You are on page 1of 20

Research in Science & Technological Education

ISSN: 0263-5143 (Print) 1470-1138 (Online) Journal homepage: http://www.tandfonline.com/loi/crst20

Modeling the influences of upper-elementary


school students’ digital reading literacy,
socioeconomic factors, and self-regulated learning
strategies

Shin-Feng Chen

To cite this article: Shin-Feng Chen (2017): Modeling the influences of upper-elementary school
students’ digital reading literacy, socioeconomic factors, and self-regulated learning strategies,
Research in Science & Technological Education, DOI: 10.1080/02635143.2017.1314958

To link to this article: http://dx.doi.org/10.1080/02635143.2017.1314958

Published online: 26 Apr 2017.

Submit your article to this journal

Article views: 18

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=crst20

Download by: [Oklahoma State University] Date: 28 May 2017, At: 16:58
Research in Science & Technological Education, 2017
https://doi.org/10.1080/02635143.2017.1314958

Modeling the influences of upper-elementary school


students’ digital reading literacy, socioeconomic factors, and
self-regulated learning strategies
Shin-Feng Chen
Department of Education, National Pingtung University, Pingtung, Taiwan, ROC

ABSTRACT KEYWORDS
Background: Reading is an interactive and constructive process of Digital reading literacy; self-
making meaning by engaging a variety of materials and sources and regulated learning; structural
by participating in reading communities at school or in daily life. equation modeling
Aim: The purpose of this study was to explore the factors affecting
digital reading literacy among upper-elementary school students.
Method: A 3-stage stratified cluster sampling was implemented that
resulted in a sample of 592 upper-elementary students from 29 classes
in 7 schools. Self-Regulated Learning Strategies Assessment (S-RLSA),
Digital Reading Literacy Assessment (DRLA), and student reports of
their parents’ education backgrounds were used to collect data on the
outcome and predictor variables. Interpretation of these data involved
two highly regarded statistical techniques. First, structural equation
modeling was used to explore relationships amongst the constructs.
Second, multi-group invariance (MI) analyses were used to assess the
influence of parental education and self-regulated learning strategies
on students’ digital reading literacy.
Results: Enriching students’ family learning resources and strengthening
their self-regulated learning abilities could have very important influences
on promoting upper-elementary school students' digital reading literacy
-webpage information retrieval, reading and communication abilities.
Conclusions: This study also provides information on how teachers
can address student resources to improve digital reading literacy and
self-regulated strategies.

Introduction
Reading is a meaning-making process in which readers access information from the materials
read, their prior knowledge, and the sociocultural context of the reading task to construct
an integrated interpretation of these inputs. Contemporary reading materials include print
and digital texts, images, tables, diagrams, labels, and animations. Prior knowledge includes
accurate and inaccurate ideas about the target contents and discipline and the discourse
conventions of the reading material. The sociocultural context might include classrooms,
informal environments, family values, cultural beliefs, and individual personal factors.
Therefore, reading is an interactive and constructive process of making meaning by engaging

CONTACT  Shin-Feng Chen  chensf@mail.nptu.edu.tw


© 2017 Informa UK Limited, trading as Taylor & Francis Group
2   S.-F. CHEN

a variety of materials and sources and by participating in reading communities at school or


in daily life that increases the range of cognitive and metacognitive demands and the neces-
sity of reading. Readers placed in this textual environment need to be strategic, critical, and
self-regulated. The purpose of this study was to investigate the factors relating to family
attributes, regulated-learning strategies, and digital reading literacy among upper-elemen-
tary school students in order to understand the key capabilities and relationships affecting
digital reading literacy.

Background
Recently, digital or electronic learning (i.e. e-learning) of science has increased with the
application of the Internet and information communication technologies (ICT) in schools.
The underlying assumption about e-learning is that it allows teachers and students freedom
to engage in teaching/learning whenever and wherever they are; schooling becomes a
construct decoupled from the brick and mortar construction called schools. Understanding
the influences and effects on science reading literacy in such environments is only the start
to exploring the needs of the Net generation – students who have been born into the ICT
environment. The Program for International Student Assessment (PISA) assessed 15-year-old
students’ reading literacy as forming a broad understanding, retrieving information, devel-
oping an interpretation, reflecting on and evaluating the content of text, and reflecting on
and evaluating the form of text (OECD 2006). This study focused on five factors believed to
affect digital reading literacy of science materials: metacognition and self-regulated learning,
reading science language and materials, digital reading literacy and self-regulated learning
strategy, existing reading awareness measures, and sociocultural and socioeconomic
factors.

Metacognition and self-regulated learning


Metacognition involves thinking about your learning as you are learning to improve your
learning; it involves two distinct clusters: self-awareness and self-management (Ford and
Yore 2012). The importance of metacognition becomes apparent when applied to an inter-
active-constructive interpretation of reading involving meaning making from multiple infor-
mation sources (Hsu et al. 2016). Metacognitive awareness in science reading involves
declarative, procedural, and conditional knowledge about science reading; self-management
of reading involves strategic planning, monitoring, and regulating the science reading act
in real time (Wang, and Chen, 2014; Wang, Chen, Fang, and Chen, 2013).
ICT-infused instruction, e-learning, and Internet environments have opened up learning
beyond the teacher-structured classroom walls to wherever learners are located with access
to the Internet – McDonalds, libraries, museums, and other informal environments – and to
a much wider variety and validity of information sources. These environments and sources
require that students possess self-regulated learning strategies to meet learning demands
and to achieve positive effects (Schraw, Crippen, and Hartley 2006). Expert self-regulated
learners develop learning strategies that allow them to critique the quality of information,
test their understanding, make improvement to areas of misunderstanding, and adjust effort
and action according to their monitoring and evaluation progress based on personal obser-
vations and exterior feedback.
RESEARCH IN SCIENCE & TECHNOLOGICAL EDUCATION   3

Self-regulated learning (SRL), an inclusive framework based on information processing,


involves aspects of metacognition, motivation, task conditions, goal setting, volition control,
behavior adaptation, and emotion reaction. Learners evaluate the learning contents then
adopt strategies, exercise personal control, then monitor and reflect on their own learning
behavior to solve the problem and achieve their goal (Ackerman and Goldsmith 2011;
Bannert 2004; Efklides and Vlachopoulos 2012; Mokhtari and Reichard 2002). Digital reading
literacy includes links, searching diverse information stores, and the combination of multi-
media texts, which require more self-regulation strategies than highly controlled
textbooks.

Science language and reading science materials


Language has important communicative, constructive, and argumentative functions in sci-
ence; especially in written form, it is essential in doing and learning science (Carlsen 2007;
Yore and Tippett 2014). Science language and text are composed of a mixture of words,
visuals, representations, signs, symbols, and notations organized into task-specific genre
(i.e. form/function) designed to describe, explain, and connect ideas formed as networks of
propositions and the scientific enterprise (Carnap 2002). Chen and Yang (2006) indicated
that science texts are like movies and the actors are the technical and metalinguistic terms
in the text. The contextual development and disciplinary attributes of the plot become
relatively vital if the audience is to understand the movie’s content. Technical terms in science
texts usually have a unique and specific definition or connected idea. Science metalanguage
denotes and describes the unique nature of the enterprise and the epistemic and ontological
features of science. The genres of science text (e.g. description, argument, cause-effect,
directions, etc.) are not common in most popular narrative texts and elementary school
reading programs.
Many students will not be capable of constructing understanding of science text without
(a) an explanation of technical terms, (b) integrating the messages from the verbal and visual
tracks in the text, (c) an understanding of conceptual networks of natural phenomenon and
principles behind the technical terms, and (d) a familiarity with the dominant genres
(Schellings, Aarnoutse, and Leeuwe 2006; Tippett, and Anthony 2011). Several models of
science reading comprehension have built on the dual-coding model in which readers make
parallel interpretations of the verbal message in the printed words, signs, and symbols with
the genre used and of the visual message in the pictures, diagrams, graphs, and representa-
tions. The critical feature is the synergy in the cross-links between these two messages that
enhance and elaborate each other and anchor to experiences (Tippett, and Anthony, 2011).
Therefore, teaching to enhance science conceptual achievement and reading comprehen-
sion must help students construct initial understanding of the central concepts and explain,
give examples, and apply the terminology and concept in a variety of different situations
(Samuelstuen and Braten 2005).

Digital reading literacy and learning-regulation strategies


Reading literacy in the digital era is not simply obtaining knowledge from static written and
visual information where the dual-coding model anticipates the diversity of dynamic signs,
symbols, graphics, and sounds in digital information sources and the influences of prior
4   S.-F. CHEN

knowledge and sociocultural context. Digital reading literacy (DRL) emphasizes learning in
the ever-changing Internet and ICT (Coiro and Dobler 2007). The Progress in International
Reading Literacy Study (PIRLS) aimed to document the process of reading comprehension,
the goals of reading, and reading behaviors and attitudes involving information retrieval,
information inference, interpretation and integration, and comparison and evaluation (Mullis
et al. 2009); while PISA’s (OECD 2006) interpretation of reading literacy emphasized infor-
mational text and consisted of retrieving information, forming a broad understanding, devel-
oping an interpretation, and reflecting on and evaluating the contents and form of text. This
study adopted the PISA definition as the operational definition of digital science reading
literacy. The influence of previous knowledge and sociocultural context means that readers
can form different broad understandings of the text while identifying the main idea and
explaining the text’s purpose. The processes of searching for and retrieving information from
digital texts involve socially and culturally derived inferences about a range of text and sit-
uational cues. While constructing meaning of the digital text, skilled readers use various
processes, critical thinking, and strategies to foster, monitor, evaluate, and maintain under-
standing (OECD 2013), which eventually leads to processing multiple information sources
in a logical manner and showing understanding of what is read.
Readers are capable of selecting relevant information from digital texts and making sense
of these information sources even if they contain discrepant ideas. Obstructions to DRL
include readers’ cognitive load inherent in the text/task conditions and inefficient reading
caused by excessively free and unbounded links within the text (Leu 2007), thus resulting
in interpretation difficulties (e.g. confirming the vital problem, searching for information,
analyzing information, integrating information, and communicating information). Schmar-
Dobler (2003) indicated that the demand for strategies are different for confronting problems,
monitoring and repairing comprehension, and browsing in reading of print text and digital
text. The major difference between digital and traditional texts is that the digital text is
dynamic with blurred boundaries; it also requires using browsing tools to confront significant
problems, search the Internet for information, and integrate information from continuous,
non-continuous, mixed, and multiple texts. Mixed texts include continuous and non-­
continuous texts while multiple texts require cross-page reading (intertextuality) and include
simple continuous or non-continuous and mixed texts.
DRL combines reading strategies of books and understanding the features of the Internet,
ICT, or digital reading environment. Coiro and Dobler (2007) proposed that the factors affect-
ing DRL are prior knowledge source, inferential reasoning strategies, and self-regulated
reading processes. Besides drawing upon their prior knowledge of topics and text structures,
digital readers must draw upon their prior knowledge of Internet texts, website structures,
and web-based search engines. The inferential reasoning strategies of DRL are characterized
by a frequent use of forward inferential reasoning and processes including 3-dimensional
Internet spaces (i.e. drilling into multiple information sources revealed during an Internet
search) where literal matching skills, structural cues, and context clues are also required.
The massive size and diversity of digital information resources require that students be
able to locate, select, and retrieve the important messages consistent with their purpose.
By cultivating this searching ability, students can read more efficiently and effectively in the
environment of digital reading and transform the read information into their own knowledge.
Therefore, DRL in this study involves searching, identifying, selecting, and integrating the
digital reading information.
RESEARCH IN SCIENCE & TECHNOLOGICAL EDUCATION   5

Existing reading awareness measures


Several scholars have proposed different views about the assessment of metacognitive SRL
applicable in the field of digital learning. The Index of Reading Awareness (IRA; Jacobs and
Paris 1987) assessed metacognitive awareness and self-management, the Motivated
Strategies for Learning Questionnaire (MSLQ; Pintrich and de Groot 1990) assessed strategies
and self-regulation, and the Metacognitive Awareness Inventory (MAI; Schraw and Dennison
1994) assessed self-regulation strategies. Assessment methods for metacognitive awareness
have involved inventories, tests, and interviews; assessment of metacognitive SRL have
involved thinking-aloud exercises, interviews, checklists, teacher assessments, and self-report
inventories (Sperling et al. 2002). The SRL strategies defined by this study addressed planning,
information management, monitoring, adjustment effort/action, and evaluation and, addi-
tionally, developed an assessment to document these five strategies.

Sociocultural and socioeconomic factors


Learning is contextually bound by societal, cultural, and economic features of the family,
community, school, classroom, and peer groups. Taiwan’s sociocultural and sociopolitical
context involves a hybrid form of spiritualism and political philosophy assumed from an
integration of the Taoism and Buddhism religions and the Confucianism philosophy of life,
family, and learning (Huang and Yore 2003). The values and priorities of the culture and
family influence school priorities and practices. Asian society and culture provided the soci-
ocultural context for this study where schooling and achievement are highly valued and
expected.
Family context and conditions are vital factors affecting students’ learning achievements.
It was reported that parents’ metacognitive knowledge influenced their interactions with
their children (Thomas and Anderson 2013). Lee and Yu (2005) proposed that parental soci-
oeconomic status, education level, and occupation directly influenced the education
achievements of their children, which was also indirectly influenced by the number of siblings
and family education resources. Scott (2004) found that family socioeconomic status, back-
grounds, employment status of the mother, ways of raising children, and the inclination of
gender role have great influence on students’ learning achievement. Secondary analysis of
TIMSS 2003 and PISA 2006 data-sets revealed that family socioeconomic status and parental
educational level were associated with students’ science achievement (Lin 2009). Chen (2013)
indicated that parental education level has a significant influence on the development of
elementary school students’ science language and, therefore, was a good proxy for socio-
economic status.

Research questions
This study investigated how parental education level and self-regulated learning strategies
influence digital reading literacy. The research questions are:

1.  How does parental education level influence the self-regulated learning strategies
of upper-elementary school students?
2.  How do parental education level and students’ self-regulated strategies affect the
digital reading literacy of upper-elementary school students?
6   S.-F. CHEN

Methodology
This study involved two parts. The first part established a conceptual model composed of
four specific factors: parental education level, self-regulated learning strategies, digital read-
ing communication, and webpage information retrieval. These four factors and their defini-
tions were chosen based on established literature. The second part tested the fitness of this
conceptual model using collected empirical data from Grades 5 and 6 students in 241 ele-
mentary schools in a southern metropolitan city in Taiwan.

Model of digital reading


This study investigated the relevant structural equation modeling (SEM) affecting upper-
elementary school students’ DRL and proposed a cause–effect model among its affecting
factors. SEM is a statistical method for managing and verifying theoretical cause–effect
relationships against empirical data. The first step involves model definition deduced from
a specific theory and concept or induced from experience and reviewed literatures (Chen,
2014; Kline 2011).
The preliminary model structure (Figure 1) contained four independent variables (i.e.
mother’s education level, father’s education level, SRL strategies in general situations, and
SRL strategies in test situations) and four dependent variables (i.e. basic digital reading
communication, advanced digital reading communication, basic webpage information
retrieval, and advanced webpage information retrieval). The circles represent error variables
(i.e. unspecified/undefined) influences on observable variables. The hypothesized model
was then tested against empirical data. Mplus Version 7.0 was applied to do confirmatory
factor analysis in which the main latent independent variables were parental education level
and SRL strategies and the latent dependent variable was DRL.

Research samples
Established research ethics for informing and recruiting participants were applied to each
research sample before data collection. The acceptance letters were signed by the students,
parents, teachers, and their schools. A stratified sampling was used to identify a verification
sample from amongst the elementary schools in southern Taiwan utilizing informed consent
of the school officials. A 3-layer stratified, cluster-sampling design was implemented to
achieve a representative sample of sufficient size to reflect the complexity of the anticipated
model, number of variables and potential relationships (Hair et al. 2014). The samples for
the first layer were extracted depending on the size of schools in Kaohsiung City. The schools
were categorized into three groups according to number of classes: <12 = small-scale,
13–30 = medium-scale, and > 31 = large-scale. The second layer is the specific school, and
the third layer is the specific class. All the students from the chosen classes were regarded
as research samples.
Of the 241 public elementary schools in the Greater Kaohsiung City School District, 84
were small scale, 84 were medium scale, and 73 were large scale. The total Grade 5 population
was 27,115 students in 1,058 classes, and the total Grade 6 population was 31,043 students
in 1,138 classes. Grade 5 averaged 25.6 students, and Grade 6 averaged 27.3 students, with
an overall average of 26.5 students in each grade-level class. The student population
RESEARCH IN SCIENCE & TECHNOLOGICAL EDUCATION   7

E3 E4

1 1

General Testing
situations situations
1 Father 1
E1 Education Basic Reading E7
Level 1
1 1

Self-Regulated
Parent Digital Reading
Learning
Education Level Communication
Strategies 1

1 Mother 1 D3 Advanced 1
E8
E2 Education D1 Reading
Level

Webpage
1 Information
Retrieval
D2
1

Basic Advanced
Retrieval Retrieval
1 1

E5 E6

Figure 1. Preliminary model structure for the influence of elementary school students’ parental education
level and self-regulated learning strategies on digital reading literacy.

distribution indicates a ratio of 1:3:6 for small-, medium-, and large-scale schools. The sample
consisted of 592 students from 29 classes of 7 public schools in the school district. There
were 319 boys (54%) and 273 girls (46%), and 260 (44%) were in Grade 5 and 332 (56%) were
in Grade 6.

Data collection
This study required data on participants’ reading literacy, self-regulation strategies, and par-
ents’ educational backgrounds. Therefore, an online DRL assessment and a SRL strategies
assessment system for upper-grade elementary school students were developed. Data on
parents’ educational backgrounds were collected from the students in accordance with the
research ethics approval.

Digital reading literacy assessment


The digital reading literacy assessment (DRLA) for upper-grade elementary school students
was developed and established within an Internet environment. The DRLA system provided
respondents with an online assessment context and response alternatives in a real-time
environment to generate a response database on the back-end server immediately. MySQL
was adopted as the database in the system structure while PHP, HTML, and Flash CS5.5 were
adopted as the access interface.
The contents of the DRLA were designed by consulting the assessment structure of PISA
and a published instrument (Henry 2006). Three aspects of PISA assessment (i.e. information
retrieval, interpretation integration, reflection, and evaluation) and Henry’s locating
8   S.-F. CHEN

information on the Internet were the central components in the DRLA for upper-elementary
school students. The first part of the assessment consisted of five webpage information
retrieval, multiple-choice questions: two information retrieval and three interpretation inte-
gration questions. The second part consisted of 13 digital reading, communication reaction
questions: four information retrieval, five interpretation integration, and four reflection and
evaluation questions. This part had two multiple-choice questions with six options (where
more than one option is correct) and 11 multiple-choice questions with four options (only
one correct option). Therefore, the total possible score was 23 points (part 1 was worth 10
points – five 2-point items, and part 2 was worth 13 points – 1-point items). DRLA items were
divided into two difficulty levels: basic and advanced. The basic questions included three
information retrieval, four interpretation integration, and one reflection and evaluation ques-
tions; the advanced questions included three information retrieval, four interpretation inte-
gration, and three reflection and evaluation questions.
Three advanced questions required students to make inferences from the information
provided by single texts; the other three advanced questions required interpretation based
on tables and figures, among which two questions required transtext information integration
(i.e. using information from two or more sources) and one question required actively locating
texts. Three of the basic questions involved retrieving information from single texts; the
other three basic questions involved integrating information from designated texts.
The internal consistency of the DRLA revealed an α = 0.70; the test–retest reliability of a
3-week interval was 0.76. DeVellis (2011) suggested that an α ≥ 0.70 means high reliability.
Therefore, the DRLA’s reliability was judged to be acceptable.
Content validity of the DRLA was judged by five content experts, and their suggestions
were used to modify some items. Construct validity was explored using correlations of the
18 individual questions and total DRLA scores. Results indicated that each item was signifi-
cantly (p ≤ 0.01) and positively associated with the total score with coefficients between
0.14 and 0.56. Discrimination of the DRLA (critical ratio, CR) was explored by dividing the
respondents into a low-score group (<27%) and a high-score group (>73%) in accordance
with the total score ranking and conducting t-tests on the differences between low- and
high-achieving groups for each of the 18 questions. The results revealed significant (p ≤ 0.01)
differences on each question between low- and high-achieving students.

Self-regulated learning strategy assessment


The students’ self-regulated learning strategies assessment (S-RLSA) was constructed using
the same real-time Internet environment and design features as the DRLA. The Internet
deployment allows this assessment system to be conveniently utilized anytime and any-
where. The S-RLSA was based on the MAI (Schraw and Dennison 1994) and the MARSI
(Mokhtari and Reichard 2002). The pool of questions from these established measures
included 52 questions about SRL strategies and 30 questions about general reading strate-
gies, problem-solving strategies, and supportive reading strategies. The S-RLSA selected
questions from the 82-question pool that focused on a series of control strategies to help
learning: planning, information management, monitoring, adjustment, and evaluation strat-
egies. The final 14-question inventory was divided into two situations based on the related
inventories and theories: DRL in an assessment situation (7 questions: 1 planning strategy,
2 information management, 2 monitoring, 1 regulation strategy, and 2 evaluation strategy)
and DRL in a general situation (7 questions: 1 planning strategy, 2 information management,
RESEARCH IN SCIENCE & TECHNOLOGICAL EDUCATION   9

Table 1. Polytomous scoring model for the S-RLSA questions.


Assessed domain
Interpretation Reflection and
Observable variable # of items Information retrieval integration evaluation
Basic retrieval 3 1 2 0
Advanced retrieval 2 1 1 0
Basic reading 5 2 2 1
Advanced reading 8 2 3 3
Total 18 6 8 4

1 monitoring, 2 regulation strategy, and 1 evaluation strategy). The S-RLSA had two planning
strategy, four information management, three monitoring, three regulation strategy, and
three evaluation strategy questions. A polytomous scoring model was adopted as the grad-
ing standard. This grading mechanism assigns up to 3 points to each question (total possible
score = 42 points), depending on the quality of the response (Table 1).
Psychometrics for the S-RLSA were explored using the same techniques as for the DRLA.
The internal consistency (reliability) for the S-RLSA for upper-grade elementary school stu-
dents revealed an α = 0.62. The test–retest reliability of a 3-week interval is 0.59. DeVellis
(2011) suggested that 0.35 < α < 0.70 means medium reliability. Therefore, the reliability of
the digital S-RLSA is within the acceptable range for low-risk purposes.
Content validity of the S-RLSA was judged by five content experts, and their suggestions
were used to modify some items. Construct validity, correlations of the 14 individual ques-
tions, and total S-RLSA scores were between 0.29 and 0.57 and significant (p ≤ 0.01). All CR
values of the questions for low- and high-scoring groups of students were significant
(p ≤ 0.01), which indicated that all questions discriminated.

Parental education level


Socioeconomic status is mainly defined by educational level, occupational prestige, and
income (Yang and Gustafsson 2004). The measure of socioeconomic status in this study
involved the educational level of parents. Participants reported their parents’ education level
regarding the highest degree or diploma, which is a common practice in Taiwan. A question
with nine options was scored as follows: 1 = dropped out of elementary school or never
went to school, 2 = graduated from elementary school, 3 = graduated from junior high
school, 4 = graduated from senior high school or vocational school, 5 = graduated from
5-year college, 6 = graduated from 2-year technical school, 7 = graduated from college,
8 = holds a master’s degree or above, and 0 = I don’t know. The coding mechanism assigned
higher values to indicate higher parental education. The ‘I don’t know’ option was coded as
missing data and not included in the related analyses.

Data analysis
The Statistical Program for the Social Science, version 20 was used to compute descriptive
statistics such as mean, standard deviation, skewness, and kurtosis. These statistics were
calculated and used to explore whether the data fit the basic assumption of normal distri-
bution. In order to investigate how upper-grade elementary students’ parental education
level influenced their capability of SRL and DRL, SEM was adopted as the main analytical
10   S.-F. CHEN

method. MI analyses of SEM were also employed in order to analyze the gender dependency
of the influence SRL strategies had on DRL. MI analyses were used to assess the influence of
parental education and self-regulated learning strategies on students’ digital reading literacy.
MI is a statistical technique that assures that comparisons made between groups represent
authentic differences in the psychological constructs being assessed. Mplus 7.0 was used
to test the theoretical model.

Results
Research results from the DRLA and S-RLSA included descriptive statistics of measurement
variables, model test results, and the model fit under multi-sample factors invariance analysis.
The means, standard deviations, and distribution features are summarized in Table 2. The
average education backgrounds for fathers and mothers, represented by the parental edu-
cation level variable, were similar with a mean of between graduation from junior high
school, senior high school, or vocational school. The mean of the father’s education level
(M = 3.18) was very close to the mean of the mother’s education level (M = 3.14). The skew-
ness data for the eight observed variables indicated a range of −1.50 (minimum) to 0.24
(maximum), and the kurtosis data ranged from −1.42 (minimum) to 1.54 (maximum). These
data support the assumption of normality and justify the use of maximum likelihood esti-
mation to test the model (Kline 2011).
The covariance matrix of the eight observed variables in the preliminary model is sum-
marized in Table 3. Reporting covariance matrix results are important when reporting SEM
results (Kline 2011). Based on the covariance matrix results, Mplus 7.0 software was used to
estimate the parameters of the structural equation model and to test the goodness of fit for
the model.
Table 4 summarizes the primitive parameters of the hypothesized model inferred from
the collected data by Mplus 7.0 estimations. These results suggest that the factor loadings
are between 0.36 and 0.90. Error variances are all positive numbers; thus, a level of signifi-
cance was achieved. Therefore, it is justified to carry out the parameter estimating for testing
the model (Kline 2011).
The overall fit of the model was very good. Therefore, to evaluate the fit of the model in
greater depth, Kline (2011) recommended that several fit indices be used. First, the Chi-square
statistic was employed as a fit index that addresses how well the variances and covariances
implied by the model match the observed variances and covariances in the data. The Chi-
square, (df = 16) = 17.22, p = 0.37, was not significant, which indicated that the model is a
good representation of the underlying covariance matrix. Furthermore, the Chi-square/
degree of freedom ratio was 1.07, suggesting a good fit according to Kline’s (2011) rule that
says a ratio of ≤ 3.0 demonstrates a good fit.
The root mean square error of approximation (RMSEA) assesses a lack of fit of the data to
the estimated model as an index that includes adjustments for model complexity such that
evaluation of fit is not overly influenced by the number of parameters in the model. Kline
(2011) suggested that a RMSEA value of ≤ 0.05 designates a good model fit. The RMSEA for
this model was 0.01.
The standardized root mean square residual (SRMR) is an index based on the residuals
between the observed and estimated covariance matrices, which are sensitive to inaccurate
RESEARCH IN SCIENCE & TECHNOLOGICAL EDUCATION   11

Table 2. Summary of descriptive statistics and data characteristics for the questionnaire.
Variable and content Min Max M SD Skew Kurt
Parental education level
 A1–Father’s education 1 6 3.18 1.75 −0.09 −1.42
 A2–Mother’s education 1 6 3.14 1.68 −0.15 −1.41
Self-regulated learning
  B1–General circumstance 8 21 15.91 2.67 −0.51 −0.13
  B2–Testing circumstance 7 21 15.87 2.48 −0.15 −0.27
Webpage information retrieval
 C1–Basic 0 1 0.85 0.24 −1.50 1.54
 C2–Advanced 0 1 0.51 0.33 −0.14 −0.69
Digital reading communication
 D1–Basic 0 1 0.53 0.27 0.04 −0.81
 D2–Advanced 0 1 0.39 0.23 0.24 −0.63

Table 3.  Covariance matrix indices for parental education level, self-regulated learning, and digital
science reading.
Observable variable A1 A2 B1 B2 C1 C2 D1 D2
A1–Father’s education 3.06
A2–Mother’s education 2.24 2.83
B1–Self-regulated general 0.46 0.57 7.14
B2–Self-regulated testing 0.81 0.74 3.13 6.15
C1–Basic retrieval 0.03 0.02 0.11 0.13 0.06
C2–Advanced retrieval 0.06 0.04 0.07 0.16 0.01 0.11
D1–Basic reading 0.08 0.05 0.17 0.27 0.02 0.02 0.07
D2–Advanced reading 0.06 0.05 0.19 0.25 0.02 0.01 0.03 0.06

Table 4. Primitive parameter estimates: loading factors, standard error (SE), and p-values.
Observable variable Factor loading SE p-value
A1–Father’s education 0.90 0.07 <0.001
A2–Mother’s education 0.85 0.06 <0.001
B1–Self-regulated general 0.56 0.04 <0.001
B2–Self-regulated testing 0.84 0.04 <0.001
C1–Basic retrieval 0.50 0.07 <0.001
C2–Advanced retrieval 0.36 0.06 <0.001
D1–Basic reading 0.70 0.03 <0.001
D2–Advanced reading 0.74 0.03 <0.001

specifications of the model. Worthington and Whittaker (2006) suggested that a SRMR value
of ≤ 0.08 means a good model fit. The SRMR for this model was 0.02.
The Bentler comparative fit index (CFI) was used to compare the model to a baseline
model in which the covariances among all the variables are assumed to be zero. The CFI
ranges from 0 to 1, with larger values representing a better fit. A value of ≥ 0.95 is considered
to be a good fit (Kline 2011). The CFI value for this model was 0.99.
Finally, the adjusted goodness of fit index (AGFI) is a measure of the proportion of the
observed covariance that is accounted for by the model, with values ranging from 0 to 1. An
AGFI value of ≥ 0.90 is considered to be a good fit (Kline 2011). The AGFI value for this model
was 0.98.
12   S.-F. CHEN

Table 5. Decomposition of effects in the hypothesized model.


Effect
Direct Indirect Total
Predictor Criterion PC t PC t PC
Parent’s education level Self-regulated learning 0.26 5.27 0.26
Webpage information retrieval 0.14 3.87 0.14
Digital reading communication 0.18 4.79 0.18
Self-regulated learning Webpage information retrieval 0.55 6.08 0.55
Digital reading communication 0.36 2.78 0.34 2.89 0.70
Webpage information retrieval Digital reading communication 0.62 4.32 0.62
Note: PC = path coefficients; t = t-test value.

Therefore, the hypothesized model (Figure 1) was supported by the empirical data.
Parental education level, webpage information retrieval capability, and SRL have quite a
large influence on DRL.
Path analysis was used to estimate the direct and indirect effects in the empirical model,
control for the correlations among the hypothesized variables, and decompose the observed
correlations into their component parts. A direct effect represents the direct influence of
one variable on another variable. An indirect effect, on the other hand, represents one var-
iable indirectly influencing another variable by way of a third variable. The path coefficients
(PC) and their associated t-values for the model are reported in Table 5. A cutoff t-value of
1.96 for a 2-tailed test was used to determine if direct and indirect paths were significant
(p ≤ 0.05). PC ranging from 0.05 to 0.10 are considered small but meaningful influences, PC
ranging from 0.11 to 0.25 are moderate in size and influence, and PC greater than 0.25 are
large in size and influence (Kline 2011). The criterion R2 (proportion of variance explained)
by SRL was 0.07, by webpage information retrieval was 0.31, and by digital reading commu-
nication was 0.75.
These results indicated four direct relationships among parental education background,
self-regulation ability, webpage information retrieval, and digital reading communication
as indicated in the hypothesized model. Parental education level had a significant and large
influence on SRL (PC = 0.26). SRL had a significant and large influence on both webpage
information retrieval (PC = 0.55) and digital reading communication (PC = 0.36). Finally, web-
page information retrieval had a significant and large influence on digital reading commu-
nication (PC = 0.62).
Table 5 also illustrates that three indirect paths were statistically significant (p ≤ 0.05).
Parental education level had a significant moderate indirect influence on both webpage
information retrieval (PC = 0.14) and digital reading communication (PC = 0.18) through its
intermediate influence on SRL, indicating its positive influence on digital reading commu-
nication. SRL had a significant large influence on digital reading communication through its
intermediate influence on webpage information retrieval (PC = 0.34). It was thus demon-
strated that students’ SRL had a positive influence on digital reading communication.
Collectively, these results are incorporated into a network of direct and indirect weighed
relationships. This empirical model (Figure 2) illustrates the four direct paths between pre-
dictor variables and dependent variables and the three indirect paths amongst triads of
predictor variables, intermediate variables, and dependent variables.
RESEARCH IN SCIENCE & TECHNOLOGICAL EDUCATION   13

Table 6. The means for predictor and outcome variables of different genders.
Mean
Variable Content Boy Girl
ξ1–Parental education level A1–Father’s education level 3.10 3.25
A2–Mother’s education level 3.02 3.26
ξ2–Self-regulated learning B1–General circumstance 15.67 16.19
B2–Testing circumstance 15.69 16.07
ξ3–Webpage information retrieval C1–Basic retrieval 0.81 0.88
C2–Advanced retrieval 0.49 0.51
η1–Digital reading communication D1–Basic reading 0.50 0.56
D2–Advanced reading 0.36 0.42

E3 E4

.69 .29

General Testing
situations situations
.19 Father .51
E1 Education Basic Reading E7
Level .56 .84
.90 .70

.26 Self-Regulated .36


Parent Digital Reading
Learning
Education Level Communication
Strategies .25

.74
.29 Mother .85 .94 D3 Advanced .46
E8
E2 Education D1 .55 Reading
Level
.62

Webpage
.69 Information
Retrieval
D2
.50 .36

Basic Advanced
Retrieval Retrieval
.75 .87

E5 E6

2
(16)=17.222,p=.371,RMSEA=.011,CFI=.999,SRMR=.019

Figure 2. The empirical model of digital reading communication, parental education level, webpage
information retrieval capability, and self-regulated learning.

This model illustrates the normalized effect size for the empirical data on upper-elemen-
tary school students’ parental education level and family learning resource affecting each
latent variable of science language, that parental education background has a direct positive
influence on SRL (factor loading = 0.26). These Grades 5 and 6 students have better SRL
capability in the webpage information retrieval capability category (factor loading = 0.55)
than for the digital reading communication capability category (factor loading = 0.36). Also,
SRL capability has a direct positive influence on both the webpage information retrieval
capability and digital reading communication capability while webpage information retrieval
capability has a positive influence on digital reading communication (factor loading = 0.62).
In conclusion, these upper-elementary school students’ parental education level, SRL
14   S.-F. CHEN

.22(.29) Self-Regulated .49(.14)


Parental Digital Reading
Learning
Education Level Communication
Strategies .20(.03)

.95(.91) D3
D1 .52(.67)

.53(.71)

Webpage
.73(.55) Information
Retrieval
D2

2
(36)=30.029,p=.748,RMSEA<.001,CFI=1.000,TLI=1.008,SRMR=.029
AIC=9522.962, BIC=9680.646, Adjusted BIC=9566.358

Figure 3.  Identical multi-sample factor structural equation model for males and females (female
weightings are in brackets).

capability, and webpage information retrieval capability have quite a positive influence on
digital reading communication.
Therefore, it is worth paying attention to the consideration of a multi-sample confirmatory
factor analysis as its measure of invariance on items within the same construct across differ-
ent groups to demonstrate whether the members of these groups have the same under-
standing of the items measured and whether predictor variables between different groups
were equal to each other in the same structure under multi-sample SEM analysis (Chen,
2014; Kline 2011). To perform identical multi-sample analysis, this research set the loading
between different genders to equal each other using an identical-factor metric model. Table 6
provides the descriptive statistics for the established parameter groups.
The standardized parameter estimate of identical multi-sample factor model is shown in
Figure 3. The total data-set had a good fit to the model (χ2 = 17.22, df = 16, p = 0.37). Analyses
of the data-sets from splitting the original data-set into two sets according to gender also
demonstrated good model fit to the data (boys’ χ2 = 8.81, df = 16, p = 0.92; girls’ χ2 = 19.01,
df = 16, p = 0.27). Critical to this study was assessing the differences in factor loading indices
between boys and girls. The four differences between these two groups measured in this
study indicated the extent one latent factor had on another latent factor. The effect of paren-
tal education level on the students’ SRL had less of an effect on boys (0.22) than on girls
(0.29). Students’ SRL effect on (a) digital reading communication had a greater effect on boys
(0.49) than on girls (0.14) and (b) webpage information retrieval had less of an effect on boys
(0.52) than on girls (0.67). Webpage information retrieval effect on digital reading commu-
nication was found also to have less of an effect on boys’ (0.53) than on girls’ (0.71).
There are four latent variables (i.e. parental education level, SRL capability, webpage infor-
mation retrieval capability, and digital reading communication) in the models for upper-­
elementary males and females. The parental education level result shows that the father’s
education level loading is greater than the mother’s education regardless of the gender of
their child. The SRL result showed that the performance in testing circumstance is better
than in general circumstance regardless of the student’s gender. Basic capability loading is
better than advanced capability in webpage information retrieval, while the basic capability
loading is better than advanced capability in digital reading communication.
RESEARCH IN SCIENCE & TECHNOLOGICAL EDUCATION   15

The gender-specific structural model indicates that both upper-elementary school boys
and girls have the best prediction in the webpage information retrieval capability of digital
reading communication ability. Without the best prediction mentioned above, the boys’ and
girls’ predictions in the SRL ability of webpage information retrieval capability were both
the best. The difference between boys and girls only shows in the prediction loading of
parental education level in SRL and SRL capability in digital reading communication.
Therefore, all the estimated coefficients reached a level of significance. That is, parental
education level, SRL capability, and webpage information retrieval capability are significant
factors influencing digital reading communication capability for both upper-elementary
school boys and girls.

Discussion
This study presents a theoretical model as a generalizable approach to measure the inner
cause–effect relationships between four specific factors. Measurement invariance and struc-
ture equation modeling were used to statistically validate that model comparisons between
boys and girls represented authentic differences in the factors being assessed. Measurement
invariance results demonstrate that all assessment items were interpreted equally among
both boys and girls which insured that comparisons made between groups represented
authentic differences in the constructs that were measured. Model fit analysis results revealed
a robust model supporting and validity and dependability of the proposed relationships of
the factors being assessed. This study therefore offers this two-step assessment approach
as a generalizable method for managing and verifying theoretical cause–effect relationships
against empirical data.
The hypothesized model proposed as a foundation of this study was supported by the
empirical data collected. However, there are two findings that we believe are noteworthy
for science education. The first finding is that the parents’ educational level has the greatest
influence on these elementary students’ SRL strategies, which has been supported by other
researchers (Korat 2011; Lee and Yu 2005; Lin 2009; Miser and Hupp 2012; Scott 2004).
This literature posits that a family’s socioeconomic status, which is closely related to educa-
tion status and family resources, is strongly correlated with the child’s science literacy.
Chen (2013) specifically indicated that upper-elementary school students’ parental education
level significantly affected their science language learning, which coincided with this study’s
finding. The value of this study is that we also found that students’ digital reading commu-
nication ability and webpage information retrieval also have a strong influence on the stu-
dents’ digital reading component of science literacy.
Students’ family socioeconomic status, education background, and home-based educa-
tion resources directly impact children’s SRL ability, webpage information retrieval ability,
and digital reading communication ability in this ICT environment. Apparently, home and
family environments rich in resources and educational priorities provide cultural and physical
contexts that allow self-directed freedom and opportunities to engage and enhance digital
reading. This research found that the SRL ability in both general and test situations has
positive influences on webpage information retrieval ability and digital reading communi-
cation ability. These findings support the claim that SRL abilities greatly influence print-based,
text-reading ability (Sperling et al. 2002). Other researchers have identified learning regula-
tion as a metacognitive ability that has considerable influence on DRL (Bannert 2004).
16   S.-F. CHEN

Ackerman and Goldsmith (2011) found that the contribution of metacognitive regulatory
processes to on-screen learning was less than the contribution it made to paper-based
learning. However, digital text and on-screen reading have become an inevitable trend in
the digital classroom era as Net generation students populate most classrooms today. This
study verified that SRL ability has substantial influence upon webpage information retrieval
ability and digital reading communication ability for upper-elementary level students.
Reading comprehension – a critical factor in learning of all subjects and, therefore,
regarded as one of the most important basic components of reading and on-screen reading
– is quickly becoming a key component as reading transitions from a paper-based process
to a digital process. Digital multi-source situations redistribute the emphases in reading to
locating, selecting, and critiquing information; webpage information retrieval is just a new
form of reading activity, which may result in changes to models and interpretations of read-
ing. When using the Internet webpage as an information resource for learning, one must
apply epistemological beliefs and metacognitive and critical judgments to identify valid
sources and construct understanding (Schraw, Crippen, and Hartley 2006). Henry (2006)
indicated that, even with the abundance of webpage information resources, learners may
easily get lost in the maze of the unfamiliar structures of digital texts if they are not equipped
with sufficient on-line reading and searching strategies and critical thinking. Therefore, it is
not surprising that this study found metacognitive learning regulation strategies have a
profound consequence upon webpage information retrieval ability and digital reading com-
munication ability.
Another finding of this research demonstrated that amongst these upper-elementary
school students, both girls and boys had basically a good model fit in the structural model
analysis with similar results. The only difference between boys and girls was in the prediction
loading of parental education level on SRL and SRL capability on digital reading communi-
cation. Nevertheless, all the estimated coefficients reached a level of significance; and both
parental education level and SRL capability are important factors affecting webpage infor-
mation retrieval and digital reading communication capability for upper-elementary school
boys or girls. As other studies did, this research also pointed out the gender differences
regarding the use of computers. Busch (1995) revealed that the complexity of the computer
task affects the gender difference of the self-efficacy and attitudes toward computers; gender
differences were only found in computer self-efficacy and attitudes regarding complex com-
puter tasks. From this perspective, SRL strategies may be considered as an important factor
in complex digital reading communication tasks on computers.
Digital text reading has many special characteristics, including non-linearity, interactivity,
immediacy, and anonymity. Under these interacting characteristics, obtaining knowledge
by DRL is only possible through cultivation and judgment of DRL. These new literacy abilities
will allow learners to make use of digital reading auxiliary tools to identify important ques-
tions, locate information, and evaluate the effectiveness of the information. Integrating
multi-sources of digital information requires learners to construct answers for their focus
questions and to communicate their answers to others (Coiro and Dobler 2007; Henry 2006).
Lee and Wu (2012) reported that learners who have a positive attitude toward and high
confidence in high-level information and communications technology tasks usually have a
more active engagement in DRL. This study also found that SRL ability has significant influ-
ence on webpage information retrieval ability and digital reading communication ability
for both girls and boys in upper-elementary school. The webpage information retrieval items
required students to demonstrate metacognition and critical thinking ability to enact SRL
RESEARCH IN SCIENCE & TECHNOLOGICAL EDUCATION   17

strategies and high-order cognition skills. Clearly, critique and critical analysis are required
with multiple sources and potential discrepant information, which are not apparent in using
a single authorized source such as the textbook.
Though the subjects mainly observed in this study were upper elementary school stu-
dents from Taiwan, it could be inferred that the physical and psychological developments
of elementary school students are similar worldwide. Therefore, a certain degree of gener-
alizability to other countries could be found in the research findings of this study.

Conclusion
The results confirmed that these upper-elementary school students’ parental education affects
their SRL ability, webpage information retrieval ability, and digital reading communication
ability. Parents with higher levels of education might be able to provide more abundant
learning resources (e.g. textual materials, technologies, etc.), opportunities, and encourage-
ment to their children. Hence, future studies should focus on helping students from lower
socioeconomic status and especially low education level to close the technology gap and
enhance ICT opportunities with supportive guidance that might not be available in their
family and home so as to develop these multiple abilities, which are critical to students’ future
learning. In addition, school authorities have to pay more attention to the allocation of learning
resources to ensure each student’s ICT opportunities; it is never too early to provide students
with ICT opportunities in elementary school. This study was conducted within a sample of
students in southern Taiwan. Further investigations using the model and methods presented
in this study are needed in populations within and outside of Taiwan for the purpose of
cross-validating the findings herein and to extend the knowledge on digital reading literacy
gained in this investigation. Another limitation involves the measure of socioeconomic status
using only the educational level of parents rather than their occupational prestige and income.

Acknowledgments
We acknowledge and express appreciation to Professor Larry Yore (University of Victoria) and Mrs.
Shari Yore for their mentoring assistance in relation to this article.

Disclosure statement
No potential conflict of interest was reported by the author.

Funding
This work was supported by the National Science Council of Taiwan [grant number NSC
103-2420-H-153-001-MY2].

References
Ackerman, R., and M. Goldsmith. 2011. “Metacognitive Regulation of Text Learning: On Screen versus
on Paper.” Journal of Experimental Psychology: Applied 17 (1): 18–32.
Bannert, M. 2004. “Designing Metacognitive Support for Hypermedia Learning.” In Instructional Design
for Multimedia-Learning, edited by H. M. Niegemann, D. Leutner, and R. Brunken, 19–30. Munster:
Waxmann.
18   S.-F. CHEN

Busch, T. 1995. “Gender Differences in Self-Efficacy and Attitudes toward Computers.” Journal of
Educational Computing Research 12 (2): 147–158.
Carlsen, W. S. 2007. “Language and Science Learning.” In Handbook of Research on Science Education,
edited by S. K. Abell and N. G. Lederman, 57–74. Mahwah, NJ: Lawrence Erlbaum.
Carnap, R. 2002. The Logical Syntax of Language. Translated by A. Smeaton. Chicago, IL: Open Court.
Chen, S.-W., and W.-G. Yang. 2006. “The Impact of a Systemic Functional Linguistics-Based Science Text
and a Conventional Science Text on Students’ Reading Comprehension.” Journal of Taiwan Normal
University: Mathematics & Science Education 51 (1–2): 107–124.
Chen, S.-F. 2013. “Structural equation modeling analysis of factors affecting science language
achievement among elementary students.” Research and Development in Science Education Quarterly
67: 1–22.
Chen, S.-F. 2014. Structural Equation Modeling: The Application of Mplus. Taipei, TW: Psychological
Publishing Co.
Coiro, J., and E. Dobler. 2007. “Exploring the Online Reading Comprehension Strategies Used by Sixth-
Grade Skilled Readers to Search for and Locate Information on the Internet.” Reading Research
Quarterly 42 (2): 214–257.
DeVellis, R. F. 2011. Scale Development: Theory and Applications. Thousand Oaks, CA: SAGE.
Efklides, A., and S. P. Vlachopoulos. 2012. “Measurement of Metacognitive Knowledge of Self, Task, and
Strategies in Mathematics.” European Journal of Psychological Assessment 28 (3): 227–239.
Ford, C. L., and L. D. Yore. 2012. “Toward Convergence of Critical Thinking, Metacognition, and Reflection:
Illustrations from Natural and Social Sciences Teacher Education, and Classroom Practice.” In
Metacognition in Science Education: Trends in Current Research. 40 vols, edited by A. Zohar and Y. J.
Dori, 251–271. Dordrecht: Springer.
Hair, J. F., W. C. Black, B. J. Babin, and R. E. Anderson. 2014. Multivariate Data Analysis. 7th ed. Harlow:
Pearson Education Limited.
Henry, L. A. 2006. “SEARCHing for an Answer: The Critical Role of New Literacies While Reading on the
Internet.” The Reading Teacher 59 (7): 614–627.
Hsu, Y.-S., M.-H. Yen, W.-H. Chang, C.-Y. Wang, and S. Chen. 2016. “Content Analysis of 1998–2012
Empirical Studies in Science Reading Using a Self-Regulated Learning Lens.” International Journal
of Science and Mathematics Education 14 (1): 1–27.
Huang, H.-P., and L. D. Yore. 2003. “A Comparative Study of Canadian and Taiwanese Grade 5 Children’s
Environmental Behaviors, Attitudes, Concerns, Emotional Dispositions, and Knowledge.” International
Journal of Science and Mathematics Education 1 (4): 419–448.
Jacobs, J. E., and S. G. Paris. 1987. “Children’s Metacognition about Reading: Issues in Definition,
Measurement, and Instruction.” Educational Psychologist 22 (3–4): 255–278.
Kline, R. B. 2011. Principles and Practice of Structural Equation Modeling. New York: Guilford Press.
Korat, O. 2011. “Mothers’ and Teachers’ Estimations of First Graders’ Literacy Level and Their Relation
to the Children’s Actual Performance in Different SES Groups.” Education and Treatment of Children
34 (3): 347–371.
Lee, D.-R., and M.-N. Yu. 2005. “The Verification of a Structural Equation Model on SES, Siblings,
Household Education Resources and Educational Achievement: Using the Empirical Data of the
2001 TEPS.” Taiwan Journal of Sociology of Education 5 (2): 1–48.
Lee, Y.-H., and J.-Y. Wu. 2012. “The Effect of Individual Differences in the Inner and Outer States of ICT on
Engagement in Online Reading Activities and PISA 2009 Reading Literacy: Exploring the Relationship
between the Old and New Reading Literacy.” Learning and Individual Differences 22 (3): 336–342.
Leu, D. J. 2007. Expanding the Reading Literacy Framework of PISA 2009 to Include Online Reading
Comprehension. A working paper commissioned by the PISA 2009 Reading Expert Group. Princeton,
NJ: Educational Testing Services.
Lin, H.-F. 2009. “The Study of Junior High School Students’ Scientific Literacy: The Cases of Taiwan, Japan,
Korea and Hong Kong in PISA 2006.” Journal of Educational Research and Development 5 (4): 77–108.
Miser, T. M., and J. M. Hupp. 2012. “The Influence of Socioeconomic Status, Home Environment, and
Childcare on Child Language Abilities.” Current Psychology 31: 144–159.
Mokhtari, K., and C. A. Reichard. 2002. “Assessing Students’ Metacognitive Awareness of Reading
Strategies.” Journal of Educational Psychology 94 (2): 249–259.
RESEARCH IN SCIENCE & TECHNOLOGICAL EDUCATION   19

Mullis, I. V. S., M. O. Martin, A. M. Kennedy, K. L. Trong, and M. Sainsbury. 2009. The PIRLS 2011 Assessment
Framework. Chestnut Hill, MA: TIMSS & PIRLS International Study Center, Lynch School of Education,
Boston College.
Organization for Economic Co-operation and Development. 2006. Assessing Scientific, Reading and
Mathematical Literacy. Paris: Author.
Organization for Economic Co-operation and Development. 2013. PISA 2015 Draft Science Framework.
Paris: Author.
Pintrich, P. R., and E. V. de Groot. 1990. “Motivational and Self-Regulated Learning Components of
Classroom Academic Performance.” Journal of Educational Psychology 82 (1): 33–40.
Samuelstuen, M. S., and I. Braten. 2005. “Decoding, Knowledge, and Strategies in Comprehension of
Expository Text.” Scandinavian Journal of Psychology 46 (2): 107–117.
Schellings, G., C. Aarnoutse, and J. V. Leeuwe. 2006. “Third-Grader’s Think-Aloud Protocols: Types of
Reading Activities in Reading an Expository Text.” Learning and Instruction 16 (6): 549–568.
Schmar-Dobler, E. 2003. “Reading on the Internet: The Link between Literacy and Technology.” Journal
of Adolescent & Adult Literacy 47 (1): 80–85.
Schraw, G., K. Crippen, and K. Hartley. 2006. “Promoting Self-Regulation in Science Education:
Metacognition as Part of a Broader Perspective on Learning.” Research in Science Education 36 (1–2):
111–139.
Schraw, G., and R. S. Dennison. 1994. “Assessing Metacognitive Awareness.” Contemporary Educational
Psychology 19: 460–475.
Scott, J. 2004. “Family, Gender, and Educational Attainment in Britain: A Longitudinal Study.” Journal of
Comparative Family Studies 35 (4): 565–589.
Sperling, R. A., B. C. Howard, L. A. Miller, and C. Murphy. 2002. “Measures of Children’s Knowledge and
Regulation of Cognition.” Contemporary Educational Psychology 27 (1): 51–79.
Thomas, G. P., and D. Anderson. 2013. “Parents’ Metacognitive Knowledge: Influences on Parent–Child
Interactions in a Science Museum Setting.” Research in Science Education 43 (3): 1245–1265.
Tippett, C. D., & R. J. Anthony, (2011). “Changing from users to producers of multimodal texts: A
theoretical framework based on cognition, metacognition, semiotics, and systemic functional
linguistics. Paper presented at the annual international conference of the National Association for
Research in Science Teaching, Orlando, April 7.
Wang, J.-R., S.-F. Chen, I. Fang, and C.-T. Chou. 2013. “Comparison of Taiwanese and Canadian Students'
Metacognitive Awareness of Science Reading, Text, and Strategies.” International Journal of Science
Education 36 (4): 693–713.
Wang, J.-R., and S.-F. Chen. 2014. “Exploring mediating effect of metacognitive awareness on
comprehension of science texts through structural equation modeling analysis.” Journal of Research
in Science Teaching 51 (2): 175–191.
Worthington, R., and T. Whittaker. 2006. “Scale Development Research: A Content Analysis and
Recommendations for Best Practices.” The Counseling Psychologist 34: 806–838.
Yang, Y., and J.-E. Gustafsson. 2004. “Measuring Socioeconomic Status at Individual and Collective
Levels.” Educational Research and Evaluation 10 (3): 259–288.
Yore, L. D., and C. D. Tippett. 2014. “Reading and Science Learning.” In Encyclopedia of Science Education,
edited by R. Gunstone, 1–9. Dordrecht: Springer.

You might also like