our participants showed a preference for reading on screen at post-test but not at pre-test. This result can be explained by the fact that children predominantly use the traditional medium for reading at school while families have no or few digital books for children at home. We may therefore speculate that the participants perceived reading on screen as more attractive than reading on paper because for them digital texts represented a relatively novel experience in the school context.

Our second research question addressed the role of traditional literacy skills (working memory and inference skills) and digital skills (basic digital skills) in text comprehension in the two mediums. Our hypotheses about longitudinal predictors were partially confirmed. Specifically, digital skills assessed at the end of preschool were a significant predictor of digital text comprehension one year later, over and above cognitive predictors. The higher were the children’s basic digital skills to use a laptop and navigate within a digital context in preschool, the higher was their comprehension of digital texts at the end of first grade. This finding is aligned with those showing the importance of these skills as a resource when reading online (Fajardo et al., 2016 ; Neumann et al., 2016 ). They are also particularly important in the youngest readers tackling linear texts.

In addition, our longitudinal models show that inference skills assessed at the end of preschool accounted for digital and printed text comprehension one year later, after controlling for word reading skills. The latter, moreover, was confirmed as a significant predictor of beginner readers’ text comprehension in both mediums (Florit et al., 2021 ) and both text genres. Contrary to our expectations, working memory was also a unique longitudinal predictor of printed and digital text comprehension, even if higher-order inference skills were considered.

Our results extend to beginner readers the research that has explored the roles played by foundational cognitive and language component skills in the comprehension of printed and digital texts in older independent readers (Delgado & Salmerón, 2021 ) and the comprehension of printed and digital books in younger preschoolers (Lauricella et al., 2014 ). We provide empirical evidence for a model of linear digital text comprehension in which basic digital skills are a specific component, along with word reading, working memory, and inference skills, which are components common to both printed and digital text comprehension. We considered basic digital skills related to computer and laptop use because they are the most commonly used digital devices in primary schools, at least in Italy. As shown by parents’ responses to the questionnaire about the home environment, children at the end of preschool are more familiar at home with other digital devices, such as tablets and/or smartphones (cfr. Rideout & Robb, 2020 ; Smahel et al., 2020 ). In younger readers or pre-readers, digital skills depend more on the medium used. However, all items in the checklist on digital skills, except the subset assessing mechanical skills use, enable children to understand and use both computer/laptop and tablet as mediums through which a text may be presented. This claim is supported by the fact that the checklist on computer/laptop use correlated significantly ( r  = .58) with a second checklist with similar items but related to the use of a tablet, which was administered as part of the wider project on text comprehension in beginner readers. The findings of the present study, though, do not support a test of whether working memory and inference skills can predict potential differences in text comprehension across mediums in beginner readers.

Educational implications

This study has significant educational implications. First, our results underline the importance of acquiring basic digital skills to support linear digital text comprehension. Child-adult interactions with digital devices and digital texts may sustain the development of emergent digital literacy (Hisrich & Blanchard, 2009 ). Children start interacting with digital texts using electronic storybooks whose technological enhancements make the reading experience qualitatively different from their experience with traditional paper books. For instance, Lauricella and colleagues ( 2014 ) found that parents focused more on the mechanics of the device when reading digital than printed books when interacting with their four-year-old children. Thus, child-adult interactions with digital texts should be regarded as additional contextual factors that may promote basic digital skills acquisition. As shown by the descriptive data of the present study, preschoolers start on-screen activities using laptops and computers that are present at home. These activities require some adult guidance and may also support basic digital skills acquisition. Notably, the digital skills considered by our checklist, especially navigational skills, are also needed for interacting with digital devices other than laptops or computers, such as smartphones and tablets, to which children are also exposed. As research has indicated, digital natives do not necessarily have the digital skills to benefit fully from digital reading (Fajardo et al., 2016 ; Neumann et al., 2016 ). It is, therefore, crucial to provide children from an early age with a range of ways of making sense of reading in digital environments. These opportunities should also help beginner readers to develop awareness of what digital reading requires, as distinct from traditional paper reading.

Second, our results show that the involvement of working memory and inference skills in the construction of meaning in texts is independent of the medium through which the text is presented. It has been shown that adults’ mediation during print book reading influences children’s text comprehension more effectively than the enhancements provided by digital books read by children independently, particularly when the enhancements are not related to the content of the book (Furenes et al., 2021 ). However, the superiority of paper over digital books in stimulating text comprehension disappears when adult mediation is the same between the mediums and when the enhancements in digital books are content-related (Furenes et al., 2021 ). There is also evidence that parents can adjust their interactions based on the characteristics of the medium and their children’s skills to facilitate meaning construction (Lauricella et al., 2014 ). In other words, the quality of parent-child social interactions is crucial to the promotion of children’s text comprehension regardless of the medium through which the book is presented. In sum, previous evidence suggests that, when the digital books selected are unlikely to cause children’s distraction or increase cognitive load, child-adult literacy activities in the digital medium may help to promote the acquisition of working memory and inference skills, as is the case with literacy activities in the traditional medium.

Limitations

First, the sample is small, although the attrition from T1 and T2 is acceptable. More participants would allow for more complex statistical analyses and more solid data. Second, only one year elapsed from T1 and T2, so the longitudinal nature of the study is somewhat limited, although we did include a crucial transitional phase examining the role of the precursors of text comprehension in readers who have just learned to read. Future studies should also cover a broad age range to further test our interpretations concerning the differences in the use of the two mediums between students at the lower and higher end of primary school. This will make it possible to further clarify how differences in the use of mediums relate to the medium effect on text comprehension. Third, while we considered some of the critical precursors, other components are also involved in children’s text comprehension. Future research that includes additional components will illuminate any differences in reading comprehension related to reading medium and text genre. In this regard, whereas we included traditional measures of word reading skills as a control variable, we did not collect measures of word reading in both mediums. Studies of readers who have mastered word reading skills, mainly undergraduates, as in Clinton ( 2019 ; for a study on adolescents, see Ronconi et al., 2022 ), found no differences in reading times between the two mediums. Longer times, though, were found for reading on screen than on paper in children at the end of primary school (Kerr & Symons, 2006 ; but see Lenhard et al., 2017 ). In these studies, text comprehension performance was better on paper than on screen. These results, therefore, indicate that reading from paper is more efficient than reading from a screen. Traditional measures of beginner readers’ word reading skills have previously been shown to play a stronger role in text comprehension on screen than on paper at the literal and inferential level of text comprehension (Florit et al., 2021 ). Based on this evidence, future research should assess word reading accuracy and fluency in both mediums, in order to deepen our understanding of the potential differences in reading between the mediums. Finally, we observed few differences in reading strategies (i.e., metacognitive strategies such as going back to the text while answering comprehension questions) between reading on paper and screen. However, meta-analytic studies on university students have suggested that reading medium effects may be related to differences at the metacognitive level (e.g., Clinton 2019 ). Therefore, more systematic observations on a larger sample are required if we are to form firm conclusions on the existence of such differences in beginner readers.

Fourth, we considered only one digital device, the computer, as children had some familiarity with computers and laptops in primary school. Future research may also compare types of digital device used for reading on screen, as differences could emerge in relation to their affordances (Salmerón et al., 2021 ). Fifth, because we considered comparable linear printed and digital texts appropriate for beginner readers, our results cannot be generalized to the types of non-linear, complex texts that characterize multimedia environments.

Sixth, because we used only a few texts, and they were relatively brief (as was deemed appropriate to the children’s grade level), we used only a limited number of questions to assess text comprehension.

Conclusions

Despite these limitations, the present longitudinal study extends current research by examining the role of traditional and digital literacy skills in the comprehension of printed and digital texts, both narrative and expository texts, in the crucial transition from preschool to the first grade. The study shows that in beginner readers, at the end of the first grade, there was no evident inferiority for reading from a screen, at least when linear printed and digital texts were considered. Word reading, working memory, and inference skills were common precursors of text comprehension in the two mediums, while basic digital skills were specific precursors of digital text comprehension. Digital skills should be considered in comprehensive models of emergent digital literacy.

Ackerman, R., & Lauterman, T. (2012). Taking reading comprehension exams on screen or on paper? A metacognitive analysis of learning texts under time pressure. Computers in Human Behavior , 28 (5), 1816–1828. https://doi.org/10.1016/j.chb.2012.04.023

Article   Google Scholar  

Annisette, L. E., & Lafreniere, K. D. (2017). Social media, texting, and personality: A test of the shallowing hypothesis. Personality and Individual Differences , 115 , 154–158. https://doi.org/10.1016/j.paid.2016.02.043

Ariasi, N., & Mason, L. (2014). From covert processes to overt outcomes of refutation text reading: The interplay of science text structure and working memory through eye fixations. International Journal of Science and Mathematics Education , 12 (3), 493–523. https://doi.org/10.1007/s10763-013-9494-9

Barr, D. J., Levy, R., Scheepers, C., & Tily, H. J. (2013). Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language , 68 (3), 255–278. https://doi.org/10.1016/j.jml.2012.11.001

Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software , 67 (1), 1–48. https://doi.org/10.18637/jss.v067.i01

Bos, L. T., De Koning, B. B., Wassenburg, S. I., & van der Schoot, M. (2016). Training inference making skills using a situation model approach improves reading comprehension. Frontiers in Psychology , 7 , 32016. https://doi.org/10.3389/fpsyg.2016.00116

Cain, K., Oakhill, J. V., & Bryant, P. E. (2004). Children’s reading comprehension ability: Concurrent prediction by working memory, verbal ability, and component skills. Journal of Educational Psychology , 96 , 31–42. https://doi.org/10.1037/0022-0663.96.1.31

Carretti, B., Caldarola, N., Tencati, C., & Cornoldi, C. (2014). Improving reading comprehension in reading and listening settings: The effect of two training programmes focusing on metacognition and working memory. British Journal of Educational Psychology , 84 , 194–210. https://doi.org/10.1111/bjep.12022

Clinton, V. (2019). Reading from paper compared to screens: A systematic review and meta-analysis. Journal of Research in Reading , 42 , 288–325. https://doi.org/10.1111/1467-9817.12269

Clinton, V., Taylor, T., Bajpayee, S., Davison, M. L., Carlson, S. E., & Seipel, B. (2020). Inferential comprehension differences between narrative and expository texts: A systematic review and meta-analysis. Reading & Writing , 33 , 2223–2248. https://doi.org/10.1007/s11145-020-10044-2

Currie, N. K., & Cain, K. (2015). Children’s inference generation: The role of vocabulary and working memory. Journal of Experimental Child Psychology , 137 , 57–75. https://doi.org/10.1016/j.jecp.2015.03.005

Dahan Golan, D., Barzillai, M., & Katzir, T. (2018). The effect of presentation mode on children’s reading preferences, performance, and self-evaluations. Computers & Education , 126 , 346–358. https://doi.org/10.1016/j.compedu.2018.08.001

Delgado, P., Vargas, C., Ackerman, R., & Salmerón, L. (2018). Don’t throw away your printed books: A meta-analysis on the effects of reading media on reading comprehension. Educational Research Review , 25 , 23–38. https://doi.org/10.1016/j.cedpsych.2020.101870

Delgado, P., & Salmerón, L. (2021). The inattentive on-screen reading: Reading medium affects attention and reading comprehension under time pressure. Learning and Instruction , 71 , 101396. https://doi.org/10.1016/j.learninstruc.2020.101396

Diakidoy, I. A. N., Stylianou, P., Karefillidou, C., & Papageorgiou, P. (2005). The relationship between listening and reading comprehension of different types of text at increasing grade levels. Reading Psychology , 26 , 55–80. https://doi.org/10.1080/02702710590910584

Dicataldo, R., Florit, E., & Roch, M. (2020). Fostering broad oral language skills in preschoolers from low SES background. International Journal of Environmental Research and Public Health , 17 , 4495. https://doi.org/10.3390/ijerph17124495

Fajardo, I., Villalta, E., & Salmerón, L. (2016). ¿Son realmente tan buenos los nativos digitales? Relación entre las habilidades digitales y la lectura digital? [Are really digital natives so good? Relationship between digital skills and digital reading]. Anales de Psicologia , 32 , 89–97. https://doi.org/10.6018/analesps.32.1.185571

Florit, E., De Carli, P., Lavelli, M., & Mason, L. (2021). Digital reading in beginner readers: Advantage or disadvantage for text comprehension? Department of Human Sciences, University of Verona.

Florit, E., Roch, M., & Levorato, M. C. (2014). Listening text comprehension in preschoolers: A longitudinal study on the role of semantic components. Reading and Writing. An Interdisciplinary Journal , 27, 793–817. https://doi.org/10.1007/s11145-013-9464-1

Furenes, M. I., Kucirkova, N., & Bus, A. G. (2021). A Comparison of children’s reading on paper versus screen: A meta-analysis. Review of Educational Research , 91 , 483–517. https://doi.org/10.3102/0034654321998074

Halamish, V., & Elbaz, E. (2020). Children’s reading comprehension and metacomprehension on screen versus on paper. Computers & Education , 145 , 1–11. https://doi.org/10.1016/j.compedu.2019.103737 . )

Hisrich, K., & Blanchard, J. (2009). Digital media and emergent literacy. Computers in the Schools , 26 , 240–255. https://doi.org/10.1080/07380560903360160

ISTAT (2009). Istituto nazionale di statistica [National Institute of Statistics]. https://www.istat.it/it/files/2016/05/Volume2008-edizione2009.pdf

Javorsky, K. H. (2014). Digital print concepts: Conceptualizing a modern framework for measuring emerging knowledge.  Theses, Student Research, and Creative Activity, Department of Teaching, Learning & Teacher Education, University of Nebraska, Lincoln. http://digitalcommons.unl.edu/teachlearnstudent/39

Google Scholar  

Kerr, M. A., & Symons, S. E. (2006). Computerized presentation of text: Effects on children’s reading of informational material. Reading and Writing , 19 , 1–19. https://doi.org/10.1007/s11145-003-8128-y

Kim, Y. S. G. (2017). Why the simple view of reading is not simplistic: Unpacking the simple view of reading using a direct and indirect effect model of reading (DIER). Scientific Studies of Reading , 21 , 310–333. https://doi.org/10.1080/10888438.2017.1291643

Kintsch, W. (1998). Comprehension: A paradigm for cognition . Cambridge University Press

Kong, Y., Seo, Y. S., & Zhai, L. (2018). Comparison of reading performance on screen and on paper: A meta-analysis. Computers & Education , 123 , 138–149. https://doi.org/10.1016/j.compedu.2018.05.005

Kuznetsova, A., Brockhoff, P. B., Haubo, R., & Christensen, B. (2014). lmerTest: Tests in Linear Mixed Effects Models. R package version 2.0–20. http://cran.r-project.org/package=lmerTest

Language and Reading Research Consortium, & Chiu, Y. D. (2018). The simple view of reading across development: Prediction of grade 3 reading comprehension from prekindergarten skills. Remedial and Special Education , 39 (5), 289–303. https://doi.org/10.1177/0741932518762055

Language and Reading Research Consortium, & Logan, J. (2017). Pressure points in reading comprehension: A quantile multiple regression analysis. Journal of Educational Psychology , 109 , 451–464. https://doi.org/10.1037/edu0000150

Lauricella, A. R., Barr, R., & Calvert, S. L. (2014). Parent-child interactions during traditional and computer storybook reading for children’s comprehension: Implications for electronic storybook design. International Journal of Child-Computer Interaction , 2 , 17–25. https://doi.org/10.1016/j.ijcci.2014.07.001

Lenhard, W., Schroeders, U., & Lenhard, A. (2017). Equivalence of screen versus print reading comprehension depends on task complexity and proficiency. Discourse Processes , 54 (5–6), 427–445. https://doi.org/10.1080/0163853X.2017.1319653

Lepola, J., Lynch, J., Kiuru, N., Laakkonen, E., & Niemi, P. (2016). Early oral language comprehension, task orientation, and foundational reading skills as predictors of grade 3 reading comprehension. Reading Research Quarterly , 51 (4), 373–390. https://doi.org/10.1002/rrq.145

Lucisano, P. (1992). Misurare le parole [Measuring words]. Kepos

Mayer, R. E. (2009). Multimedia learning (2nd ed.).). Cambridge University Press

Nakagawa, S., Johnson, P. C. D., & Schielzeth, H. (2017). The coefficient of determination R 2 and intra-class correlation coefficient from generalized linear mixed-effects models revisited and expanded. Journal of The Royal Society Interface , 14 (134), https://doi.org/10.1098/rsif.2017.0213

Neumann, M. M., Finger, G., & Neumann, D. L. (2016). A conceptual framework for emergent digital literacy. Early Childhood Education Journal , 45 , 1–9. https://doi.org/10.1007/s10643-016-0792-z

Oakhill, J. V., & Cain, K. (2012). The precursors of reading ability in young readers: Evidence from a four-year longitudinal study. Scientific Studies of Reading , 16 , 91–121. https://doi.org/10.1080/10888438.2010.529219

Piemontese, M. E. (1996). Capire e farsi capire. Teorie e tecniche della scrittura controllata [To understand and make oneself to be understood. Theories and techniques of controlled writing]. Tecnodid

Prensky, M. (2001). Digital natives, digital immigrants. On the Horizon , 9 , 1–6. https://doi.org/10.1016/j.appdev.2017.01.002

R Development Core Team (2021). The R Project for Statistical Computing (R. D. C. Team, Ed.). http://www.r-project.org/

Reading Study Group. (2002). Reading for understanding: Toward a research and development program in reading comprehension . Rand Education

Rideout, V., & Robb, M. B. (2020). The Common Sense census: Meia use by kids age zero to eight , 2020 . Common Sense Media. https://www.commonsensemedia.org/research/the-common-sense-census-media-use-by-kids-age-zero-to-eight-2020

Ronconi A., Veronesi V., Mason L., Manzione L., Florit E., Anmarkrud O., & Bråten I. (2022). Effects of reading medium on the processing, comprehension, and calibration of adolescent readers. Computers & Education , 185 . https://doi.org/10.1016/j.compedu.2022.104520

Salmerón, L., Delgado, P., Vargas, C., & Gil, L. (2021). Tablets for all? Testing the screen inferiority effect with upper primary school students. Learning and Individual Differences , 86 , 101975. https://doi.org/10.1016/j.lindif.2021.101975

Sanchez, C. A., & Wiley, J. (2009). To scroll or not to scroll: Scrolling, working memory capacity, and comprehending complex texts. Human Factors , 51 (5), 730–738. https://doi.org/10.1177/%2F0018720809352788

Sartori, G., Job, R., & Tressoldi, P. E. (2007). Batteria per la valutazione della dislessia e della disortografia evolutiva – 2 [Test battery for the evaluation of developmental dyslexia and dysorthography – 2]. Giunti Organizzazioni Speciali

Seigneuric, A., & Ehrlich, M. F. (2005). Contribution of working memory capacity to children’s reading comprehension: A longitudinal investigation. Reading and Writing , 18 (7–9), 617–656. https://doi.org/10.1007/s11145-005-2038-0

Silva, M., & Cain, K. (2015). The relations between lower and higher level comprehension skills and their role in prediction of early reading comprehension. Journal of Educational Psychology , 107 (2), 321–331. https://doi.org/10.1037/a0037769

Singer, L. M., & Alexander, P. A. (2017). Reading across mediums: Effects of reading digital and print texts on comprehension and calibration. Journal of Experimental Education , 85 , 155–172. https://doi.org/10.1080/00220973.2016.1143794

Smahel, D., Machackova, H., Mascheroni, G., Dedkova, L., Staksrud, E., Ólafsson, K., Livingstone, S., & Hasebrink, U. (2020). EU Kids Online 2020: Survey results from 19 countries . EU Kids Online . http://eprints.lse.ac.uk/id/eprint/103294

Stella, G., Pizzoli, C., & Tressoldi, P. E. (2000). Peabody test di vocabolario recettivo . Omega Edizioni

Støle, H., Mangen, A., & Schwippert, K. (2020). Assessing children’s reading comprehension on paper and screen: A mode-effect study. Computers & Education , 151 , Article e103861. https://doi.org/10.1016/j.compedu.2020.103861 .

Tabachnick, B. G., & Fidell, L. S. (2013). Using multivariate statistics (6th ed.). Pearson

Takacs, Z., Swart, E., & Bus, A. (2015). Benefits and pitfalls of multimedia and interactive features in technology-enhanced storybooks. Review of Educational Research , 85 (4), 698–739. https://doi.org/10.3102/0034654314566989

Tobia, V., & Bonifacci, P. (2015). The simple view of reading in a transparent orthography: The stronger role of oral comprehension. Reading and Writing , 28 (7), 939–957. https://doi.org/10.1007/s11145-015-9556-1

Urgesi, C., Campanella, F., & Fabbro, F. (2011). NEPSY-II- Second edition (Italian standardization) . Giunti Organizzazioni Speciali

Wickham, H. (2009). ggplot2. Elegant graphics for data analysis . Springer Science & Business Media

Download references

Acknowledgements

We thank the school principals, the children, the parents, and the teachers for taking part in the study.

This research was funded by University of Padova with a postdoc scholarship (Assegni Junior 2017) to Dr. Elena Florit (supervisor: Prof. Lucia Mason).

Open access funding provided by Università degli Studi di Verona within the CRUI-CARE Agreement.

Author information

Authors and affiliations.

Department of Human Sciences, University of Verona, Via S. Francesco 22, 37129, Verona, Italy

Elena Florit

Department of Psychology, University of Milano-Bicocca, Milano, Italy

Pietro De Carli

Department of Information Engineering, University of Padova, Padova, Italy

Antonio Rodà

Department of Developmental Psychology and Socialization, University of Padova, Padova, Italy

Samantha Domenicale & Lucia Mason

You can also search for this author in PubMed   Google Scholar

Contributions

Elena Florit: Conceptualization, Methodology, Analysis, Writing-Original Draft. Pietro De Carli: Analysis, Writing-Original Draft. Antonio Rodà: Methodology. Samantha Domenicale: Data Collection. Lucia Mason: Conceptualization, Writing-Original Draft, Supervision.

Corresponding author

Correspondence to Elena Florit .

Ethics declarations

The authors declare no conflict of interest.

The study was approved by the ethics committee of Psychological Research Area of the University of Padova (ethical approval code: 7F35AB627AB88DEF551318A3539E38EE).

Children participated with their own verbal assent and their parents’ written consent.

The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Elena Florit and Pietro De Carli were previously at the Department of Developmental Psychology and Socialization, University of Padova, where this work was done.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Florit, E., De Carli, P., Rodà, A. et al. Precursors of reading text comprehension from paper and screen in first graders: a longitudinal study. Read Writ 36 , 1821–1843 (2023). https://doi.org/10.1007/s11145-022-10327-w

Download citation

Accepted : 11 June 2022

Published : 08 October 2022

Issue Date : September 2023

DOI : https://doi.org/10.1007/s11145-022-10327-w

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Text comprehension
• Digital reading
• Reading on paper
• Longitudinal study
• Beginner readers
• Find a journal
• Publish with us
• Track your research

Academia.edu no longer supports Internet Explorer.

To browse Academia.edu and the wider internet faster and more securely, please take a few seconds to  upgrade your browser .

•  We're Hiring!
•  Help Center

Reading Comprehension

• Most Cited Papers
• Most Downloaded Papers
• Newest Papers
• Save to Library
• Last »
• Reading Follow Following
• Literacy Follow Following
• Psycholinguistics Follow Following
• Reading Strategies and Cognition Follow Following
• Reading Assessment Follow Following
• Teaching and Learning Writing and Reading Follow Following
• Vocabulary Follow Following
• Adolescent Literacy Follow Following
• New Literacies Follow Following
• Writing Follow Following

Enter the email address you signed up with and we'll email you a reset link.

• Academia.edu Publishing
•   We're Hiring!
•   Help Center
• Find new research papers in:
• Health Sciences
• Earth Sciences
• Cognitive Science
• Mathematics
• Computer Science
• Academia ©2024

SYSTEMATIC REVIEW article

Levels of reading comprehension in higher education: systematic review and meta-analysis.

• 1 Departamento de Métodos de Investigación y Diagnóstico en Educación, Universidad Internacional de la Rioja, Logroño, Spain
• 2 Department of Theory and History of Education and Research Methods and Diagnosis in Education, University of Malaga, Málaga, Spain

Higher education aims for university students to produce knowledge from the critical reflection of scientific texts. Therefore, it is necessary to develop a deep mental representation of written information. The objective of this research was to determine through a systematic review and meta-analysis the proportion of university students who have an optimal performance at each level of reading comprehension. Systematic review of empirical studies has been limited from 2010 to March 2021 using the Web of Science, Scopus, Medline, and PsycINFO databases. Two reviewers performed data extraction independently. A random-effects model of proportions was used for the meta-analysis and heterogeneity was assessed with I 2 . To analyze the influence of moderating variables, meta-regression was used and two ways were used to study publication bias. Seven articles were identified with a total sample of the seven of 1,044. The proportion of students at the literal level was 56% (95% CI = 39–72%, I 2 = 96.3%), inferential level 33% (95% CI = 19–46%, I 2 = 95.2%), critical level 22% (95% CI = 9–35%, I 2 = 99.04%), and organizational level 22% (95% CI = 6–37%, I 2 = 99.67%). Comparing reading comprehension levels, there is a significant higher proportion of university students who have an optimal level of literal compared to the rest of the reading comprehension levels. The results have to be interpreted with caution but are a guide for future research.

Introduction

Reading comprehension allows the integration of knowledge that facilitates training processes and successful coping with academic and personal situations. In higher education, this reading comprehension has to provide students with autonomy to self-direct their academic-professional learning and provide critical thinking in favor of community service ( UNESCO, 2009 ). However, research in recent years ( Bharuthram, 2012 ; Afflerbach et al., 2015 ) indicates that a part of university students are not prepared to successfully deal with academic texts or they have reading difficulties ( Smagorinsky, 2001 ; Cox et al., 2014 ), which may limit academic training focused on written texts. This work aims to review the level of reading comprehension provided by studies carried out in different countries, considering the heterogeneity of existing educational models.

The level of reading comprehension refers to the type of mental representation that is made of the written text. The reader builds a mental model in which he can integrate explicit and implicit data from the text, experiences, and previous knowledge ( Kucer, 2016 ; van den Broek et al., 2016 ). Within the framework of the construction-integration model ( Kintsch and van Dijk, 1978 ; Kintsch, 1998 ), the most accepted model of reading comprehension, processing levels are differentiated, specifically: A superficial level that identifies or memorizes data forming the basis of the text and a deep level in which the text situation model is elaborated integrating previous experiences and knowledge. At these levels of processing, the cognitive strategies used, are different according to the domain-learning model ( Alexander, 2004 ) from basic coding to a transformation of the text. In the scientific literature, there are investigations ( Yussof et al., 2013 ; Ulum, 2016 ) that also identify levels of reading comprehension ranging from a literal level of identification of ideas to an inferential and critical level that require the elaboration of inferences and the data transformation.

Studies focused on higher education ( Barletta et al., 2005 ; Yáñez Botello, 2013 ) show that university students are at a literal or basic level of understanding, they often have difficulties in making inferences and recognizing the macrostructure of the written text, so they would not develop a model of a situation of the text. These scientific results are in the same direction as the research on reading comprehension in the mother tongue in the university population. Bharuthram (2012) indicates that university students do not access or develop effective strategies for reading comprehension, such as the capacity for abstraction and synthesis-analysis. Later, Livingston et al. (2015) find that first-year education students present limited reading strategies and difficulties in understanding written texts. Ntereke and Ramoroka (2017) found that only 12.4% of students perform well in a reading comprehension task, 34.3% presenting a low level of execution in the task.

Factors related to the level of understanding of written information are the mode of presentation of the text (printed vs. digital), the type of metacognitive strategies used (planning, making inferences, inhibition, monitoring, etc.), the type of text and difficulties (novel vs. a science passage), the mode of writing (text vs. multimodal), the type of reading comprehension task, and the diversity of the student. For example, several studies ( Tuncer and Bahadir, 2014 ; Trakhman et al., 2019 ; Kazazoglu, 2020 ) indicate that reading is more efficient with better performance in reading comprehension tests in printed texts compared to the same text in digital and according to Spencer (2006) college students prefer to read in print vs. digital texts. In reading the written text, metacognitive strategies are involved ( Amril et al., 2019 ) but studies ( Channa et al., 2018 ) seem to indicate that students do not use them for reading comprehension, specifically; Korotaeva (2012) finds that only 7% of students use them. Concerning the type of text and difficulties, for Wolfe and Woodwyk (2010) , expository texts benefit more from the construction of a situational model of the text than narrative texts, although Feng (2011) finds that expository texts are more difficult to read than narrative texts. Regarding the modality of the text, Mayer (2009) and Guo et al. (2020) indicate that multimodal texts that incorporate images into the text positively improve reading comprehension. In a study of Kobayashi (2002) using open questions, close, and multiple-choice shows that the type and format of the reading comprehension assessment test significantly influence student performance and that more structured tests help to better differentiate the good ones and the poor ones in reading comprehension. Finally, about student diversity, studies link reading comprehension with the interest and intrinsic motivation of university students ( Cartwright et al., 2019 ; Dewi et al., 2020 ), with gender ( Saracaloglu and Karasakaloglu, 2011 ), finding that women present a better level of reading comprehension than men and with knowledge related to reading ( Perfetti et al., 1987 ). In this research, it was controlled that all were printed and unimodal texts, that is, only text. This is essential because the cognitive processes involved in reading comprehension can vary with these factors ( Butcher and Kintsch, 2003 ; Xu et al., 2020 ).

The Present Study

Regardless of the educational context, in any university discipline, preparing essays or developing arguments are formative tasks that require a deep level of reading comprehension (inferences and transformation of information) that allows the elaboration of a situation model, and not having this level can lead to limited formative learning. Therefore, the objective of this research was to know the state of reading comprehension levels in higher education; specifically, the proportion of university students who perform optimally at each level of reading comprehension. It is important to note that there is not much information about the different levels in university students and that it is the only meta-analytic review that explores different levels of reading comprehension in this educational stage. This is a relevant issue because the university system requires that students produce knowledge from the critical reflection of scientific texts, preparing them for innovation, employability, and coexistence in society.

Materials and Methods

Eligibility criteria: inclusion and exclusion.

Empirical studies written in Spanish or English are selected that analyze the reading comprehension level in university students.

The exclusion criteria are as follows: (a) book chapters or review books or publications; (b) articles in other languages; (c) studies of lower educational levels; (d) articles that do not identify the age of the sample; (e) second language studies; (f) students with learning difficulties or other disorders; (g) publications that do not indicate the level of reading comprehension; (h) studies that relate reading competence with other variables but do not report reading comprehension levels; (i) pre-post program application work; (j) studies with experimental and control groups; (k) articles comparing pre-university stages or adults; (l) publications that use multi-texts; (m) studies that use some type of technology (computer, hypertext, web, psychophysiological, online questionnaire, etc.); and (n) studies unrelated to the subject of interest.

Only those publications that meet the following criteria are included as: (a) be empirical research (article, thesis, final degree/master’s degree, or conference proceedings book); (b) university stage; (c) include data or some measure on the level of reading comprehension that allows calculating the effect size; (d) written in English or Spanish; (e) reading comprehension in the first language or mother tongue; and (f) the temporary period from January 2010 to March 2021.

Search Strategies

A three-step procedure is used to select the studies included in the meta-analysis. In the first step, a review of research and empirical articles in English and Spanish from January 2010 to March 2021. The search is carried out in online databases of languages in Spanish and English, such as Web of Science (WoS), Scopus, Medline, and PsycINFO, to review empirical productions that analyze the level of reading comprehension in university students. In the second step, the following terms (titles, abstracts, keywords, and full text) are used to select the articles: Reading comprehension and higher education, university students, in Spanish and English, combined with the Boolean operators AND and OR. In the last step, secondary sources, such as the Google search engine, Theseus, and references in publications, are explored.

The search reports 4,294 publications (articles, theses, and conference proceedings books) in the databases and eight records of secondary references, specifically, 1989 from WoS, 2001 from Scopus, 42 from Medline, and 262 of PsycINFO. Of the total (4,294), 1,568 are eliminated due to duplications, leaving 2,734 valid records. Next, titles and abstracts are reviewed and 2,659 are excluded because they do not meet the inclusion criteria. The sample of 75 publications is reduced to 40 articles, excluding 35 because the full text cannot be accessed (the authors were contacted but did not respond), the full text did not show specific statistical data, they used online questionnaires or computerized presentations of the text. Finally, seven articles in Spanish were selected for use in the meta-analysis of the reading comprehension level of university students. Data additional to those included in the articles were not requested from the selected authors.

The PRISMA-P guidelines ( Moher et al., 2015 ) are followed to perform the meta-analysis and the flow chart for the selection of publications relevant to the subject is exposed (Figure 1) .

Figure 1 . Flow diagram for the selection of articles.

Encoding Procedure

This research complies with what is established in the manual of systematic reviews ( Higgins and Green, 2008 ) in which clear objectives, specific search terms, and eligibility criteria for previously defined works are established. Two independent coders, reaching a 100% agreement, carry out the study search process. Subsequently, the research is codified, for this, a coding protocol is used as a guide to help resolve the ambiguities between the coders; the proposals are reflected and discussed and discrepancies are resolved, reaching a degree of agreement between the two coders of 97%.

For all studies, the reference, country, research objective, sample size, age and gender, reading comprehension test, other tests, and reading comprehension results were coded in percentages. All this information was later systematized in Table 1 .

Table 1 . Results of the empirical studies included in the meta-analysis.

In relation to the type of reading comprehension level, it was coded based on the levels of the scientific literature as follows: 1 = literal; 2 = inferential; 3 = critical; and 4 = organizational.

Regarding the possible moderating variables, it was coded if the investigations used a standardized reading comprehension measure (value = 1) or non-standardized (value = 0). This research considers the standardized measures of reading comprehension as the non-standardized measures created by the researchers themselves in their studies or questionnaires by other authors. By the type of evaluation test, we encode between multiple-choice (value = 0) or multiple-choices plus open question (value = 1). By type of text, we encode between argumentative (value = 1) or unknown (value = 0). By the type of career, we encode social sciences (value = 1) or other careers (health sciences; value = 0). Moreover, by the type of publication, we encode between article (value = 1) or doctoral thesis (value = 0).

Effect Size and Statistical Analysis

This descriptive study with a sample k = 7 and a population of 1,044 university students used a continuous variable and the proportions were used as the effect size to analyze the proportion of students who had an optimal performance at each level of reading comprehension. As for the percentages of each level of reading comprehension of the sample, they were transformed into absolute frequencies. A random-effects model ( Borenstein et al., 2009 ) was used as the effect size. These random-effects models have a greater capacity to generalize the conclusions and allow estimating the effects of different sources of variation (moderating variables). The DerSimonian and Laird method ( Egger et al., 2001 ) was used, calculating raw proportion and for each proportion its standard error, value of p and 95% confidence interval (CI).

To examine sampling variability, Cochran’s Q test (to test the null hypothesis of homogeneity between studies) and I 2 (proportion of variability) were used. According to Higgins et al. (2003) , if I 2 reaches 25%, it is considered low, if it reaches 50% and if it exceeds 75% it is considered high. A meta-regression analysis was used to investigate the effect of the moderator variables (type of measure, type of evaluation test, type of text, type of career, and type of publication) in each level of reading comprehension of the sample studies. For each moderating variable, all the necessary statistics were calculated (estimate, standard error, CI, Q , and I 2 ).

To compare the effect sizes of each level (literal, inferential, critical, and organizational) of reading comprehension, the chi-square test for the proportion recommended by Campbell (2007) was used.

Finally, to analyze publication bias, this study uses two ways: Rosenthal’s fail-safe number and regression test. Rosenthal’s fail-safe number shows the number of missing studies with null effects that would make the previous correlations insignificant ( Borenstein et al., 2009 ). When the values are large there is no bias. In the regression test, when the regression is not significant, there is no bias.

The software used to classify and encode data and produce descriptive statistics was with Microsoft Excel and the Jamovi version 1.6 free software was used to perform the meta-analysis.

The results of the meta-analysis are presented in three parts: the general descriptive analysis of the included studies; the meta-analytic analysis with the effect size, heterogeneity, moderating variables, and comparison of effect sizes; and the study of publication bias.

Overview of Included Studies

The search carried out of the scientific literature related to the subject published from 2010 to March 2021 generated a small number of publications, because it was limited to the higher education stage and required clear statistical data on reading comprehension.

Table 1 presents all the publications reviewed in this meta-analysis with a total of students evaluated in the reviewed works that amounts to 1,044, with the smallest sample size of 30 ( Del Pino-Yépez et al., 2019 ) and the largest with 570 ( Guevara Benítez et al., 2014 ). Regarding gender, 72% women and 28% men were included. Most of the sample comes from university degrees in social sciences, such as psychology and education (71.42%) followed by health sciences (14.28%) engineering and a publication (14.28%) that does not indicate origin. These publications selected according to the inclusion criteria for the meta-analysis come from more countries with a variety of educational systems, but all from South America. Specifically, the countries that have more studies are Mexico (28.57%) and Colombia, Chile, Bolivia, Peru, and Ecuador with 14.28% each, respectively. The years in which they were published are 2.57% in 2018 and 2016 and 14.28% in 2019, 2014, and 2013.

A total of 57% of the studies analyze four levels of reading comprehension (literal, inferential, critical, and organizational) and 43% investigate three levels of reading comprehension (literal, inferential, and critical). Based on the moderating variables, 57% of the studies use standardized reading comprehension measures and 43% non-standardized measures. According to the evaluation test used, 29% use multiple-choice questions and 71% combine multiple-choice questions plus open questions. 43% use an argumentative text and 57% other types of texts (not indicated in studies). By type of career, 71% are students of social sciences and 29% of other different careers, such as engineering or health sciences. In addition, 71% are articles and 29% with research works (thesis and degree works).

Table 2 shows the reading comprehension assessment instruments used by the authors of the empirical research integrated into the meta-analysis.

Table 2 . Reading comprehension assessment tests used in higher education.

Meta-Analytic Analysis of the Level of Reading Comprehension

The literal level presents a mean proportion effect size of 56% (95% CI = 39–72%; Figure 2 ). The variability between the different samples of the literal level of reading comprehension was significant ( Q = 162.066, p < 0.001; I 2 = 96.3%). No moderating variable used in this research had a significant contribution to heterogeneity: type of measurement ( p = 0.520), type of test ( p = 0.114), type of text ( p = 0.520), type of career ( p = 0.235), and type of publication ( p = 0.585). The high variability is explained by other factors not considered in this work, such as the characteristics of the students (cognitive abilities) or other issues.

Figure 2 . Forest plot of literal level.

The inferential level presents a mean proportion effect size of 33% (95% CI = 19–46%; Figure 3 ). The variability between the different samples of the inferential level of reading comprehension was significant ( Q = 125.123, p < 0.001; I 2 = 95.2%). The type of measure ( p = 0.011) and the type of text ( p = 0.011) had a significant contribution to heterogeneity. The rest of the variables had no significance: type of test ( p = 0.214), type of career ( p = 0.449), and type of publication ( p = 0.218). According to the type of measure, the proportion of students who have an optimal level in inferential administering a standardized test is 28.7% less than when a non-standardized test is administered. The type of measure reduces variability by 2.57% and explains the differences between the results of the studies at the inferential level. According to the type of text, the proportion of students who have an optimal level in inferential using an argumentative text is 28.7% less than when using another type of text. The type of text reduces the variability by 2.57% and explains the differences between the results of the studies at the inferential level.

Figure 3 . Forest plot of inferential level.

The critical level has a mean effect size of the proportion of 22% (95% CI = 9–35%; Figure 4 ). The variability between the different samples of the critical level of reading comprehension was significant ( Q = 627.044, p < 0.001; I 2 = 99.04%). No moderating variable used in this research had a significant contribution to heterogeneity: type of measurement ( p = 0.575), type of test ( p = 0.691), type of text ( p = 0.575), type of career ( p = 0.699), and type of publication ( p = 0.293). The high variability is explained by other factors not considered in this work, such as the characteristics of the students (cognitive abilities).

Figure 4 . Forest plot of critical level.

The organizational level presents a mean effect size of the proportion of 22% (95% CI = 6–37%; Figure 5 ). The variability between the different samples of the organizational level of reading comprehension was significant ( Q = 1799.366, p < 0.001; I 2 = 99.67%). The type of test made a significant contribution to heterogeneity ( p = 0.289). The other moderating variables were not significant in this research: type of measurement ( p = 0.289), type of text ( p = 0.289), type of career ( p = 0.361), and type of publication ( p = 0.371). Depending on the type of test, the proportion of students who have an optimal level in organizational with multiple-choices tests plus open questions is 37% higher than while using only multiple-choice tests. The type of text reduces the variability by 0.27% and explains the differences between the results of the studies at the organizational level.

Figure 5 . Forest plot of organizational level.

Table 3 shows the difference between the estimated effect sizes and the significance. There is a larger proportion of students having an optimal level of reading comprehension at the literal level compared to the inferential, critical, and organizational level; an optimal level of reading comprehension at the inferential level vs. the critical and organizational level.

Table 3 . Results of effect size comparison.

Analysis of Publication Bias

This research uses two ways to verify the existence of bias independently of the sample size. Table 4 shows the results and there is no publication bias at any level of reading comprehension.

Table 4 . Publication bias results.

This research used a systematic literature search and meta-analysis to provide estimates of the number of cases of university students who have an optimal level in the different levels of reading comprehension. All the information available on the subject at the international level was analyzed using international databases in English and Spanish, but the potentially relevant publications were limited. Only seven Spanish language studies were identified internationally. In these seven studies, the optimal performance at each level of reading comprehension varied, finding heterogeneity associated with the very high estimates, which indicates that the summary estimates have to be interpreted with caution and in the context of the sample and the variables used in this meta-analysis.

In this research, the effects of the type of measure, type of test, type of text, type of career, and type of publication have been analyzed. Due to the limited information in the publications, it was not possible to assess the effect of any more moderating variables.

We found that some factors significantly influence heterogeneity according to the level of reading comprehension considered. The type of measure influenced the optimal performance of students in the inferential level of reading comprehension; specifically, the proportion of students who have an optimal level in inferential worsens if the test is standardized. Several studies ( Pike, 1996 ; Koretz, 2002 ) identify differences between standardized and non-standardized measures in reading comprehension and a favor of non-standardized measures developed by the researchers ( Pyle et al., 2017 ). The ability to generate inferences of each individual may difficult to standardize because each person differently identifies the relationship between the parts of the text and integrates it with their previous knowledge ( Oakhill, 1982 ; Cain et al., 2004 ). This mental representation of the meaning of the text is necessary to create a model of the situation and a deep understanding ( McNamara and Magliano, 2009 ; van den Broek and Espin, 2012 ).

The type of test was significant for the organizational level of reading comprehension. The proportion of students who have an optimal level in organizational improves if the reading comprehension assessment test is multiple-choice plus open questions. The organizational level requires the reordering of written information through analysis and synthesis processes ( Guevara Benítez et al., 2014 ); therefore, it constitutes a production task that is better reflected in open questions than in reproduction questions as multiple choice ( Dinsmore and Alexander, 2015 ). McNamara and Kintsch (1996) identify that open tasks require an effort to make inferences related to previous knowledge and multidisciplinary knowledge. Important is to indicate that different evaluation test formats can measure different aspects of reading comprehension ( Zheng et al., 2007 ).

The type of text significantly influenced the inferential level of reading comprehension. The proportion of students who have an optimal level in inferential decreases with an argumentative text. The expectations created before an argumentative text made it difficult to generate inferences and, therefore, the construction of the meaning of the text. This result is in the opposite direction to the study by Diakidoy et al. (2011) who find that the refutation text, such as the argumentative one, facilitates the elaboration of inferences compared to other types of texts. It is possible that the argumentative text, given its dialogical nature of arguments and counterarguments, with a subject unknown by the students, has determined the decrease of inferences based on their scarce previous knowledge of the subject, needing help to elaborate the structure of the text read ( Reznitskaya et al., 2007 ). It should be pointed out that in meta-analysis studies, 43% use argumentative texts. Knowing the type of the text is relevant for generating inferences, for instance, according to Baretta et al. (2009) the different types of text are processed differently in the brain generating more or fewer inferences; specifically, using the N400 component, they find that expository texts generate more inferences from the text read.

For the type of career and the type of publication, no significance was found at any level of reading comprehension in this sample. This seems to indicate that university students have the same level of performance in tasks of literal, critical inferential, and organizational understanding regardless of whether they are studying social sciences, health sciences, or engineering. Nor does the type of publication affect the state of the different levels of reading comprehension in higher education.

The remaining high heterogeneity at all levels of reading comprehension was not captured in this review, indicating that there are other factors, such as student characteristics, gender, or other issues, that are moderating and explaining the variability at the literal, inferential, critical, and organizational reading comprehension in university students.

To the comparison between the different levels of reading comprehension, the literal level has a significantly higher proportion of students with an optimal level than the inferential, critical, and organizational levels. The inferential level has a significantly higher proportion of students with an optimal level than the critical and organizational levels. This corresponds with data from other investigations ( Márquez et al., 2016 ; Del Pino-Yépez et al., 2019 ) that indicate that the literal level is where university students execute with more successes, being more difficult and with less success at the inferential, organizational, and critical levels. This indicates that university students of this sample do not generate a coherent situation model that provides them with a global mental representation of the read text according to the model of Kintsch (1998) , but rather they make a literal analysis of the explicit content of the read text. This level of understanding can lead to less desirable results in educational terms ( Dinsmore and Alexander, 2015 ).

The educational implications of this meta-analysis in this sample are aimed at making universities aware of the state of reading comprehension levels possessed by university students and designing strategies (courses and workshops) to optimize it by improving the training and employability of students. Some proposals can be directed to the use of reflection tasks, integration of information, graphic organizers, evaluation, interpretation, nor the use of paraphrasing ( Rahmani, 2011 ). Some studies ( Hong-Nam and Leavell, 2011 ; Parr and Woloshyn, 2013 ) demonstrate the effectiveness of instructional courses in improving performance in reading comprehension and metacognitive strategies. In addition, it is necessary to design reading comprehension assessment tests in higher education that are balanced, validated, and reliable, allowing to have data for the different levels of reading comprehension.

Limitations and Conclusion

This meta-analysis can be used as a starting point to report on reading comprehension levels in higher education, but the results should be interpreted with caution and in the context of the study sample and variables. Publications without sufficient data and inaccessible articles, with a sample of seven studies, may have limited the international perspective. The interest in studying reading comprehension in the mother tongue, using only unimodal texts, without the influence of technology and with English and Spanish has also limited the review. The limited amount of data in the studies has limited meta-regression.

This review is a guide to direct future research, broadening the study focus on the level of reading comprehension using digital technology, experimental designs, second languages, and investigations that relate reading comprehension with other factors (gender, cognitive abilities, etc.) that can explain the heterogeneity in the different levels of reading comprehension. The possibility of developing a comprehensive reading comprehension assessment test in higher education could also be explored.

This review contributes to the scientific literature in several ways. In the first place, this meta-analytic review is the only one that analyzes the proportion of university students who have an optimal performance in the different levels of reading comprehension. This review is made with international publications and this topic is mostly investigated in Latin America. Second, optimal performance can be improved at all levels of reading comprehension, fundamentally inferential, critical, and organizational. The literal level is significantly the level of reading comprehension with the highest proportion of optimal performance in university students. Third, the students in this sample have optimal performance at the inferential level when they are non-argumentative texts and non-standardized measures, and, in the analyzed works, there is optimal performance at the organizational level when multiple-choice questions plus open questions are used.

The current research is linked to the research project “Study of reading comprehension in higher education” of Asociación Educar para el Desarrollo Humano from Argentina.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Author Contributions

Cd-l-P had the idea for the article and analyzed the data. ML-R searched the data. Cd-l-P and ML-R selected the data and contributed to the valuable comments and manuscript writing. All authors contributed to the article and approved the submitted version.

This paper was funded by the Universidad Internacional de la Rioja and Universidad de Málaga.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling editor declared a shared affiliation though no other collaboration with one of the authors ML-R at the time of the review.

Publisher’s Note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Afflerbach, P., Cho, B.-Y., and Kim, J.-Y. (2015). Conceptualizing and assessing higher-order thinking in reading. Theory Pract. 54, 203–212. doi: 10.1080/00405841.2015.1044367

CrossRef Full Text | Google Scholar

Alexander, P. A. (2004). “A model of domain learning: reinterpreting expertise as a multidimensional, multistage process,” in Motivation, Emotion, and Cognition: Integrative Perspectives on Intellectual Functioning and Development. eds. D. Y. Dai and R. J. Sternberg (Mahwah, NJ: Erlbaum), 273–298.

Google Scholar

Amril, A., Hasanuddin, W. S., and Atmazaki, (2019). The contributions of reading strategies and reading frequencies toward students’ reading comprehension skill in higher education. Int. J. Eng. Adv. Technol. (IJEAT) 8, 593–595. doi: 10.35940/ijeat.F1105.0986S319

Baretta, L., Braga Tomitch, L. M., MacNair, N., Kwan Lim, V., and Waldie, K. E. (2009). Inference making while reading narrative and expository texts: an ERP study. Psychol. Neurosci. 2, 137–145. doi: 10.3922/j.psns.2009.2.005

Barletta, M., Bovea, V., Delgado, P., Del Villar, L., Lozano, A., May, O., et al. (2005). Comprensión y Competencias Lectoras en Estudiantes Universitarios. Barranquilla: Uninorte.

Bharuthram, S. (2012). Making a case for the teaching of reading across the curriculum in higher education. S. Afr. J. Educ. 32, 205–214. doi: 10.15700/saje.v32n2a557

Borenstein, M., Hedges, L. V., Higgins, J. P. T., and Rothstein, H. R. (2009). Introduction to Meta-Analysis. United Kingdom: John Wiley and Sons, Ltd, 45–49.

Butcher, K. R., and Kintsch, W. (2003). “Text comprehension and discourse processing,” in Handbook of Psychology: Experimental Psychology. 2nd Edn . Vol . 4. eds. A. F. Healy, R. W. Proctor, and I. B. Weiner (New Jersey: John Wiley and Sons, Inc.), 575–595.

Cain, K., Oakhill, J., and Bryant, P. (2004). Children’s reading comprehension ability: concurrent prediction by working memory, verbal ability, and component skills. J. Educ. Psychol. 96, 31–42. doi: 10.1037/0022-0663.96.1.31

Campbell, I. (2007). Chi-squared and Fisher-Irwin tests of two-by-two tables with small sample recommendations. Stat. Med. 26, 3661–3675. doi: 10.1002/sim.2832

PubMed Abstract | CrossRef Full Text | Google Scholar

Cartwright, K. B., Lee, S. A., Barber, A. T., DeWyngaert, L. U., Lane, A. B., and Singleton, T. (2019). Contributions of executive function and cognitive intrinsic motivation to university students’ reading comprehension. Read. Res. Q. 55, 345–369. doi: 10.1002/rrq.273

Channa, M. A., Abassi, A. M., John, S., and Sahito, J. K. M. (2018). Reading comprehension and metacognitive strategies in first-year engineering university students in Pakistan. Int. J. Engl. Ling. 8, 78–87. doi: 10.5539/ijel.v8n6p78

Cox, S. R., Friesner, D. L., and Khayum, M. (2014). Do Reading skills courses help underprepared readers achieve academic success in college? J. College Reading and Learn. 33, 170–196. doi: 10.1080/10790195.2003.10850147

Del Pino-Yépez, G. M., Saltos-Rodríguez, L. J., and Moreira-Aguayo, P. Y. (2019). Estrategias didácticas para el afianzamiento de la comprensión lectora en estudiantes universitarios. Revista científica Dominio de las Ciencias 5, 171–187. doi: 10.23857/dc.v5i1.1038

Dewi, R. S., Fahrurrozi, Hasanah, U., and Wahyudi, A. (2020). Reading interest and Reading comprehension: a correlational study in Syarif Hidayatullah State Islamic University, Jakarta. Talent Dev. Excell. 12, 241–250.

Diakidoy, I. N., Mouskounti, T., and Ioannides, C. (2011). Comprehension and learning from refutation and expository texts. Read. Res. Q. 46, 22–38. doi: 10.1598/RRQ.46.1.2

Dinsmore, D. J., and Alexander, P. A. (2015). A multidimensional investigation of deep-level and surface-level processing. J. Exp. Educ. 84, 213–244. doi: 10.1080/00220973.2014.979126

Egger, M., Smith, D., and Altmand, D. G. (2001). Systematic Reviews in Health Care: Meta-Analysis in Context. London: BMJ Publishing Group.

Feng, L. (2011). A short analysis of the text variables affecting reading and testing reading. Stud. Lit. Lang. 2, 44–49.

Figueroa Romero, R. L., Castañeda Sánchez, W., and Tamay Carranza, I. A. (2016). Nivel de comprensión lectora en los estudiantes del primer ciclo de la Universidad San Pedro, filial Caraz, 2016. (Trabajo de investigación, Universidad San Pedro). Repositorio Institucional USP. Available at: http://repositorio.usanpedro.edu.pe/bitstream/handle/USANPEDRO/305/PI1640418.pdf?sequence=1andisAllowed=y (Accessed February 15, 2021).

Guevara Benítez, Y., Guerra García, J., Delgado Sánchez, U., and Flores Rubí, C. (2014). Evaluación de distintos niveles de comprensión lectora en estudiantes mexicanos de Psicología. Acta Colombiana de Psicología 17, 113–121. doi: 10.14718/ACP.2014.17.2.12

Guo, D., Zhang, S., Wright, K. L., and McTigue, E. M. (2020). Do you get the picture? A meta-analysis of the effect of graphics on reading comprehension. AERA Open 6, 1–20. doi: 10.1177/2332858420901696

Higgins, J. P., and Green, S. (2008). Cochrane Handbook for Systematic Reviews of Interventions. The Cochrane Collaboration.

Higgins, J. P., Thompson, S. G., Deeks, J. J., and Altman, D. G. (2003). Measuring inconsistency in meta-analyses. BMJ 327, 327–557. doi: 10.1136/bmj.327.7414.557

Hong-Nam, K., and Leavell, A. G. (2011). Reading strategy instruction, metacognitive awareness, and self-perception of striving college developmental readers. J. College Literacy Learn. 37, 3–17.

Kazazoglu, S. (2020). Is printed text the best choice? A mixed-method case study on reading comprehension. J. Lang. Linguistic Stud. 16, 458–473. doi: 10.17263/jlls.712879

Kintsch, W. (1998). Comprehension: A Paradigm for Cognition. New York: Cambridge University Press.

Kintsch, W., and van Dijk, T. A. (1978). Toward a model of text comprehension and production. Psychol. Rev. 85, 363–394. doi: 10.1037/0033-295X.85.5.363

Kobayashi, M. (2002). Method effects on reading comprehension test performance: test organization and response format. Lang. Test. 19, 193–220. doi: 10.1191/0265532202lt227oa

Koretz, D. (2002). Limitations in the use of achievement tests as measures of educators’ productivity. J. Hum. Resour. 37, 752–777. doi: 10.2307/3069616

Korotaeva, I. V. (2012). Metacognitive strategies in reading comprehension of education majors. Procedural-Social and Behav. Sci. 69, 1895–1900. doi: 10.1016/j.sbspro.2012.12.143

Kucer, S. B. (2016). Accuracy, miscues, and the comprehension of complex literary and scientific texts. Read. Psychol. 37, 1076–1095. doi: 10.1080/02702711.2016.1159632

Livingston, C., Klopper, B., Cox, S., and Uys, C. (2015). The impact of an academic reading program in the bachelor of education (intermediate and senior phase) degree. Read. Writ. 6, 1–11. doi: 10.4102/rw.v6i1.66

Márquez, H., Díaz, C., Muñoz, R., and Fuentes, R. (2016). Evaluación de los niveles de comprensión lectora en estudiantes universitarios pertenecientes a las carreras de Kinesiología y Nutrición y Dietética de la Universidad Andrés Bello, Concepción. Revista de Educación en Ciencias de la Salud 13, 154–160.

Mayer, R. E. (ed.) (2009). “Modality principle,” in Multimedia Learning. United States: Cambridge University Press, 200–2020.

McNamara, D. S., and Kintsch, W. (1996). Learning from texts: effects of prior knowledge and text coherence. Discourse Process 22, 247–288. doi: 10.1080/01638539609544975

McNamara, D. S., and Magliano, J. (2009). “Toward a comprehensive model of comprehension,” in The psychology of learning and motivation. ed. E. B. Ross (New York: Elsevier), 297–384.

Moher, D., Shamseer, L., Clarke, M., Ghersi, D., Liberati, A., Petticrew, M., et al. (2015). Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Syst. Rev. 4:1. doi: 10.1186/2046-4053-4-1

Ntereke, B. B., and Ramoroka, B. T. (2017). Reading competency of first-year undergraduate students at the University of Botswana: a case study. Read. Writ. 8:a123. doi: 10.4102/rw.v8i1.123

Oakhill, J. (1982). Constructive processes in skilled and less skilled comprehenders’ memory for sentences. Br. J. Psychol. 73, 13–20. doi: 10.1111/j.2044-8295.1982.tb01785.x

Parr, C., and Woloshyn, V. (2013). Reading comprehension strategy instruction in a first-year course: an instructor’s self-study. Can. J. Scholarship Teach. Learn. 4:3. doi: 10.5206/cjsotl-rcacea.2013.2.3

Perfetti, C. A., Beck, I., Bell, L. C., and Hughes, C. (1987). Phonemic knowledge and learning to read are reciprocal: a longitudinal study of first-grade children. Merrill-Palmer Q. 33, 283–319.

Pike, G. R. (1996). Limitations of using students’ self-reports of academic development as proxies for traditional achievement measures. Res. High. Educ. 37, 89–114. doi: 10.1007/BF01680043

Pyle, N., Vasquez, A. C., Lignugaris, B., Gillam, S. L., Reutzel, D. R., Olszewski, A., et al. (2017). Effects of expository text structure interventions on comprehension: a meta-analysis. Read. Res. Q. 52, 469–501. doi: 10.1002/rrq.179

Rahmani, M. (2011). Effects of note-taking training on reading comprehension and recall. Reading Matrix: An Int. Online J. 11, 116–126.

Reznitskaya, A., Anderson, R., and Kuo, L. J. (2007). Teaching and learning argumentation. Elem. Sch. J. 107, 449–472. doi: 10.1086/518623

Sáez Sánchez, B. K. (2018). La comprensión lectora en jóvenes universitarios de una escuela formadora de docentes. Revista Electrónica Científica de Investigación Educativa 4, 609–618.

Sanabria Mantilla, T. R. (2018). Relación entre comprensión lectora y rendimiento académico en estudiantes de primer año de Psicología de la Universidad Pontificia Bolivariana. (Trabajo de grado, Universidad Pontificia Bolivariana). Repositorio Institucional UPB. Available at: https://repository.upb.edu.co/bitstream/handle/20.500.11912/5443/digital_36863.pdf?sequence=1andisAllowed=y (Accessed February 15, 2021).

Saracaloglu, A. S., and Karasakaloglu, N. (2011). An investigation of prospective teachers’ reading comprehension levels and study and learning strategies related to some variables. Egit. ve Bilim 36, 98–115.

Smagorinsky, P. (2001). If meaning is constructed, what is it made from? Toward a cultural theory of reading. Rev. Educ. Res. 71, 133–169. doi: 10.3102/00346543071001133

Spencer, C. (2006). Research on learners’ preferences for reading from a printed text or a computer screen. J. Dist. Educ. 21, 33–50.

Trakhman, L. M. S., Alexander, P., and Berkowitz, L. E. (2019). Effects of processing time on comprehension and calibration in print and digital mediums. J. Exp. Educ. 87, 101–115. doi: 10.1080/00220973.2017.1411877

Tuncer, M., and Bhadir, F. (2014). Effect of screen reading and reading from printed out material on student success and permanency in introduction to computer lesson. Turk. Online J. Educ. Technol. 13, 41–49.

Ulum, O. G. (2016). A descriptive content analysis of the extent of Bloom’s taxonomy in the reading comprehension questions of the course book Q: skills for success 4 reading and writing. Qual. Rep. 21, 1674–1683.

UNESCO (2009). “Conferencia mundial sobre la Educación Superior – 2009.” La nueva dinámica de la educación superior y la investigación para el cambio social y el desarrollo; July 5-8, 2009; Paris.

van den Broek, P., and Espin, C. A. (2012). Connecting cognitive theory and assessment: measuring individual differences in reading comprehension. Sch. Psychol. Rev. 43, 315–325. doi: 10.1080/02796015.2012.12087512

van den Broek, P., Mouw, J. M., and Kraal, A. (2016). “Individual differences in reading comprehension,” in Handbook of Individual Differences in Reading: Reader, Text, and Context. ed. E. P. Afflerbach (New York: Routledge), 138–150.

Wolfe, M. B. W., and Woodwyk, J. M. (2010). Processing and memory of information presented in narrative or expository texts. Br. J. Educ. Psychol. 80, 341–362. doi: 10.1348/000709910X485700

Xu, Y., Wong, R., He, S., Veldre, A., and Andrews, S. (2020). Is it smart to read on your phone? The impact of reading format and culture on the continued influence of misinformation. Mem. Cogn. 48, 1112–1127. doi: 10.3758/s13421-020-01046-0

Yáñez Botello, C. R. (2013). Caracterización de los procesos cognoscitivos y competencias involucrados en los niveles de comprensión lectora en Estudiantes Universitarios. Cuadernos Hispanoamericanos de Psicología 13, 75–90. doi: 10.18270/chps.v13i2.1350

Yussof, Y. M., Jamian, A. R., Hamzah, Z. A. Z., and Roslan, A. (2013). Students’ reading comprehension performance with emotional. Int. J. Edu. Literacy Stud. 1, 82–88. doi: 10.7575/aiac.ijels.v.1n.1p.82

Zheng, Y., Cheng, L., and Klinger, D. A. (2007). Do test format in reading comprehension affect second-language students’ test performance differently? TESL Can. J. 25, 65–78. doi: 10.18806/tesl.v25i1.108

Keywords: reading comprehension, higher education, university students, systematic review, meta-analysis

Citation: de-la-Peña C and Luque-Rojas MJ (2021) Levels of Reading Comprehension in Higher Education: Systematic Review and Meta-Analysis. Front. Psychol . 12:712901. doi: 10.3389/fpsyg.2021.712901

Received: 21 May 2021; Accepted: 07 July 2021; Published: 04 August 2021.

Reviewed by:

Copyright © 2021 de-la-Peña and Luque-Rojas. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Cristina de-la-Peña, [email protected]

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Reading print is better for comprehension, study finds

Leisure reading on paper helps with text comprehension better than reading on digital devices, according to a new study.

Driving the news: "The main conclusion is that leisure reading habits on screen are minimally related to reading comprehension," researchers at the University of Valencia found.

By the numbers: The researchers, who analyzed more than two dozen studies, said "the relationship between the frequency of reading printed texts and text comprehension is much higher (between 0.30 and 0.40) than what we found for leisure digital reading habits (0.05)."

• "This means, for example, that if a student spends 10 hours reading books on paper, their comprehension will probably be 6 to 8 times greater than if they read on digital devices for the same amount of time," study co-authors Cristina Vargas and Ladislao Salmerón said .

Of note: The study also found that as students get older, the relationship between recreational reading on digital devices and text comprehension improves.

Details: The researchers analyzed 25 studies on reading comprehension published between 2000 and 2022, with more than 450,000 participants.

• "One might have expected that reading for informational purposes (i.e., visiting Wikipedia or other educational websites; reading news, or reading e-books) would be much more positively related to comprehension, but this is not the case," one researcher said.
• The study was published earlier this week in the Review of Educational Research .

Go deeper: Attempts to ban books at public libraries surge at record levels

Get the rundown of the biggest stories of the day with Axios Daily Essentials.