You are on page 1of 26

1

RUNNING HEAD: The home reading environment


Mattson &Mathew

The home reading environment: A comparison of passive physical and digital home reading

traits on students’ PISA reading performance

David Mattson & Katie Mathew

Drexel University
2
The home reading environment
Mattson & Mathew
The home reading environment: A comparison of passive physical and digital home reading

traits on students’ PISA reading performance

Introduction

Beginning in 2000, PISA test scores have ignited the international education community

with speculation as each country measures itself against the best scores around the world. The

most natural and expected questions inevitably follow such comparisons: Why do some countries

score higher than others? PISA rotates the emphasis of its testing every four years, and Reading

was the subject of highest scrutiny in the testing wave of 2018 (the data used for this study was

taken from the 2018 testing).  Many attributes of students’ lives have been measured for

correlation to higher reading scores, one of the most widely recognized is the student home

environment. While prior studies have evaluated the positive influence of strong home literacy

environment (Wiescholek et al., 2018), or emphasized the contributions of books in the home as

part of Scholarly Culture (Sikora et al., 2019), both theories recognize that digital media and

digital reading habits may change the influence of these factors on students’ reading scores

(Sikora et al., 2019, Wiescholek et al., 2018). Exactly how digital reading habits will emerge as

influential will be considered by analyzing correlations between attributes of Home Literacy

Environment and Scholarly Culture with combined measures of student familiarity with digital

resources, such as the PISA measure of ICT. 

Motivation

A wide breadth of research has been conducted demonstrating that a home environment

which encourages reading outside school is strongly correlated to higher reading scores (Hume et

al., 2015; Wiescholek et al., 2018). Likewise, educators and parents are curious about how

students develop personal reading habits outside school that may increase reading scores (Clark,
3
The home reading environment
Mattson & Mathew
2011; Hume et al., 2015).  Home literacy environment (Wiescholek et al., 2018) and Scholarly

Culture Theory (Sikora et al., 2019) are two theories that explore the relationship between

students' home environment and their reading proficiency. 

Home literacy environment theories divide relevant attributes of a students’ home

environment into passive or active traits (Wiescholek et al., 2018).  Examples of active traits are

parents reading a book to students or students reading to their parents. Passive traits are

environmental features observable to the student, such as the number of books in the home, the

presence of a desk in the house, or a students’ observation of a parent reading for personal

enjoyment. According to Home Literacy Environment theories, passive traits can be separated

from active traits and have similar strength of influence on reading proficiency (Wiescholek et

al., 2018). These theories suggest that passive Home Literacy Environment traits should be

considered significant contributors to developing reading at home and higher reading

proficiency. 

Scholarly Cultural theory emphasizes the presence of a home library as a key indicator of

routines that foster a “Scholarly Culture” (Sikora et al., 2019).  A study evaluating the home

environments of students across 31 societies shows that home library size is strongly correlated

with literacy activities of parents, parental attitudes towards literacy and student academic

attainment in standardized testing (Sikora et al., 2019).  While Home Literacy Environment

focuses on the provision of reading materials in general, Scholarly Culture theory differentiates

itself by giving primacy to the influence of a home library over the presence of other literacy

activities.  

The presence of a home library, or a high number of books at home is correlated with

higher internet familiarity (Evans et al., 2014). This suggests that students who have access to
4
The home reading environment
Mattson & Mathew
more books at home are more likely to have access to e-reading materials that support digital

literacy activities.  Scholarly Culture theories do not distinguish between the literacy activities

that directly utilize the books in the home library, but rather see this indicator as representative of

an environment that includes a diverse set of routines (including digital reading materials) that

support literacy. 

Although the presence of digital resources as complementary to traditional reading is

widely incorporated in curriculum (Lindsay, 2010), the question of whether digital reading habits

can successfully replace traditional books in the home as an indicator of academic achievement

in reading has not been established by current testing results.  Questions remain about the

viability of digital texts and e-reading to replace books as indicators of important literacy

activities (Neumann, 2016). In their research exploring the context of reading at home,

Rouncefield and Tolmie emphasize that reading is part of a complex experience, much more than

merely passing sight over text on paper, or symbols on a screen.  Students read differently when

reading online, preferring to skim text or browse (2011). Their nuanced exploration of the

reading at home “experience” suggests that many predictions of a one to one switch between

traditional books and digital reading are mistaken in that they ignore the subtleties of reading

routines that encompass much more than the object which contains the text (Rouncefield &

Tolmie, 2011). 

Nevertheless, a growing number of students with access to e-reading materials suggests

that internet access and technology may have already changed the current composition of Home

Literacy Environments, even if educational theories have yet to determine exactly how this

change will manifest in testing.  Analyzing available data for potential trade-offs in predictive
5
The home reading environment
Mattson & Mathew
value between internet familiarity and traditional HLE traits may be one way of analyzing how

these in reading routines and home literacy will affect testing scores in the future.  

In the PISA 2018 test and student questionnaire, data was obtained that sought to provide

a richer understanding of students’ experience outside of school and how these factors contribute

to their education (OCED, 2019).  As reading was the focus of the 2018 PISA test, the

questionnaire targeted information about what Weischolek et al. describe as their “home literacy

environment” (Weischolek et al., 2018).  Students, principals, and parents were asked about the

number of books in students’ homes and if they had a desk at home.  As both Home Literacy

Environment theories and Scholarly Culture theory posit that books in the home represent a

benefit towards home literacy activities (Lindsay, 2010; Schubert and Baker, 2010; Sikora 2019),

the data gathered by the 2018 PISA provides an opportunity for measuring possible relationships

between passive traits of home literacy environment and reading proficiency.

In order to evaluate the influence of digital access and digital resources on student

reading proficiency, the PISA 2018 Information Communication and Technology scale was

utilized as a measure of the influence of digital resources on students’ literacy activities.  The

2018 PISA formulated the ICT score by having students complete a questionnaire which focused

on the availability and use of information and communications technology at home and school

and on students’ ability to carry out tasks on computers and their attitudes towards using

computers (OECD, 2019a).  Although these tasks were not explicitly related to literacy activities,

logic suggests that students with low ICT scores would not be using digital resources as part of

any home literacy routine. Additionally, studies have shown that children’s exploration of online

print resources has a positive influence on emergent literacy skills (Neumann, 2016). The

alignment of these theories and their prescribed attributes of literacy activities with the data
6
The home reading environment
Mattson & Mathew
collected by the PISA questionnaire and reading test scores presents the opportunity to apply

these theories and test their predictive ability.

Research Questions

Based on the literature review and aforementioned motivation the following research questions

were proposed:

Q1: Is there a statistically significant relationship between PISA reading score outcomes

and passive home literacy traits as measured by the presence of a desk at home and books

in the home when controlling for country and gender?

Q2: Is there a statistically significant relationship between PISA reading score outcomes

and digital home literacy traits as measured by the PISA ICT score when controlling for

country and gender? 

Q3: Does the relationship between PISA reading score outcomes and desk at home differ

by gender controlling for passive and digital reading traits?

Data

A subsample of data from PISA 2018 was utilized for this paper. The dataset includes data from

a sample of 15-year-olds from the Japan (n = 4,249) and Germany (n = 6,011). PISA uses a

stratified two-stage sample design where schools are sampled using probability proportional to

size sampling, and students are sampled with equal probability within schools. Sampled students

receive a final weight which indicates how many other students from the population are

represented (Westat, 2016).


7
The home reading environment
Mattson & Mathew
Measures

The following measures were utilized for the current analysis:

Country: Data for this paper was taken from a subsample of two countries: Japan (n = 4,249)

and Germany (n = 6,011) (Germany = 0; Japan = 1).

Gender: Students indicated whether they identify as male or female (Female = 0; Male = 1).

Desk at home: Students indicated whether they have a desk at home for studying or not (No

desk= 0; Desk = 1).

Number of books at home: In the student questionnaire, students reported approximately how

many books are in their homes. This variable was originally coded into six categories, however,

previous research has shown that the benefits of number of books in the home increases reading

outcomes exponentially from 0 – 100 but tapers off after 101 books (Evans, Kelley & Sikora,

2014). For the purposes of this paper, the books variable was transformed into four categories:

0 - 10 books, 11 - 25 books, 26 - 100 books, and 101 - 500+ books. A system of dummies was

created for these variables (Books 1, Books 2, Books 3, Books 4).

Information and Communications Technology (ICT): Students completed a questionnaire which

focused on the availability and use of information and communications technology at home and

school and on student’s ability to carry out tasks on computers and their attitudes towards using

computers (OECD, 2019b). Responses to the questionnaire were combined and standardized

using weighted likelihood estimates (WLE) such that the mean of 0 represents the average

worldwide score.

Reading Score: Students completed various measures of reading performance.


8
The home reading environment
Mattson & Mathew
Methods

In this study descriptive and inferential statistics will be used to explore the research

questions. Descriptive statistics will be used to summarize the dataset with measures of

frequency, central tendency and variation reported. Inferential statistics will be utilized to draw

inferences about the sample to the larger population. Linear regression and multiple regression

will be used to model predictions of the outcome variable based on various combinations of the

independent variables. Tables and figures will be used to visualize the results. Finally, a

discussion of the results will be included at the end of the paper.

Results

A total of 10,260 15-year-old students were sampled for this paper to explore the

relevance of passive reading traits on PISA reading scores. Table 1 presents descriptive statistics

for the sample. Students came from two countries: Japan (n = 6,011) and Germany (n = 4249).

These sample countries were selected because they both have a high economic index, place

similar values on education and it was assumed that students would generally have the same

access to technology for education and leisure. One case was removed from the sample for

missing data; the final sample size was n = 10,259. Gender across the sample was mostly

balanced, with the Japanese sample comprising marginally more female students (51%) than the

German sample (48%). 

Variables measuring passive reading traits were organized into two categories: physical

home traits and digital traits. Home physical traits were measured by whether the student

indicated having a desk to study at home and the number of books they estimated being in their

home. Students in Germany were marginally more likely to report having a desk dedicated for

studying at home (96%) compared with students in Japan (94%). The number of books in the
9
The home reading environment
Mattson & Mathew
home was also used as a physical home trait. Most students across the sample estimated 26 or

more books in their home (73%); however, German students were more likely to indicate having

over 101 books (49%) compared with Japanese students (36%).

Information Communications and Technology (ICT) was used as a measure of digital

passive reading traits. ICT is a composite score of various questionnaire items. Composite scores

were standardized across the entire PISA data set with a score of zero indicating the average

across the entire sample of 79 countries. A positive score indicates a higher than average

familiarity with ICT resources, whereas a negative score indicates a lower than average.

Japanese students indicated less familiarity with ICT resources on average (M= -0.52, SD= 0.83)

compared to German students (M= 0.06, SD= 0.86).

PISA reading score was used as the outcome variable. German students performed

slightly better on PISA reading (M=516) compared with Japanese students (M= 510), although

there was slightly less variation in Japanese scores (SD= 96) compared to German scores (SD=

101).
10
The home reading environment
Mattson & Mathew
Table 1:

Descriptive statistics for 15-year-old Japanese and German students various passive reading
traits and reading score (N = 10,259)

  Total Japan Germany 

Country 10,259 4,248 6,011


Gender      
  Female 50% 51% 48%

  Male 50% 49% 52%

Desk at home      
  Yes 95% 94% 96%
  No 5% 6% 4%
Number of books at      
home
  0 – 10 books 12% 14% 10%
  11 – 25 books 15% 16% 13%
  26 – 100 books 32% 34% 28%
  101 – 500+ books 41% 36% 49%
       

  Mean SD Mean SD Mean SD


 

ICT1 -0.28 0.89 -0.52 0.83 0.06 0.86

Reading score 510 98 505 96 516 101


 

1
Standardized variable
Data source: PISA, 2018
Figure 1 is a bar graph showing mean reading score by number of books in the home for

the overall sample. As can be seen, there is a positive relationship between number of books in

the home and reading score, where more books in the home is associated with higher reading
11
The home reading environment
Mattson & Mathew
scores. A descriptive comparison of the means shows an approximately 30 point increase in

reading score based on each category of books in the home. The largest jump in mean reading

score is between 26 - 100 books and 101 - 500+ books where an average 45 point increase in

reading score is observed.

Figure 1:

Bar graph showing mean reading score by number of books in the home (N = 10,259)

Data source: PISA, 2018

Figure 2 is a clustered boxplot showing a comparison of median ICT resource familiarity

according to country and gender. Across the sample males were slightly more likely in both

countries to indicate greater familiarity with ICT resources compared with females. The boxplots

also indicate that the dispersion of ICT scores for both males and females, in both Japan and

Germany, are stretched. The stretch of the bottom and top quartile is noteworthy and a large

number of outliers can be observed at both ends of the distrubution. This may suggest that there
12
The home reading environment
Mattson & Mathew
is a wide variability in access to ICT resources in schools and at home in these countries.

Alternatively, there may have been wide variability in the interpretation of PISA ICT

questionnaire items.

Figure 2:

Information communications and technology (ICT) resources by gender reported by 15-year-


olds in Japan and Germany (N = 10,259)

Data Source: PISA, 2018

Figure 3 is a clustered bar graph showing reading score according to whether students

reported having a desk at home for studying in each country. Overall, average reading scores are

higher for students who reported having a desk at home. The increase in reading score between

not having a desk and having a desk is slightly larger in Germany than Japan.

Figure 3:

Clustered bar graph showing mean reading score by desk in the home by country (N = 10,259)
13
The home reading environment
Mattson & Mathew

Data source: PISA, 2018

Figure 4 is a scatterplot showing the relationship between ICT resources and Reading

scores. A weak, positive linear relationship can be observed between ICT and reading, where

lower ICT familiarity is weakly associated with lower reading scores and higher ICT familiarity

is weakly associated with higher reading scores.

Figure 4:

Scatterplot showing correlation between reading score and ICT resources (N = 10,259)
14
The home reading environment
Mattson & Mathew

Data source: PISA, 2018

As mentioned in the introduction, an association between books in the home and access

to internet has been previously established in the literature (Evans et al., 2014). To test whether

there is an association between books in the home and ICT, a one-way analysis of variance was

conducted. Between group comparisons were statistically significant F(3, 10255)= 300, p<0.001

and post-hoc tests showed statistically significant differences between all categories of books on

ICT. Figure 5 is a means plot of ICT based on categories of books in the home. In line with the

previous findings, more books in the home is associated with higher ICT familiarity.

Figure 5:

Means plot depicting mean ICT score by number of books in the home (N = 10,259)
15
The home reading environment
Mattson & Mathew

Data source: PISA, 2018

To test if and how passive reading traits predict reading score, a multiple regression was

conducted. The proceeding multiple regression model presents the prediction of the continuous

dependent variable, reading, based on multiple categorical (desk, books at home, country,

gender) and continuous (ICT) independent variables. Table 2 presents the results of the multiple

regression model. Assumptions were checked throughout the process of running the steps of the

model and will be discussed throughout the analysis.

Model 1 explored the predictive effects of having a desk at home on reading score. In

checking for outliers, fifteen outliers were identified below or above 3 standard deviations

beyond the mean, however, these scores were determined to be important for analysis and were

therefore retained. Homoscedasticity and multicollinearity were not assessed at this stage

because only one categorical independent variable was included in the model. The model was

statistically significant, F(1, 10257) = 48.21, p<0.001, showing that simply having a desk at
16
The home reading environment
Mattson & Mathew
home predicts a 30.30 point average increase in reading score. The variance in reading score

accounted for by desk at home, however, was very low (R = 0.005) or less than 1%.
2

Four categories of books in the home were assessed for their predictive effect on reading

scores (Model 2). Given that number of books was a categorical variable, the asssumptions of

linear relationship, homoscedastity, and multicollinearity were not evaluated. Twenty-six outliers

were observed, but were ultimately left in for the analysis. A system of dummy variables was

generated and the category 0 - 10 books was used as the reference category. The model was

statistically significant, F(3,10255) = 541.04, p<0.001 showing a main effect of books in the

home on reading score. Compared to the reference category, 0 - 10 books: having 11 - 25 books

in the home predicted a 34.23 point increase on average in reading, having 26 - 100 books in the

home predicted a 67.41 point increase on average in reading, and having over 101 books in the

home predicted a 101.87 point increase on average in reading. Number of books in the home is

associated with R = 0.137 or 13.7% of the variance in PISA reading scores.


2

Model 3 assessed the predictive value of ICT on reading scores. Figure 4 shows a linear

relationship between the continuous independent variable (ICT) and the continuous dependent

variable (Reading). Eighteen outliers were found to fall above or below 3 standard deviations

beyond the mean, however, the pattern of results appeared relatively balanced in the scatterplot

(Figure 4), therefore the outlying data was retained for analysis. A visual inspection of

standardized predictor values and standardized residuals indicated that variances along the line of

best fit are similar as you move along the line suggesting that the homoscedasticity assumption

was met. The model was statistically significant, F(1,10257) = 184.77, p<0.001 showing a main

effect of ICT on reading. ICT is a standardized variable and every one point standard deviation

increase in ICT predicts a 14.72 point increase in reading score on average. However, the amount
17
The home reading environment
Mattson & Mathew
of variance in reading score accounted for by ICT was very small, R =0.018 or approximately
2

2%.

Models 4 and 5 analyzed the predictive value of home physical traits (Desk and Books)

and digital traits (ICT) on reading respectively, controlling for country and gender. In model 4

multicollinearity was assessed via correlations and variance inflaction factor; all thresholds were

deemed to be in the realm of acceptable, meaning that the explanatory variables were not highly

related to one another. Model 4 was statistically significant F(6, 10252)= 299.72, p<0.001 and

demonstrated that, when country and gender are controlled, physical reading traits exert a

predictive influence on reading score. However, in comparing Models 1 and 2 to 4, it appears

that adding books to the model with a desk reduces the predictive influence of a desk on reading

score, suggesting that books are a more important predictor of reading than having a desk. The

variance accounted for by the model is R = 0.149 or approximately 15% of the variance in
2

reading scores can be accounted for by desk and books controlling for country and gender.

The predictive influence of digital trait (ICT) on reading score controlling for country and

gender was assessed in Model 5. A linear relationship between ICT and reading was previously

established and outliers were retained. Homoscedacity was visually assessed and deemed

acceptable and multicollinearity values were within the acceptable range. The model was

statistically significant F(3,10,255)= 107.42, p<0.001 suggesting that digital reading traits exert a

predictive influence on reading scores when country and gender are controlled. Compared to

Model 3, controlling for country and gender in Model 5 marginally increased the variance

accounted for from approximately 2% to 3%. Given that gender was significant and country was

not, this may suggest that gender and ICT may have a small interactive effect on reading. This
18
The home reading environment
Mattson & Mathew
was assessed in a later model. In models 4 and 5, country was found to not be a significant

predictor and was therefore left out of proceeding models.

In Model 6, physical traits (desk, books) and digital traits (ICT) were assessed for their

explanatory value in predicting reading, controlling for gender. The model was significant F(6,

10252)= 301.98, p<0.001. Notably, when combined with physical traits in the model, the

predictive power of ICT on reading decreased significantly. Model 7 shows that the influence of

books at home significantly reduces the predictive impact of ICT on reading suggesting that

books at home are a more important predictor of reading scores than ICT. As previously

analyzed using ANOVA, books in the home and ICT are associated where higher numbers of

books corresponding with higher ICT. It may be that books in the home and ICT are overlapping

measures.

Two interactions were tested: gender and desk, and gender and ICT. Model 8, which

included the interaction between gender and desk was statistically significant, F(7, 10251)=

259.45, p<0.001 showing that there is an average positive effect of having a desk for males. Male

students who have a desk score 15.63 points higher on reading on average. The amount of

variance accounted for in this model which includes desk, books, ICT, gender and the interaction

between gender and desk was R =0.151, or approximately 15% of the variance in reading scores.
2

This was the strongest model found in this analysis and the regression equation was:

READING’ = B + B (DESK) + B (11 - 25 BOOKS) + B (26 - 100 BOOKS) + B (101 - 500+


0 1 2 3 4

BOOKS) + B (ICT)) +B (Gender) + B (Gender*Desk) + E


5 6 7
19
RUNNING HEAD: The home reading environment
Mattson &Mathew
Table 2:
 
Multiple regression analysis of PISA reading scores on passive reading traits, country and gender among 15-year-olds in Japan and
Germany (N = 10,259)
 
  Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9

Constant 481.04 439.51 513.88 440.69 526.48 442.88 442.05 450.30 442.71
Desk at home 30.291***     13.42**   12.33** 4.59 12.33**
Books in the                
home1
11 – 25 books   34.28***   32.24***   31.52*** 33.53*** 31.35*** 31.51***
26 – 100 books   67.40***   65.34***   64.00*** 66.08*** 63.89*** 64.00***
101 – 500+ books   105.87***   104.04***   101.72*** 103.37*** 101.66*** 101.74***
ICT2     14.72***   14.97*** 3.99*** 3.49*** 3.89** 3.39*
 
Country (Japan)       -0.01 -0.01
Gender (Male)       -21.18*** -22.16*** -21.45*** -36.27*** -21.17***
Gender X Desk             15.63*  
Gender X ICT               0.99

R2 0.005 0.137 0.018 0.149 0.030 0.15 0.138 0.151 0.15


1
Reference: 0 – 10 books
2
Standardized
*p<0.05, **p<0.01, ***p<0.001
20
RUNNING HEAD: The home reading environment
Mattson &Mathew
To contextualize the results of the regression model, various simulations were calculated to

demonstrate the predicted reading score outcomes of students with or without strong home literacy

traits. In order to help give perspective to these simulations and their outcome scores, we reference

the descriptive statistics in Table 1 to show the mean PISA reading score of students in the sample

population was 510.  Also, recall that studies have shown that one additional year of schooling

equates to an increase of approximately 39-40 points in PISA reading score on average (OCED,

2010). 

First, a simulation for a male student with low physical home reading traits (no desk, 0 - 11

books) and low digital reading traits (10th percentile ICT = -1.27).

READING’ = 450.30 + 12.33(0) + 3.89(-1.27)) - 36.27(1)) + 15.63(1*0)

This student would be predicted to score 409.09 on PISA reading. 

Next, a simulation for a female student with high physical home reading traits (desk, 101-

500+ books) and high digital reading traits (90th percentile ICT = 0.77).

READING’ = 450.30 + 12.33(1) + 101.66(1) + 3.89(0.77)) - 36.27(0) + 15.63(0*1)

This student would be predicted to score 567.29 on PISA reading. Although these simulations are

designed to hypothesize two very different home literacy enviornments, the difference in predicted

reading outcome is approximately 160 points, or four full academic years of instruction (OECD,

2010).

Discussion

Symmetry between the sample countries was observed descriptively. In order to better

control for extraneous variables the researchers chose countries where a high percentage of students

had access to the internet, similar levels of reading proficiency, and approximately similar

populations of genders had taken the test. The decision to include countries with such similarities
21
The home reading environment
Mattson & Mathew
was motivated by a desire to test the literacy routines and environments of the students with the

least interference of disparate wealth or base educational opportunity between countries. Perhaps

such factors could have been controlled for by additional regression models, but the researchers felt

that a clear picture of the variables of interest would benefit from the absence of such distractions.

The equivalence between countries was also borne out in the regression model where country

differences were not found to be a significant predictor in any of the models tested.

The decision to apply linear regression modeling was made based on the research questions

and the nature of the predictive quantities the researchers intended to measure. In order to assess the

theoretical influence of Home Literacy Environment (Wiescholek et al. 2018) and Scholarly Culture

theory (Sikora, 2009) on reading score outcomes, the researchers needed to isolate the factors these

theories listed as most important to literacy activities at home while controlling for other variables

that might interfere with the interpretation of the analytic results. This required the researchers to

measure the effect of desk, books in the home and ICT as categories of influence, while controlling

for measures such as gender and country. Linear regression modeling can be applied when

researchers intend to test the relationship of a single variable with a dependent variable outcome

while controlling for other variables of influence (Lewis et al. 2004).  

After completing various linear regression models to analyze the predictive relationship

between independent variables ‘Desk at Home’ and ‘Books at Home’ (representing the passive

home literacy traits) and the PISA ‘ICT score’ (representing digital literacy traits), the results

confirmed that both passive and digital literacy traits have a statistically significant predictive

relationship with the dependent outcome variable PISA reading scores, when controlling for gender

and country. Therefore the null hypotheses were rejected for both research questions one and two,
22
The home reading environment
Mattson & Mathew
and the alternative hypotheses that these variables have statistically significant relationships to

PISA reading scores when controlling for country and gender were accepted.

However, the researchers found that the regression models of books in the home had the

strongest predictive relationship to the dependent variable, PISA Reading scores, over ICT and

desk. The regression model including desk at home resulted in a very small R square, or less than

1% of the variance in reading. The regression model of digital home literacy variable ICT

controlling for country and gender resulted in a R square of only 3%. In addition, in later regression

analysis models that included measures of both ‘Books in the home’ and ICT, the correlation

coefficient of ICT fell from 14 to 3 points. This may be due to a strong covariation between books

in the home and ICT familiarity, where books in the home act as a proxy for a combination of

passive reading traits (ICT, desk, and others). 

The strongest model predicting reading included an interaction between desk at home and

gender, controlling for desk, books, and ICT. The model was significant allowing us to reject the

null hypothesis of research question three, which was that the interaction of gender and desk do not

exert an influence on reading. However, the additional variance accounted for by the addition of this

interaction was very small (0.1%). The model was used to generate two simulations. The first

showed that a male student, with no desk, 0 - 10 books, and who fell in the 10th percentile on ICT

(low home reading traits) would be expected to score 409 on PISA reading. Conversely, a female

student, with a desk, 101 - 500+ books, and who fell in the 90th percentile on ICT (high home

reading traits) would be expected to score 567 on PISA reading. The difference between these two

theoretical students is 158 points, which is roughly equivalent to four grade levels (OCED, 2010).

The comparison of these models provide strong evidence that passive home literacy traits exert a
23
The home reading environment
Mattson & Mathew
significant influence on reading performance. In other words, the home environment plays an

important role in students' reading proficiency.

Ultimately, the highest correlation coefficient consistently found across the models was

associated with students who reported more than 101 Books in the home. On average, students who

reported over 100 books at home were predicted to score over 100 points higher in PISA reading

scores than students with 0-10 books in the home. This suggests that traditional passive reading

traits, like having a plentiful home library, continues to be impactful for students’ reading

proficiency in 2018.

The results of the regression analysis correspond to previous studies that indicate a positive

association between home libraries and higher scores in standardized reading tests. However, the

lack of evidence to suggest that digital passive traits as measured by ICT have equally powerful

predictive ability in relationship to PISA reading scores should not dissuade researchers from

pursuing further research into the influence of digital media on home literacy activities and literacy

proficiency. 

Limitations

One of the most important limitations of this study relates to the variable used as

representative of digital home literacy, the Information and Communications Technology score.

While the ICT score does accurately represent students' access to the internet and familiarity with

technology that could be used for e-reading or literacy activities at home, access and familiarity

does not imply the type of routines promoted by Home Literacy Environment or Scholarly Culture

Theory. The selection of the ICT variable as representative of a home digital literacy trait should

adumbrate the complexity of creating a measure to test precisely how students use digital

technology at home. 
24
The home reading environment
Mattson & Mathew
Another important limitation was the inability to combine categories of the Books at home

variable in such a way that prohibited using each subcategory as an entirely separate category

within the regression analysis. Because Books at home was originally a categorical variable, the

researchers could only employ the dummy variables as a complete set of measures within the

regression model. 

One further limitation was the lack of experience and time necessary to create a composite

variable of passive home literacy traits to be measured as a single variable within the regression

models.  The creation of composite variables for both passive home literacy traits and digital home

literacy traits would improve the validity of the variables under theoretical examination in this

analysis. 

Conclusion

In conclusion, deeper analysis of how digital literacy traits and traditional passive literacy

traits relate to reading outcomes will continue to interest parents and educators. As students interact

with more technology in their homes, educators will need to understand how Scholarly Culture is

being reproduced through e-reading at home so that classroom activities can facilitate student

reading experiences that foster strong home reading habits for all students.  Additionally, future

PISA tests should continue to adapt their data collection processes to capture the changing

educational culture of the home environment. Analyses that demonstrates the purpose and

limitations of current data collection measures will help ensure that future efforts improve our

vision of what students need to succeed at home and at school.


25
The home reading environment
Mattson & Mathew
References

Anderson, Wilson, & Fielding (1988). Growth in reading and how children spend their time

outside of school. Reading Research Quarterly, 23(3), 285 – 303.

Clark, C. (2011). Setting the baseline: The national literacy trust’s first annual survey into

reading - 2010. London: National Literacy Trust.

Evans, M. D., Kelley, J., & Sikora, J. (2014). Scholarly culture and academic performance in 42 

nations. Social Forces, 92(4), 1573-1605.

Hume, L. E., Lonigan, C. J., & McQueen, J. D. (2015). Children's literacy interest and its

relation to parents’ literacy‐promoting practices. Journal of Research in Reading, 38(2),

172-193.

Lewis-Beck, M. S., Bryman, A., & Futing Liao, T. (2004). The SAGE encyclopedia of social 

science research methods (Vols. 1-0). Thousand Oaks, CA: Sage Publications, Inc.

doi: 10.4135/9781412950589

Lindsay, J. (2010). Children’s access to print material and education-related

outcomes: Findings from a meta-analytic review. Napervile, IL: Learning Point

Associates.

Neumann, M. M. (2016). Young children's use of touch screen tablets for writing and reading at 

home: Relationships with emergent literacy. Computers & Education, 97, 61-68

OECD (2010), PISA 2009 Results: Executive Summary. 

OECD. (2018). Equity in Education: Breaking down barriers to social mobility.

https://doi.org/10.1787/9789264073234-en
26
The home reading environment
Mattson & Mathew
OECD. (2019a). PISA 2018 Assessment and Analytical Framework. pp. 21–72. Paris. PISA; OECD

Publishing.

OECD (2019b). PISA 2018 Results (Volume II): Where All Students Can Succeed. PISA. OECD

Publishing, Paris, https://doi.org/10.1787/b5fd1b8f-en.

Rouncefield, M., & Tolmie, P. (2011). Digital words: Reading and the 21st century home. In The

 connected home: The future of domestic life (pp. 133-162). Springer, London.

Schubert, F. and Becker, R. (2010). Social inequality of reading literacy: A longitudinal analysis

with cross-sectional data of PIRLS 2001and PISA 2000 utilizing the pair wise matching

procedure. Research in Social Stratification and Mobility, 29, 109-133.

Westat (2016). Sampling in PISA. First Meeting of the PISA 2018 National Project Managers.

Retrieved from: https://www.oecd.org/pisa/pisaproducts/SAMPLING-IN-PISA.pdf

Wiescholek, S., Hilkenmeier, J., Greiner, C., & Buhl, H. M. (2018). Six-year-olds' perception of

home literacy environment and its influence on children's literacy enjoyment, frequency,

and early literacy skills. Reading Psychology, 39(1),

41-68.https://doi.org/10.1016/j.ijer.2019.101512

You might also like