Introduction

Reading literacy is gaining more worldwide interest as an important contributor to the development of human language and understanding of the world (e.g., Cai and Zhu, 2016; Ho and Lau, 2018; Nadori, 2020; Solheim and Lundetræ, 2016). In international education, where reading has become one of the main subjects in the field of assessment, the fact that reading is highly valued (e.g., Eelly, 1992; Lee, 2012; Mak et al. 2017; Ho and Lau, 2018; Expósito-Casas et al. 2022; Ma et al. 2023; Mehmet et al. 2023; Bu and Chen, 2023), which is demonstrated in a series of international assessments of reading literacy such as Program for International Student Assessment (PISA), Progress International Reading Literacy Study (PIRLS), and National Assessment of Educational Progress (NAEP).

PISA 2018 is an international assessment instrument measuring 15-year-old students’ literacy in reading, mathematics, and science as well as ability to apply knowledge to solve problems, thereby reflecting students’ potential for participation in life. The core test of PISA 2018 is reading literacy whose results have attracted global attention since its inception in 2019. Certain scholars have devoted themselves to the study of reading literacy (e.g., Yan et al. 2016; Vazquez-Lopez, 2021; Expósito-Casas et al. 2022; Gutiérrez-de-Rozas et al. 2022; Ma et al. 2023; Bernardo and Mante-Estacio, 2023) but previous studies have mainly focused on events in developed countries such as Europe and the United States or countries where students tend to score highly, such as Mainland China. Less attention has been paid to countries and regions where low reading scores are obtained. Only a small number studies have explored reading literacy in low reading performing countries and regions, especially African countries (Trinidad, 2020; Haw et al. 2021; Gutiérrez-de-Rozas et al. 2022).

A large number of studies have shown that students’ reading literacy is influenced by numerous factors (e.g., Lee, 2012; Mak et al. 2017; Ho and Lau, 2018; Miyamoto et al. 2019; Vazquez-Lopez, 2021; Expósito-Casas et al. 2022; Ma et al. 2023). Traditionally, reading literacy studies have been divided into two categories: one has explored the influence of a specific factor on reading literacy whereas the other combined two or more factors to explore the joint and interactive effects of these different influences on reading literacy. The previous studies demonstrated that students’ traits in reading were investigated and that specific cognitive or emotional factors such as reading strategy, reading motivation, and reading meta-cognitive ability had each been shown to have a specific and significant impact on reading literacy (e.g., Lau and Chan, 2003; Miyamoto et al. 2019; Bernardo and Mante-Estacio, 2023). The latter type of research indicated that investigations that were based on multiple perspectives and that centered on the impact of the individual, family, school, society, and even national factors on students’ reading literacy were more fruitful compared with single-perspective studies (e.g., Brozo et al. 2014; Ho and Lau, 2018; Vazquez-Lopez, 2021 Expósito-Casas et al. 2022; Ma et al. 2023; Ma et al. 2023).

Broadly, the present study set out to investigate how multiple factors influence reading literacy in Morocco, a multilingual country where different languages are widely spoken at home. Aside from those mentioned, language is also an important factor affecting reading literacy (Khoudi et al. 2024). As stated earlier, African countries have rarely participated in previous PISA investigations. In PISA 2018, only one African country, Morocco, took part in the PISA survey (OECD, 2019). The mean scores of Moroccan students’ reading literacy were 359 points compared with the average of 487 points for OECD countries, ranking Morocco 73th among 79 countries and regions (OECD, 2019). The present study may help the international audience understand reading education in Morocco in greater depth and also foster the exploration of ways to improve Moroccan reading literacy.

Literature review

Reading literacy

The term ‘reading literacy’ originated from a study conducted by the International Association for the Evaluation of School Achievement in 1991, which expanded the definition of reading skills to include the ability to reflect on and adopt reading as a tool for personal and social purposes (Elley, 1992). PISA 2018 defined reading literacy as understanding, using, evaluating, reflecting on, and engaging with texts in order to achieve one’s goals, develop one’s knowledge and potential, and participate in society (OECD, 2019). The emphasis of the PISA 2018 reading assessment is not to test whether students are capable of reading professionally, but to focus on the constructive, expansive, and critically reflective skills that students apply in the process of reading in order to make meaning from text or to interact with texts. In the present study, the definition of reading literacy used in PISA 2018 was adopted for the exploration of how certain factors affect reading literacy.

Factors predicting reading literacy

Students’ personal factors

Students’ reading literacy, when considered as an expression of achievement, is inevitably influenced by both cognitive and non-cognitive factors (Yonghan, 2011). For example, a large body of research has cast light on the impact of gender on reading (e.g., Stoet and Geary, 2013; Zuze and Reddy, 2014; Mak et al. 2017; Alivernini et al. 2020; Mehmet et al. 2023). Stoet and Geary (2013) revealed a significant difference in reading literacy for boys versus girls, where girls’ scores were significantly higher than those of boys, and Mak et al. (2017) showed that gender and school gender composition mediated students’ reading achievement.

Meta-cognition is also a factor affecting reading. Strong correlations between scores on reading meta-cognition and reading ability have been demonstrated (Artelt et al. 2001; Mak et al. 2017; Bu and Chen, 2023), and meta-cognitive strategies, such as understanding and remembering, summarizing, and assessing credibility, have been found to have specific and strong effects on on-line reading comprehension (Hu and Wang, 2022). These results suggest that the quantity and quality of meta-cognitive strategies utilized by students could affect their learning outcomes.

A further factor affecting reading literacy that has been employed in research is the self-concept of reading (Yonghan 2011; Locher et al. 2021; Walgermo et al. 2018; Bernardo and Mante-Estacio, 2023). It has been shown that the amount of reading, choice of type of book, inherent difficulty and length of reading choice, and intrinsic motivation to read are affected by the self-concept of reading ability (Locher et al. 2021). It has been suggested that students’ self-educational expectations may positively influence their willingness to put in effort, which in turn promotes reading literacy (Kuan, 2011). Similarly, students’ interest in learning has a significant impact on reading literacy (Artelt et al. 2001; Walgermo et al. 2018; Mak et al. 2017).

Family context‐related factors

The family environment is of course one of the most important for students as they mature and have been shown to be a key factor influencing their physical and mental development, and the effect of family on students’ reading literacy cannot be ignored (Yeo et al. 2014; Coleman et al. 1966; Schiff and Lotem, 2011; Charity et al. 2004; Li and Li, 2022). Studies have shown that students’ socio-economic and cultural family background has a significant impact on reading literacy (Coleman et al. 1966; Schiff and Lotem, 2011; Gustafsson et al. 2018; Richter et al. 2021), and on student achievement as well (Coleman et al. 1966). Subsequent research indicated that the causes of academic differences vary according to types of family background (Schiff and Lotem, 2011; Gustafsson et al. 2018). Thus, the home language environment may have a significant impact on students’ reading literacy (MacPhee et al. 2017; Charity et al. 2004). It has also been suggested that parental language influences students’ academic performance and that inconsistency between the home and school language environments puts students in a relatively disadvantaged position, as shown in a lack of support provided at home for the school language environment (MacPhee et al. 2017).

School context‐related factors

As the primary place of learning, schools of course have an impact on students’ reading literacy. School location and type, according to socioeconomic status, play a role in students’ reading literacy (Khoudi et al. 2024; Young and Dolzhenko, 2022; Rutter and Maughan, 2002). As shown in peer group effects, average school socio-economic and cultural backgrounds have a great impact on student academic achievement (Gustafsson et al. 2018). Zuze and Reddy (2014) shed light on the strong links between resources (both material and human) and achievement in reading. The classroom was the main spot where school education was conducted. Teaching effects were shown to largely determine the effectiveness of the function of education, and teacher-student interactions were shown to play a key role in students’ learning and in their feelings about study at school (Mehmet et al. 2023; Hafen et al. 2015). Teacher support indirectly affected students’ academic performance through students’ attitudes toward learning (Dahbi, 2023; Chatri et al. 2021; Hughes et al. 2008); other studies showed that teacher support both directly and indirectly affected student achievement by increasing students’ sense of belonging at school (Lee, 2012). Scores on reading instructional strategies were positively correlated with those on students’ reading literacy, indicating that comprehension at the macro-level has a significant impact on students’ reading literacy, and that collaboration among teachers is conducive to the implementation of reading instructional strategies (Yan and Cai, 2022).

Reading literacy in Morocco

Morocco is a country with a unique culture and a multilingual populace, mainly due to its history and geographical position at the crossroads of sub-Saharan Africa, North Africa, and Europe, and its proximity to the Middle East. The official languages of the country are Arabic and Amazigh. However, the mother tongue and community language spoken by most children is the local dialect Darija or regional variants of Amazigh. The characteristic feature of their language is diglossia, with one language used in formal education and the other used in daily communication (Dahbi, 2023; Rao et al. 2024). Amazigh has its own letters, and although it is recognized as an official language by the Moroccan Constitution, the method of its integration into schools, government offices, and society is under negotiation (Oumar and Chris, 2022).

Morocco participated in PISA 2018 for the first time with Arabic as the test language (OECD, 2019). Its test results ranked relatively low, with even the highest-performing students scoring only around the OECD (2019) average. Education in Morocco is dispersed over a wide range of grade levels compared with certain other countries and regions (e.g., Iceland, Japan, and Norway) where almost 100% of students are enrolled in grade 10 when they participate in the PISA test (OECD, 2019). Moroccan students’ reading literacy is affected by individual, family, and school factors. Therefore, it is worth exploring the effects of multilingualism on students’ reading literacy in Morocco, where there are two official languages: Arabic and Amazigh.

Theoretical framework and research questions

In the present study, Ecological Systems Theory was adopted as a framework for analyzing students’ reading literacy (Bronfenbrenner, 1979). Ecological Systems Theory, as opposed to, for example, Self-Regulated Learning Theory (Bu and Chen, 2023) and Multiple Intelligences Theory (Kallenbach and Viens, 2004) was used, firstly, because it is widely used in empirical research on reading literacy. For example, Li et al. (2023), using 85,102 individuals as their sample, applied the theory to explore the effect of family socioeconomic status (SES) on children’s reading ability. Hu et al. (2022) categorized situational factors regarding fourth graders’ reading achievement for high and low-performing students in the Progress in International Reading Literacy Study of 2016, based on Ecological Systems Theory. In a study based on socio-ecological systems theory, Chiu and Chow (2015) explored the influence of peer family micro-, meso- and national macro-systems on children’s reading literacy. Secondly, the structure of PISA data is nested and has a good fit with Ecological Systems Theory whereby the social environment is viewed in terms of certain hierarchical levels, and individual development is nested within the interactive environmental systems (Zastrow and Kirst-Ashman, 2004), and reading literacy is held to be affected by the interaction between individual behaviors and the environment. Thus, children’s reading literacy is said to be affected not only by the cultural, linguistic, and institutional context of a country, but also by interactive effects among individuals, families, and schools.

The research questions for the present study were as follows:

  1. (1)

    How do student factors (e.g., gender, meta-cognition, reading self-concept, and expected career status) and family factors (e.g., ESCS) affect Moroccan students’ reading literacy?

  2. (2)

    How do school factors (e.g., school location, school type, and school size) affect Moroccan students’ reading literacy?

  3. (3)

    How does the interaction between school and student influence Moroccan students’ reading literacy?

Method

Data and sample

The data used in this study were primarily drawn from the PISA 2018 database which was publicly released in December 2019 by the Organization for Economic Co-operation and Development (OECD). The PISA 2018 reading literacy database covered student background information (e.g., gender, ESCS) as well as reading literacy and its influencing factors for students in 79 countries and regions around the world. The study used the data from PISA 2018 of Morocco, in which a total of 5800 15-year-old students from 179 schools participated. Background information for the sample is shown in Table 1.

Table 1 The basic information of the sample.
Full size table

Variables

The questionnaires were developed by the OECD where the variables employed were selected to meet research needs and statistical requirements.

The dependent variable

The reading literacy test for PISA 2018 contains sub-scales, including cognitive progress and text structure, each of which is given one of ten Plausible Values (PVs) in the database, and a total scale, also given a PV. Cognitive progress consists of locate information, understand, and evaluate, and reflect (OECD, 2019). Text structure was divided into single text and multiple texts (OECD, 2019). The present study used the average of the 5 PV values of reading literacy for data analysis. Morocco’s test language in PISA 2018 was Arabic, but most of the students speak other languages in their homes (Dahbi, 2023). To ensure the validity of the study, we deleted the sample whose usual language at home was Arabic (N = 871), leaving a sample size of 5800 students who did not speak Arabic at home.

Individual student-level variables (Level-1)

Individual student-level variables employed in the study were comprised of two parts: demographic variables, and objective versus subjective variables. The two demographic variables used in the student questionnaire were gender and ESCS, which were control variables. The variables used in the questionnaire were meta-cognition, joy/like reading, self-concept of reading, student expected occupational status, teacher instruction strategies, teacher support in test language lessons, and duration in early childhood education and care, which were continuous variables.

School level variables (Level-2)

School-level variables were obtained from the headmaster’s questionnaire. School location, school type, and school size were selected as continuous variables for data analysis; shortage of educational material, shortage of educational staff, and proportion of highly educated teachers were selected as continuous variable for data analysis.

Specific variable information is shown in Table 2.

Table 2 Descriptive statistics of participants.
Full size table

Hierarchical linear modeling

Hierarchical linear modeling (HLM) was mainly used to analyze the effects of the variables at different levels (Hox, 2010). Considering students were nested in schools in this study, multilevel linear models were used for the analysis. HLM has been used in a number of PISA studies due to the structural characteristics of the data, (e.g., Hu and Yu, 2021; Hu and Wang, 2022; Wu et al. 2019). For example, Hu and Yu (2021) constructed hierarchical linear models to investigate the effects of ICT-based social media on adolescents’ digital reading performance; and Hu and Wang (2022) adopted three-level hierarchical linear modeling to explore the influence of students’ perceptions of instructional quality on their digital reading performance. In the present study, in order to investigate factors influencing the reading literacy of Moroccan students, four models, that is, the null, random coefficients regression, intercept, and full models, were used in the analyses. All models were via HLM 6.08 (Raudenbush et al. 2004). Missing data were deleted before the models were set.

The null model tested whether there were differences in reading literacy among schools and estimated the contribution of between-group variance. The intraclass Correlation Coefficient, i.e., ICC (1), ρ = \({\tau }_{00}\)/(\({\tau }_{00}\) + \({\delta }^{2}\)), indicated the strength of the between-school variance. This model divided the total variation in students’ reading literacy into within-school and between-school variance, and was mainly used to explore whether there were significant differences in Moroccan students’ reading literacy among schools. The null model was expressed as:

$${{\rm{Reading\; literacy\_}}}_{{ij}}={\beta }_{0j}+{r}_{{ij}}$$
$${\beta }_{0j}={\gamma }_{00}+{\mu }_{0j}$$

In the equation, the dependent variable was\(\,{{\rm{Reading\; literacy\_}}}_{{ij}}\), which represented student i in school j. \({\beta }_{0j}\) was the intercept and represented the average score of reading literacy in school j. \({\gamma }_{00}\) was the intercept indicating the grand mean of reading literacy. \({r}_{{ij}}\) showed the within-school variance of reading literacy, and \({\mu }_{0j}\) indicated the variance between schools.

The second model was a random coefficient regression model. This model examined the effect of student-level variables on reading literacy by adding student-level variables to the null model, with the assumption that the effect of school variables was constant across schools. To eliminate covariance, the model group-centered on all independent variables except control variables, and the level-student model was represented as:

$${\displaystyle{\begin{array}{ll}{{\rm{Reading}}\;{\rm{literacy}}}_{{ij}}=\\{\beta }_{0j}+{\beta }_{1j}{\rm{Gender}}+{\beta }_{2j}{\rm{ESCS}}+{\beta }_{3j}{\rm{U}}{\rm{nderstanding}}\;{\rm{and}}\;{\rm{remembering}}\\+{\beta }_{4j}{\rm{S}}{\rm{ummarising}}+{\beta }_{5j}{\rm{A}}{\rm{ssess}}\;{\rm{credibility}}+{\beta }_{6j}{\rm{S}}{\rm{elf}}-{\rm{concept}}\;{\rm{of}}\;{\rm{reading}}\\+{\beta }_{7j}{\rm{J}}{\rm{oy}}/{\rm{L}}{\rm{ike}}\;{\rm{reading}}+{\beta }_{8j}{\rm{SEI}}+{\beta }_{9j}{\rm{Teacher}}-{\rm{directed}}\;{\rm{instruction}}\\+{\beta }_{10j}{\rm{Perceived}}\;{\rm{feedback}}+{\beta }_{11j}{\rm{Adaptation}}\;{\rm{of}}\;{\rm{instruction}}\\+{\beta }_{12j}{\rm{Teacher}}\;{\rm{support}}\;{\rm{in}}\;{\rm{test}}\;{\rm{language}}\;{\rm{lessons}}\\+{\beta }_{13j}{\rm{Duration}}\;{\rm{in}}\;{\rm{early}}\;{\rm{childhood}}\;{\rm{education}}\;{\rm{and}}\;{\rm{care}}+{r}_{{ij}}\end{array}}}$$

In the equation, \({\beta }_{1j} \sim {\beta }_{13j}\) were the student-level regression coefficients indicating the extent to which student-level variables affected reading literacy.

The third model was the intercept-as-outcomes regression, which added a school-level variable to the null model to examine the effect of school factors on students’ reading literacy. To eliminate collinearity, the model grand-centered on all independent variables except control variables, and the model was displayed as:

$${{\rm{Reading}}\; {\rm{literacy}}}_{{ij}}={\beta }_{0j}+{\beta }_{1j}{\rm{Gender}}+{\beta }_{2j}{\rm{ESCS}}+{r}_{{ij}}$$
$$\begin{array}{l}{\beta }_{0j}={\gamma }_{00}+{\gamma }_{01}{\rm{School}}\;{\rm{location}}\\\qquad\qquad\,+{\gamma }_{02}{\rm{School}}\;{\rm{type}}+{\gamma }_{03}{\rm{School}}\;{\rm{size}}\\\qquad\qquad\,+{\gamma }_{04}{\rm{Shortage}}\;{\rm{of}}\;{\rm{educational}}\;{\rm{material}}\\\qquad\qquad\,+{\gamma }_{05}{\rm{Shortage}}\;{\rm{of}}\;{\rm{educational}}\;{\rm{staff}}\\\qquad\qquad\,+{\gamma }_{06}{\rm{Proportion}}\;{\rm{of}}\;{\rm{highly}}\;{\rm{educated}}\;{\rm{teachers}}+{u}_{{ij}}\end{array}$$
$${\beta }_{1j}={\gamma }_{10}+{u}_{1j}$$
$${\beta }_{2j}={\gamma }_{20}+{u}_{2j}$$

In the equation, \({\gamma }_{01} \sim {\gamma }_{06}\) were the school-level regression coefficients, indicating the extent to which school-level variables affected reading literacy.

The fourth model was the complete model, which was built on the random coefficient regression and the intercept-as-outcomes regression to construct the complete model. It examined the joint effect of student-level and school-level variables on reading literacy. To eliminate covariance, the model group-centered on the student-level variables and grand-centered on school-level variables. The full model was shown as:

$$\begin{array}{ll}{{\rm{Reading}}\;{\rm{literacy}}}_{{ij}}\\={\beta }_{0j}+{\beta }_{1j}{\rm{Gender}}+{\beta }_{2j}{\rm{ESCS}}+{\beta }_{3j}{\rm{U}}{\rm{nderstanding}}\;{\rm{and}}\;{\rm{remembering}}\\+{\beta }_{4j}{\rm{S}}{\rm{ummarising}}+{\beta }_{5j}{\rm{A}}{\rm{ssess}}\;{\rm{credibility}}+{\beta }_{6j}{\rm{S}}{\rm{elf}}-{\rm{concept}}\;{\rm{of}}\;{\rm{reading}}\\+{\beta }_{7j}{\rm{J}}{\rm{oy}}/{\rm{L}}{\rm{ike}}\;{\rm{reading}}+{\beta }_{8j}{\rm{SEI}}+{\beta }_{9j}{\rm{Teacher}}-{\rm{directed}}\;{\rm{instruction}}\\+{\beta }_{10j}{\rm{Perceived}}\;{\rm{feedback}}+{\beta }_{11j}{\rm{Adaptation}}\;{\rm{of}}\;{\rm{instruction}}\\+{\beta }_{12j}{\rm{Teacher}}\;{\rm{support}}\;{\rm{in}}\;{\rm{test}}\;{\rm{language}}\;{\rm{lessons}}\\+{\beta }_{13j}{\rm{Duration}}\;{\rm{in}}\;{\rm{early}}\;{\rm{childhood}}\;{\rm{education}}\;{\rm{and}}\;{\rm{care}}+{r}_{{ij}}\end{array}$$
$${\beta }_{0j}={\gamma }_{00}+{\gamma }_{01}{\rm{School\; location}}+{\gamma }_{02}{\rm{School\; type}}+{\gamma }_{03}{\rm{School\; size}}+{\gamma }_{04}{\rm{Shortage\; of\; educational\; material}}+{\gamma }_{05}{\rm{Shortage\; of\; educational\; staff}}+{\gamma }_{06}{\rm{Proportion\; of\; highly\; educated\; teachers}}+{u}_{{ij}}$$
$${\beta }_{1j}={\gamma }_{10}+{u}_{1j}$$
$${\beta }_{2j}={\gamma }_{20}+{u}_{2j}$$
$${\beta }_{3j}={\gamma }_{30}+{u}_{3j}$$
$${\beta }_{5j}={\gamma }_{40}+{\gamma }_{41}{\rm{Proportion\; of\; highly\; educated\; teachers}}+{u}_{4j}$$
$${\beta }_{6j}={\gamma }_{50}+{\gamma }_{51}{\rm{School\; location}}+{u}_{5j}$$
$${\beta }_{7j}={\gamma }_{60}+{u}_{6j}$$
$${\beta }_{8j}={\gamma }_{70}+{\gamma }_{71}{\rm{School\; type}}+{u}_{7j}$$
$${\beta }_{9j}={\gamma }_{80}+{\gamma }_{81}{\rm{School\; location}}+{u}_{8j}$$
$${\beta }_{10j}={\gamma }_{90}+{u}_{9j}$$
$${\beta }_{11j}={\gamma }_{100}+{u}_{10j}$$
$${\beta }_{12j}={\gamma }_{110}+{\gamma }_{111}{\rm{Shortage\; of\; educational\; material}}+{u}_{11j}$$
$${\beta }_{13j}={\gamma }_{120}+{\gamma }_{121}{\rm{School\; size}}+{u}_{12j}$$
$${\beta }_{14j}={\gamma }_{130}+{\gamma }_{131}{\rm{Shortage\; of\; educational\; material}}+{u}_{13j}$$

In the equation, \({\beta }_{1j} \sim {\beta }_{13j}\) were student-level regression coefficients indicating the effect of student-level variables on reading literacy; \({\gamma }_{01} \sim {\gamma }_{06}\) were school-level regression coefficients representing the influence of school-level variables on reading literacy; and γ41, γ51, γ71, γ81, γ111, γ121, γ131 indicated the interaction effect of school-level and student-level variables.

Results

The null model

The results of the null model are shown in Table 3, Model 1. The grand mean of reading literacy (\({\gamma }_{00}\)) was 348.90. The random effect estimates demonstrated reading literacy variance among students (\({\delta }^{2}\) = 2593.53) and across schools (\({\tau }_{00}\) = 1762.07). The Intraclass Correlation Coefficient, i.e., ICC (1)\(=\rho ={\tau }_{00}\,/\,({\tau }_{00}\,+\,{\delta }^{2})\approx \,\)0.405. Cohen (1988) stated that when ρ > 0.138, it can be considered to show strong within-group correlation. Thus, it appears that the students’ reading literacy varies greatly among schools. ICC (1) also indicated that 40.5% of the total variance in students’ reading literacy is explained by school-level variables.

Table 3 HLM analysis of factors influencing students’ reading literacy.
Full size table

The random coefficient regression model

The results of the random coefficient regression model are seen in Table 3, Model 2. The variable gender was significant (β = −8.99, p = 0.014), but ESCS did not demonstrate a significant effect (β = 1.66, p = 0.333); understanding and remembering, summarizing, and assessing credibility involved in reading meta-cognition, understanding and remembering had a significant positive effect (β = 8.11, p = 0.001), that is, for each unit increase in understanding and remembering, the mean of reading literacy in Morocco increased by 8.11 points, whereas summarizing (β = 3.51, p = 0.226) and assess credibility (β = 0.47, p = 0.867) had a non-significant effect; the self-concept of reading positively predicted reading literacy (β = 11.34, p = 0.000); students’ expected occupational status had a significant positive effect on reading literacy (β = 0.36, p = 0.000); teacher support in test language lessons positively predicted reading literacy (β = 5.13, p = 0.044); in terms of teacher instructional strategies, adaptation of instruction positively predicted students’ reading literacy (β = 7.14, p = 0.005), indicating that the implementation of adaptive teaching is conducive to the promotion of students’ reading literacy. Duration in early childhood education and care positively predicted reading literacy (β = 3.69, p = 0.01).

After the addition of an independent variable at level 1, that is, at the student level, ES ≈ 0.369, showing that residual variance improved by 36.9%. According to the criteria that Cohen (1988) set where, for example, explanatory power is strong when ES > 0.35, the student variable therefore explains the variance strongly.

The intercept-as-outcomes regression model

The results of the intercept-as-outcomes regression model are shown in Model 3 of Table 3. School location, school type, and proportion of highly educated teachers were significant predictors of students’ reading literacy, whereas the remaining variables were non-significant. School location had a positive effect on students’ reading literacy (β = 11.89, p = 0.001), that is, the reading literacy of students whose schools were located in cities was significantly higher than that of students whose schools were located in towns and villages. School type also had a positive effect on reading literacy (β = 31.94, p = 0.031) whereby students in private schools performed better than the students in public schools. Further, the proportion of highly educated teachers positively affected students’ reading literacy (β = 33.87, p = 0.010). There were no significant effects for school size, shortage of educational material, and shortage of educational staff on reading literacy.

The full model

The results of the full model are shown in Table 3 Model 4. Among the student-level factors, understanding and remembering (β = 7.96, p = 0.002), self-concept of reading (β = 10.56, p = 0.000), students’ expected occupational status (β = 0.41, p = 0.000), adaptation of instruction (β = 6.95, p < 0.01), teacher support in test language lessons (β = 5.68, p = 0.005), and duration in early childhood education and care (β = 3.30, p = 0.018) significantly predicted students’ reading literacy. Remaining variables did not have a significant effect on students’ reading literacy.

The examination of the direct effect of school-level variables on students’ reading literacy using the student-level intercept as the dependent variable and school-level variables as the independent variables showed that variables school location (β = 13.22, p = 0.000) and proportion of highly educated teachers (β = 46.99, p = 0.001) significantly influenced reading literacy. The remaining school-level variables did not have significant effects on reading literacy.

The results of the interaction between student-level and school-level variables in the full model are shown in Table 4. We investigated possible interaction effects between school variables and individual and family variables on reading literacy. Summarizing of meta-cognition did not have a significant effect on students’ reading literacy (β = 3.80, p = 0.202); however, the interaction term of proportion of highly educated teachers significantly predicted reading literacy (β = 14.72, p = 0.038), showing that the greater school teachers’ education was, the better students’ summarizing competence was. Thus, students’ reading literacy was shown to benefit from their teacher’s greater educational training.

Table 4 Cross-level interactions of school-level and student-level variables.
Full size table

School location not only had a direct effect on students’ reading literacy (β = 13.22, p = 0.000) but also had a significant effect on students’ reading literacy as a moderator variable interacting with assess credibility of meta-cognition (β = 3.17, p = 0.040) and students’ expected occupational status (β = −0.14, p = 0.016). Assess credibility of meta-cognition and students’ expected occupational status did not have a direct effect on reading literacy in Model 4 but both variables significantly predicted students’ reading literacy under the moderator. Neither school type (β = 16.33, p = 0.243) nor joy/like reading (β = 2.80, p = 0.287) had a significant effect on reading literacy; however, the interaction of school type and joy/like reading did significantly affect students’ reading literacy (β = 11.07, p = 0.021).

Adaptation of instruction directly predicted students’ reading literacy (β = 6.95, p = 0.005) and shortage of educational material weakened the relationship between adaptation of instruction and reading literacy (β = −5.77, p = 0.000). These findings suggest that a lack of educational resources attenuates the positive effects of adaptive instructional strategies on reading literacy. Shortage of educational material was also moderated by duration in early childhood education and care to significantly predict students’ reading literacy (β = 2.58, p = 0.032), a finding which explains to some extent the complementary role of home and school in improving reading literacy. Teacher support in test language lessons positively predicted students’ reading literacy (β = 5.68, p = 0.031). School size also significantly affected reading literacy (β =− 0.01, p = 0.015), showing that the smaller the school size, the higher the perceived teacher support in the classroom was, and the more favorable the situation was for students’ reading literacy development.

Discussion

Using two-level HLM analysis and PISA 2018 data, the present study explored factors influencing students’ reading literacy in Morocco. The results revealed significant inter-school differences in Moroccan students’ reading literacy. Students’ reading literacy was affected by individuals, families, and schools, as discussed below in more detail.

Personal factors predicting reading literacy

In terms of personal factors, Moroccan students’ understanding and remembering of meta-cognition, self-concept of reading, and expected occupational status all affected their reading literacy. These findings are supported by the extant literature. Proficient use of meta-cognitive reading strategies has been shown to contribute to deeper comprehension of reading material and enhance the effectiveness of reading strategy selection, thereby not only improving students’ completion of reading tasks but also the amount and quality of knowledge and information they gain from texts (Phakiti, 2003). Reading motivation, as demonstrated in the self-concept of reading ability and in students’ educational expectations, has been shown to improve students’ reading literacy: the higher reading motivation is, the more improved is reading literacy (Alramamneh et al. 2023; Kuan, 2011).

Our gender effects varied: gender had significant effects on reading literacy in Model 2 and Model 3, but not in the full model. Different findings have also been produced in previous studies: for example, Stoet and Geary (2013) suggested that girls have higher reading scores than boys, although Phakiti (2003) suggested that it was not significant. In our study, after adding school variables, gender had no significant effect, perhaps showing that Moroccan education in reading literacy has succeeded in assisting girls and boys equitably.

Family factors predicting reading literacy

Our results suggest that Moroccan students’ reading literacy is not significantly influenced by their family’s socioeconomic and cultural status, findings that are inconsistent with those that suggest socioeconomic and cultural status is the main driver for academic achievement (Li and Li, 2022; Schiff and Lotem, 2011; Gustafsson et al. 2018). Some authors also propose that minority status and second language learning put students at a disadvantage (MacPhee et al. 2017). Morocco is a multilingual country; students speak a variety of languages at home with the majority using Darija as their local dialect whereas the main language taught in schools is Arabic. These practices result in differences in the daily oral and written language usage of Moroccan students. The family language environment plays a more crucial role in students’ reading literacy than family socioeconomic and cultural status (Alramamneh et al. 2023; Charity et al. 2004), perhaps explaining why Moroccan students’ reading literacy is less affected by family economic and cultural status than by family language use. Further studies are called for to explore these important issues.

Our findings suggest further that duration of early childhood education and care also affects students’ reading literacy. Early childhood education, especially at home, is critical for children. Brain, cognitive, and socio-emotional development occur early in life, and there is growing evidence that learning early in life has a significant impact on later educational success as well as lifelong well-being (Richter et al. 2021).

School-related factors predicting reading literacy

School-related factors that affect reading literacy include not only school factors that exert a direct influence but also the interaction between school and student. Our findings revealed inter-school differences in students’ reading literacy, with school location, proportion of highly educated teachers, and school type taking a relatively prominent position. These school factors either directly or indirectly affected students’ reading literacy. The reading literacy of students in urban schools was a great deal better than that of students in rural schools, reflecting the quality of socio-economic and cultural environments as well as a mismatch between urban and rural areas in the distribution of educational resources (Young and Dolzhenko, 2022). There may be less digital access to reading information offered to students in rural schools (Rutter and Maughan, 2002) and also less dedicated space for reading (Khoudi et al. 2024), contributing to a school climate which is non-conducive to reading literacy.

Location of schools had an unexpected impact on rural students’ career expectations. Students in rural schools were found to be more capable than those in urban schools of improving their reading literacy via their career expectations. The reason for this effect may concern how students’ self-educational expectations positively influence their willingness to put in effort, which in turn promotes their reading literacy (Kuan, 2011). The students in rural schools probably were more willing to improve reading literacy. Schools, as places of teaching and education, have been shown to profoundly affect the development of reading literacy, optimize the allocation of resources, provide a good environment for student growth and development, and aid students’ development (Young and Dolzhenko, 2022; Gustafsson et al. 2018; Zuze and Reddy, 2014).

Our results also showed that Moroccan children’s reading literacy was clearly affected by the proportion of highly educated teachers available. Highly qualified teachers make more efficient responses to children’s needs and involve parents more in school decisions. The presence of highly educated teachers largely explains high levels of reading literacy (Chatri et al. 2021). In our study, school teacher quality and teacher instructional strategies had a positive impact on reading literacy, a finding which is in line with those of previous research on teacher predictions of student reading literacy, further supporting the idea that high-quality teachers enhance student reading achievement (Yan and Cai, 2022).

Teachers are practical organizers and practitioners of school education, and highly educated teachers, as the ‘soft’ resources of schools, have an important impact on students’ cultivation of meta-cognitive strategies and the enhancement of their interest in reading. Our study showed that teachers’ support in the classroom had a direct positive effect on students’ reading literacy. Students’ perceived relationship with their teachers also had a positive effect on their reading literacy. These findings are consistent with previous studies (Hughes et al. 2008; Lee, 2012). After controlling for school type and location, we also found that the use of highly qualified teachers’ resources significantly influenced students’ reading literacy through understanding and remembering of meta-cognition.

The Moroccan educational system consists of nationalist schools providing solid bilingual education and traditional religious schooling aimed at reading and writing in Arabic. The schools have promoted different languages but have not used educational objectives as the criteria for choosing one language over another (Dahbi, 2023), a practice which may produce different effects on children’s reading literacy. Our results imply that both individual factors and environmental factors, especially school and teacher factors, should be taken into account in efforts to improve Moroccan children’s reading literacy.

Morocco, the only African country participating in PISA 2018, ranked 73th out of 79 participating countries or regions in Arabic reading scores. Moroccan children’s perceived teacher support in the PISA test language class was shown to have a greater impact on reading literacy compared with higher-scoring countries. Consistent with previous research (Zhao et al. 2022), understanding and remembering of meta-cognitive strategies had a greater effect on reading literacy compared with higher-scoring countries.

The goals of language education in Morocco remain unclear, leading to a focus on language forms rather than the use of language classrooms to teach language arts, critical thinking, confidence in self-expression, creativity in language use, and all the remaining skills and competencies that allow children to feel empowered in their own language (Dahbi, 2023). In addition, due to the poor reading literacy performance of most Moroccan children, there is little differentiation in the assessment of high-level meta-cognitive strategies.

Implications and limitations

The findings of the present study have several key implications. Firstly, they suggest that educators in Morocco need to focus on the development of children’s meta-cognitive comprehension and memory with a focus at first on the lower level of strategies (Dahbi, 2023). Morocco should focus on the development of meta-cognitive reading strategies to guide students in their choice of appropriate reading strategies, cultivation of their interest in reading, and enhancement of their motivation for reading.

Secondly, the Moroccan education administration should further optimize the allocation of educational resources in rural areas, strengthen support for rural areas and other areas that have a weak educational foundation, and provide a good material and institutional guarantee for the cultivation of reading literacy for students in disadvantaged areas. In addition, Moroccan teachers need to improve the quality of language teaching in the classroom, especially by enhancing children’s deep understanding of texts and training their ability to capture hidden textual meanings.

Thirdly, the appropriate Moroccan authorities should promote deeper integration between families and schools whereas parents should increase their investment of resources and emotional support in family education (Alramamneh et al. 2023). At the same time, parents should be more proactive in communicating with school teachers and pay more attention to children’s development at schools. School teachers should help children develop their own educational aspirations, form positive career expectations, cultivate their understanding of key concepts, and work with their parents to develop healthy competitiveness for the future.

Three limitations were identified in this study. Firstly, although the present study, focusing on three dimensions of student, family, and school, analyzed factors influencing reading literacy at school- and student-levels, future research could expand the level of data nesting (e.g., by including district) in order to explore factors that may influence students’ reading literacy at a higher level than the school-level. Secondly, the present study analyzed the influencing factors of reading literacy through HLM. Future studies could usefully examine the influence of relevant factors on students’ reading literacy through structural equation modeling (SEM). Thirdly, PISA 2018 used cross-sectional data and therefore did not assess changes in student reading literacy over time. Such research, using some kind of longitudinal design, would provide highly important data.

Conclusion

The present study examined the impact of both student- and school-level factors on the reading literacy of students who spoke Arabic at school but not at home. The results indicate that there are significant differences in reading literacy among Moroccan students across schools, and these differences are influenced by individual, family, and school factors. Critically, Moroccan students’ understanding and remembering of meta-cognitive strategies have a greater effect on reading literacy than students in the higher-scoring countries. In addition, the duration of students’ early education and the number of school educational materials together were shown to promote the development of students’ reading literacy. The results also showed that teachers’ educational qualifications not only had a direct impact on students’ reading literacy but had a further impact on students’ reading literacy also by moderating students’ meta-cognition.