You are on page 1of 10

Learning and Individual Differences 99 (2022) 102211

Contents lists available at ScienceDirect

Learning and Individual Differences


journal homepage: www.elsevier.com/locate/lindif

Development of writing skills within a home-based, shared reading


intervention: Re-analyses of evidence from a randomized controlled trial
Simon Calmar Andersen a, *, Helena Skyt Nielsen b, Meredith L. Rowe c
a
Department of Political Science, TrygFonden's Centre for Child Research, Aarhus University, Bartholins alle 7, 8000 Aarhus C, Denmark
b
Department of Economics and Business, TrygFonden's Centre for Child Research, Aarhus University, Fuglesangs alle 4, 8210 Aarhus V, Denmark
c
Graduate School of Education, Harvard University, 14 Appian Way, 506 Larsen Hall, Cambridge, MA 02138, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Achievement gaps in students' literacy skills by socioeconomic status (SES) are prevalent across the globe. Theory
Written narratives suggests that reading and writing skills rely on similar knowledge and cognitive processes, yet home-based
Parent reading interventions do not typically investigate effects on writing outcomes. Using a randomized controlled
Elementary school
trial, we examined the effect of a parent focused shared-reading intervention on writing skills of 1587 second-
Experiment
grade students in Denmark. Parents were provided with books and encouraged to have their children read with
them regularly. Children's writing skills at the word, sentence, and text levels were measured at the end of the
intervention. In addition to the previously reported effects on children's reading skills, the current analysis found
that the shared-reading intervention also had effects on the children's gains in writing skills at the sentence and
text levels. Initial gaps in writing skills based on maternal education and gender of the children were reduced by
the intervention.

1. Introduction beneficial for children from less-advantaged homes.


Shared reading interventions, where parents are encouraged to read
Achievement gaps in students' literacy skills by socioeconomic status and talk about books with their children are typically focused on
(SES) are prevalent across the globe and are reported to be increasing preschool-aged children and have been found to improve children's
(Chmielewski, 2019). Results from the 2018 Program for International early language, emergent reading, and writing skills (Noble et al., 2019;
Student Assessment (PISA) of 15-year-olds indicates that in Denmark, a Zevenbergen & Whitehurst, 2003), which predict their later reading
country considered above average in reading ability and equity and skills in third grade (e.g., Sénéchal & LeFevre, 2002). Whether shared-
education, the average reading score is 501, yet there is still a 78-point reading interventions with primary school-aged children also improve
difference (0.78 SD) in scores for students from more versus less- children's writing skills – and if so which types of writing skills – is to our
advantaged homes. As points of comparison, the United States has an knowledge not yet examined. It is plausible that increased book reading
average score of 505, with an SES-gap of 99 points, whereas Hong Kong at home (by parent and child together) could help improve writing skills,
has an average score of 524 with an SES-gap of just 59 points (OECD, as reading and writing draw upon shared knowledge and cognitive
2019). While there is no international writing assessment with compa­ processes (Fitzgerald & Shanahan, 2000), are found to be strongly
rable data, there is increasing evidence that significant within-country associated with one another (Proctor et al., 2020), and studies show that
SES-gaps in reading and writing skills are already robust at the begin­ school-based reading interventions are found to have effects on writing
ning of primary school (e.g., Kim et al., 2015; Reardon, 2013). These (Graham et al., 2018).
socioeconomic disparities are thus evident early on, indicating an
important role of the home literacy environment in helping children to 1.1. Theory on knowledge shared by reading and writing processes
develop literacy skills. However, few if any studies have examined
whether interventions targeting the home literacy environment affect Fitzgerald and Shanahan (2000) suggest that reading and writing
children's writing skills – and if so whether this would be especially both draw upon four forms of shared knowledge. Three of these may

* Corresponding author at: TrygFonden's Centre for Child Research, Centre for Integrated Register-based Research, CIRRAU, Department of Economics and
Business Economics, Denmark.
E-mail addresses: sca@ps.au.dk (S.C. Andersen), hnielsen@econ.au.dk (H.S. Nielsen), meredith_rowe@gse.harvard.edu (M.L. Rowe).

https://doi.org/10.1016/j.lindif.2022.102211
Received 2 November 2021; Received in revised form 23 August 2022; Accepted 29 August 2022
Available online 14 September 2022
1041-6080/© 2022 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
S.C. Andersen et al. Learning and Individual Differences 99 (2022) 102211

affect writing in a broad sense whereas the fourth may have more spe­ socioeconomic status.
cific effects at different levels of text. First, “meta-knowledge” about As noted, studies on parent-focused shared-reading interventions are
written language encompasses general knowledge about functions and mostly conducted with parents of preschool-aged children, and less is
purposes of reading and writing, knowledge that readers and writers known about whether these types of interventions can impact primary
interact, and ideas of one's own meaning making. Second, “procedural school students' writing skills. A recent meta-analysis of shared-reading
knowledge” refers to knowing how to access, use, and generate knowl­ interventions for children aged 7 years or younger shows positive effects
edge about reading and writing and about instantiating smooth inte­ on language outcomes, including expressive vocabulary, receptive vo­
gration of various processes. Third, “domain knowledge” relates to cabulary, phonological awareness, print concepts, and expressive lan­
substance and content of a specific domain which includes both prior guage (Noble et al., 2019). Few of these studies examine writing
knowledge and content knowledge that is gained while reading and outcomes, yet there is some evidence of effects on emergent writing
writing. skills such as letter knowledge and print awareness (Zevenbergen &
The fourth type of knowledge in Fitzgerald and Shanahan's (2000) Whitehurst, 2003). This suggests that a parent-focused shared-reading
model is “knowledge about text attributes”, which may analytically be intervention with older children in primary school could also potentially
separated into knowledge at the word, sentence, and text levels. This have effects on more advanced writing outcomes.
“levels of language” model, and its distinction between word, sentence, Further, evidence from summer reading programs show that home-
and text levels is a well-established analytical tool in linguistics, even if based interventions can have a significant positive impact on chil­
these three levels must be integrated in both the reading and the writing dren's reading skills in the primary grades, and that effects tend to be
process (Abbott et al., 2010, p. 281, Ahmed et al., 2014). At the word stronger for children from lower-income families (Kim et al., 2016; Kim
level, Fitzgerald and Shanahan (2000) refer to graphophonics, or letter & Quinn, 2013). It is not clear, however, whether home-based reading
and word identification and generation, phonological awareness, and interventions might also affect children's writing skills. Writing does not
awareness of the shape of letters. We suggest that word-level knowledge receive as much attention in literacy research as reading, yet writing is
is used especially for decoding in the reading process, and for spelling in challenging for students. In the U.S. context, only 27 % of 8th graders are
the writing process. At the sentence-level is knowledge of syntax, or the considered ‘proficient” writers on a National exam (U.S. Department of
grammar of sentences and punctuations. This may benefit language Education et al., 2011). Thus, it is important to understand a variety of
comprehension when listening to stories or reading and with sentence ways to support students in their writing development, as writing skills
construction when writing. Finally, knowledge about “text format” in­ help promote learning across academic domains (Graham et al., 2020)
cludes, according to Fitzgerald and Shanahan (2000), knowledge about and is essential for many workforce roles.
the syntax of larger chunks of text (e.g., story grammars and expository
text structures), and text organization (e.g., sequence of text). We sug­ 1.3. The present study
gest that this may be especially relevant for text comprehension, when
listening or reading, and for genre composition and use of discourse In the present study, we ask whether a parent-focused shared-
when writing. reading intervention for second-graders in Denmark also has effects on
The different types of shared knowledge and competences are the student's growth in writing skills over the academic year. Further­
interlinked. According to the simple view of reading, decoding and more, we ask whether the intervention closes part of the socio-
language comprehension are prerequisites for text comprehension (e.g. demographic gaps in writing skills. If the theory that reading and
Gough & Tunmer, 1986). And text comprehension, for instance estab­ writing draw upon shared knowledge and cognitive processes (Fitzger­
lishing a pre-understanding of a book by looking at the title and the front ald & Shanahan, 2000) is correct, we could expect that when parents and
page, may help the process of decoding. Similarly, having an overall their children read books together and talk about the books to gain
understanding of the composition of the story that is about to be written, comprehension, they add to the knowledge base shared by reading and
may facilitate the writing of the first words and sentences. Indeed, many writing. In this way, reading may have spill-over effects on writing, as
types of evidence suggest that that there is a bidirectional relationship even though improved writing skills were not the aim of the reading
between reading and writing (e.g., Shanahan, 2016). activities, the increased experience with shared reading at home may
help generate knowledge that has positive effects on writing skills.
1.2. Shared reading and its relationship with writing As mentioned, one of the four types of shared knowledge that Fitz­
gerald and Shanahan (2000) argue underlie both reading and writing
A recent meta-analysis shows that reading interventions that relates to specific knowledge of text attributes which can be divided into
involved teaching students specific reading skills or having students the word, sentence, and text level. If this theory is correct, we would
read or observe others interacting with text, had positive effects on expect shared reading that focuses mostly on language and text
overall measures of writing as well as more specific measure of writing comprehension, for example parents and children talking about how to
quality and spelling (Graham et al., 2018). However, the studies rep­ understand specific sentences as well as the meaning of the whole text,
resented in the meta-analysis by Graham et al. (2018) were school-based would affect writing skills at the sentence and text level more than the
studies where teachers or researchers conducted sessions with children, word (spelling) level.
and the reading activities often also targeted writing skills, such as in Furthermore, if the shared knowledge between reading and writing
studies in which students write a text, read and give feedback on each is the driving mechanism behind spill-over effects from shared reading
other's text before they revise the text (e.g. Philippakos & MacArthur, on writing skills, we should expect that children with the weakest
2016). Furthermore, the existing studies on the effect of reading in knowledge base would be more affected by shared reading. Research
school on writing skills have some limitations. For example, 27 of the 36 suggests that families with lower socioeconomic status are found to offer
studies in Graham et al.'s (2018) meta-analysis used either a simple sparser home learning environments for language and literacy devel­
measure of spelling as the outcome, or a composite measure of writing opment, on average, compared to families of higher socioeconomic
quality. Few studies examined effects on multiple writing outcomes, and status (Golinkoff et al., 2019), so we might expect children from these
none of those compare effects at the word, sentence, and text level (yet families to benefit more from a shared reading intervention. Further,
see Knudson, 1991, which assessed some of these elements). Further­ there is a small yet significant gender gap favoring girls in literacy skills
more, sample sizes of these studies were small (3060 students in total at school entry (e.g., Chatterji, 2006; Entwisle et al., 2007). The reason
across the 36 studies, ranging from 28 to 220 students). The small for the gender difference is not entirely clear, yet the parental invest­
sample sizes not only make results uncertain, but limit the statistical ment environmental hypothesis (e.g., Baker & Milligan, 2016) has been
power needed to compare effects for students with different put forth based on evidence that parents read more with young girls than

2
S.C. Andersen et al. Learning and Individual Differences 99 (2022) 102211

boys. Thus, to the extent that boys' lower skills may be determined by skills.
their early environments, an intervention of this sort may help to close The reading advice to parents primarily targeted reading skills at the
gender gaps as well. text level: The general approach in the booklet and online video was that
In a previous study, we conducted a randomized controlled trial with parents should talk with their children before, during and after they read
parents of second graders in Denmark. The treatment group in the trial the book to “help your child understand the text”. The booklet and video
received an intervention that encouraged parents to have their child also contained some advice that targeted the sentence level by suggesting
read with them 15–20 min a day and instructions placed the most that during reading parents could talk with their child about what the
emphasis on talking about the book and ensuring comprehension. The words or phrases meant. In contrast, the booklet recommended that any
intervention adopted a growth-mindset (Dweck, 2006) approach where difficulties in decoding single words should not stand in the way of
part of the intervention involved informing parents of the large impact enjoyable reading: “If there are words your child cannot read themselves
they could have on their children's literacy skills by reading with them you can say the word's first letter/sound, let your child guess based upon
and encouraging them to read regularly. Indeed, in this study (Authors), the context of the story and the pictures, read the word aloud for your
and in prior similarly designed intervention studies (e.g., Rowe & Leech, child. Do not correct your child if they read incorrectly unless it affects
2019), intervention effects were stronger for parents who had more of a their understanding of what has been read. If the text is too difficult to
fixed than growth mindset. This suggests that informing parents about read, you can read it aloud for your child”.1
the malleability of their children's skills and their own role in the process The booklet and online video were translated into 10 languages
is a particularly useful intervention approach for parents who believe because a high share of families have immigrant background. Yet, the
abilities are more fixed than malleable with effort. Further, results supplied books were in Danish, as we wanted to make sure that immi­
showed that the intervention succeeded in increasing the reading skills grant families also had the best possible opportunities of supporting
of the students by 0.26 of a standard deviation (p < 0.01) and with re­ their children's Danish reading skills. Families were encouraged to read
sults at least as strong for children with immigrant parents or low- as often as possible and received information on how to find additional
educated mothers. Effects were strongest on text comprehension and reading material at the library, at the school, in the newspaper, etc.
decoding skills and a little smaller on language comprehension. We also However, the intervention did not request any fixed frequency. After 10
found effects on writing skills at the sentence level (Andersen & Nielsen, reading sessions, children could bring the logbooks to their school­
2016). teacher, and the class would get a sticker. The class with the most
Here, drawing on the literature presented positing specific theoret­ stickers received a prize. It was not mandatory for teachers to use the
ical relations between reading and writing skills, we use the data from logbook system; 13 of 36 classrooms in the treatment group made use of
the trial to examine children's writing skills at the word, (sentence) and the logbooks. According to the logbooks, parents read, on average, 89.2
text levels. Our specific research questions are as follows: times with their children from the time they received the materials in
October to the end of May. This corresponds to about 3 times per week in
(1) Does maternal education, immigration status and gender predict about 30 weeks, which is more than average for this age group (Libraries
second graders' writing skills at the word, sentence and text level? of the Future, 2017). Yet, it is important to note that the group that used
(2) Does a parent-focused shared-reading intervention have spill- the logbook might differ from the group that did not use it; both because
over effects to children's writing skills at the word, sentence, the teachers and classrooms were a non-random sample and because the
and text level? students were affected by the nudge (the sticker) and thus perhaps read
(3) Does the intervention affect writing skills of students with low- more than those not using the logbook. Table S1 in the Supplementary
educated mothers more than students with high-educated Material shows that boys are significantly over-represented in the group
mothers? And boys more than girls? using the logbook compared to girls, and there tend to be over-
representation of students with high-educated mothers compared to
Data from the pre-test intervention assessment is used to study pre­ low-educated mothers. Thus, the logbook data may be an upper bound
dictors of writing skills in Question 1, whereas Question 2 looks at ef­ of the reading activities. Since the intervention did not request families
fects of the shared-reading intervention on children's writing skills and to read at any specific frequency, a specific measure of treatment fidelity
Question 3 tests differential socio-demographic effects from the post-test is difficult to calculate, but these figures suggest that treatment fidelity
data. was moderate.

2. Methods 2.2. Sample

In this section we present materials and methods. First, we describe We recruited 74 regular second grade classrooms (28 schools) in a
the content of the shared-reading intervention. Then we present the major city in Denmark and included all 1587 children attending those
sample, the measures, and the statistical analysis. classrooms. We employ no exclusion criteria, but we note that children
with immigrant backgrounds are only present in regular classrooms if
2.1. Intervention they have scored above a certain threshold in a mandatory language
screening test at school start. We use this particular test to stratify the
The intervention consisted of providing parents and their eight-year- classrooms based on their average Danish language skills. We used the
old children age-appropriate books together with a booklet and access to average test score to rank-order the classrooms and create strata of four
an online video with advice for parents about how to read with their classrooms each. Within each stratum two classrooms were randomly
children. Books included both fiction and non-fiction. With a limited assigned to treatment (n = 814) and two to control (n = 773). After the
number of books provided, it would be impossible to match the reading randomization, two schools were merged reducing the number of
skill level of all students. Parents were therefore advised on how to find classrooms to 72 (without affecting the number of students). Finally,
additional books (for instance at the library) that would match the 444 students were lost to follow-up, leaving a final sample of 566 stu­
reading skill level of their child. Importantly, the booklet and the video dents assigned to treatment and 577 students assigned to control.
did not give any advice on how to support children's writing skills –
neither about spelling (word level), sentence construction (sentence
level) or text composition or specific genre characteristics (text level). 1
The primary study examined the growth mindset approach, but since the
Therefore, we could use it to test whether shared reading – without any current study on the effects on writing is more exploratory, we focus on main
explicit focus on training in writing – spills over to improve writing effects of the intervention.

3
S.C. Andersen et al. Learning and Individual Differences 99 (2022) 102211

Attrition was larger among immigrants and children of low-educated posttest. Mean number of words in the pretest was 62.7 (SD = 43.5).
mothers. However, attrition did not differ significantly between treat­ Mean in posttest was 87.8 (SD = 51.5). The writing prompt and the
ment and control group (p = 0.52). See details in Fig. 1. instructions were not pilot tested, as the task itself is very familiar to
Participation was voluntary and the project is in accordance with both teachers and students. However, in a previously published study we
American Psychological Association's Ethical Principles of Psychologists validated the writing measures (Andersen et al., 2018). Below we
and Code of Conduct. When this research was carried out, no internal summarize the five measures (details about the coding are reported in
review boards for social science existed in the country, and approval Supplementary Material).
from such institutions was not required. The project was approved by
the local government that implemented the intervention. Specifically, 2.3.2.1. Word-level: spelling. We measured the students' spelling on a 4-
the Danish Data Protection Agency approved the collection and treat­ point scale. The lowest score (0) was assigned to text that showed no or
ment of all data for the project (approval no. 2013-41-1793). little phonological awareness, and the highest score (3) was assigned if
the students to some extent mastered not only morphological but also
orthographic features of Danish. Since this was a single item, we could
2.3. Measures
not compute a standard reliability score.
2.3.1. Measures of child and parent characteristics
Information on parents was obtained from administrative data linked 2.3.2.2. Sentence-level: microstructure wNAP. We adapted the Narrative
to the sample via the civil register system. The register data provided Assessment Protocol (NAP) coding procedure (Justice et al., 2010) to
detailed information on key variables such as parents' education, written narratives in a score called wNAP. It consists of 12 items
country of origin and child gender.2 Parent's education is grouped in including: sentence structure, phrase structure, use of modifiers and tier-
compulsory education (i.e. primary and lower-secondary), vocational two verbs and nouns. For each of the 12 items we coded how frequently
training, high-school only, associate's degree, bachelor's degree and a given NAP item occurred in the narrative with instances of three oc­
master's or PhD degrees. For subgroup analyses, we categorized educa­ currences or above counting as “3” as in Justice et al. (2010). Only few
tion below the bachelor's level as “low education” and bachelor's, mas­ children scored maximum on all 12 items. The wNAP score was created
ter's and PhD degrees as “high education”. by summing the child's score on the 12 from 0 to 36. (Reliability: 0.95 in
Table 1 shows selected baseline characteristics and balance between pretest and 0.94 in posttest).
treatment and control groups (full set of characteristics in Table S2).
Students were on average 8.12 years old, 45 % of mothers had at least a 2.3.2.3. Text-level (I): narrative discourse. We measured narrative
bachelor's degree and 21 % had an immigrant background.3 Information discourse using four different items measured on 2- or 3-point scales.
on parents is missing for between 0 and 8 % of parents. The background Narrative discourse organizes meaning using language that represent
characteristics were balanced across groups except that the measure of the past events and links them in time and/or cause. Mastering language
discourse writing skills (see below) was significantly lower at baseline forms that represents a series of past events involves establishing tem­
(pre-test) in the control group than in the treatment group. Therefore, poral distance between the here and now of the events in the narrative
pre-test scores were important conditioning variables in the analyses and the communicative event itself (Labov & Waletzky, 1967; Martin &
examining treatment effects. Rose, 2008). We examined temporal distance and cohesion in the
narrative with the first item. Linguistic features that create cohesion and
2.3.2. Writing measures establish temporal and causal relations within text are important traits
Student writing data for the study were collected twice: in of written discourse. We examined how the writers used pronouns (item
August–December 2013 and in March–June 2014. The two data 2) and conjunctions (item 3) to create and maintain cohesion between
collection waves included responses from 80.3 % of the students in the phrases and sentences, and whether the events depicted in the pictures
fall (n = 1274) and 72.0 % of the students in the spring (n = 1143 in the writing prompt were represented in the texts (item 4). (Cronbach's
students). Response rates were higher for girls (74 % versus 70 % for alpha is 0.61). Finally, we constructed an additive index with all four
boys), Danes (75 % versus 60 % for immigrants), and children of high- items weighted equally. (Reliability: 0.76 in pretest and 0.68 in
educated mothers (78 % versus 68 % for low-educated). Children's posttest).
writing skills were elicited using two different writing prompts (one in
Fall and one in Spring), similar in structure and layout. Both writing 2.3.2.4. Text-level (II): narrative genre. We measured narrative genre
prompts consisted of four pictures, representing steps in a narrative using four items to assess to what extent the students can compose a text
structure. The picture sequences involved the following elements: in the shape of a narrative: orientation, evaluation, complication and
Orientation (setting the scene), complication (two events linked), eval­ resolution. These items were measured on 2- or 3-point scale. (Cron­
uation (of the characters), resolution (to the complication) (Martin & bach's alpha is 0.58). We constructed an additive index with all four
Rose, 2008). One of the pictures was replaced with a question mark items weighted equally. (Reliability: 0.80 in pretest and 0.88 in
allowing the student to make up his or her own twist of the plot at a posttest).
pivotal place in the story (see Fig. 2). The task was set to last no more
than 45 min and the teachers were instructed to hand out extra paper to 2.3.2.5. Writing composite. A factor analysis based on polychoric cor­
the students if needed. relations between all items (spelling, wNAP, discourse and genre) shows
The aim of the prompt was to elicit as elaborate narratives as possible that the items tend to load on a single factor (see Table S3 in Supple­
from the participating students. Fig. S1 in the Supplementary Material mentary Material). We created a composite writing score based on the
shows the length of the texts written by students in both the pretest and factor loadings. This may be considered as a general writing quality
equivalent to what is used in many existing studies of writing (Graham
et al., 2018; see e.g. Lee & Schallert, 2016). The general writing quality
2
The reading score gaps are sizeable for these groups in Denmark: girl-boy measure is highly correlated with wNAP (0.95–0.97; see Table S5 in
gap 0.2 SD, low-high education gap 0.5 SD and Danish-Immigrant gap Supplementary Material).
0.7–1.0 SD, according to Beuchert and Nandrup (2018) and Rangvid (2010). Table S4 in the Supplementary Material presents descriptive statis­
3
With the exception of the share of immigrants, which is twice as high as the tics on all writing measures, and Table S5 shows the correlation between
national average, the sample is representative of the Danish population of the writing measures. Before using the variables in the analyses, we
second graders. It is unsurprising that the share of immigrants is higher in a standardized them with a mean of 0 and a SD of 1 in order to make
major city compared to the country average.

4
S.C. Andersen et al. Learning and Individual Differences 99 (2022) 102211

Enrollment
Signed up (n=74 classrooms,
1,587 students)

Stratification
Classrooms placed 2 classrooms placed
in 18 strata of 4 in stratum #19

6 classrooms 6 classrooms in
merged into 4 stratum #19

Allocation

Allocated to treatment group Allocated to control group


(n=36 classes, 814 students) (n=36 classes, 773 students)

Follow-Up

Lost to writing test Lost to writing test


analysis analysis
(n=248 students) (n=196 students)

Analysis

Analysed, writing test Analysed, writing test


(n=566 students) (n=577 students)

Fig. 1. Consort diagram.

results more comparable to other studies (also see Blinded for Review for gender, immigration status, and mother's education; and uijk is an error
more detail). Fig. S2 shows the distribution of the measures. The genre term. Standard errors are clustered at classroom level to allow writing
measure showed some level of ceiling effects, which may constrain the skills to be correlated within classrooms.
estimated effect sizes for this measure. Additional analyses (not shown) For our second research question, to determine effects of the reading
show that the ceiling effect on the genre measure is more pronounced for intervention on writing outcomes compared with treatment as usual in
girls and children of high-educated mothers, and to a lesser extent, for the control group, we estimate the following model by OLS:
girls on the discourse measure. spring 2014 fall 2013
Wijk = β0 + β1 Tj + β2 Wijk + X i β3 + γk + εijk
2.4. Statistical analysis
where the post-intervention realization of writing skills has superscript
spring 2014, whereas the pre-intervention realizations have superscript
For our first research question, to examine pretest differences in
fall 2013; γ k is strata fixed effects and εijk is an error term. The parameter
socio-demographic predictors of writing skills, we estimate the
of main interest is β1, which is the intention-to-treat effect. As a
following model by Ordinary Least Squares (OLS):
robustness check we include an extended vector of covariates, Xi.
Wijk = α0 + X i α1 + uijk Finally, for the third research question, to test differential socio-
demographic intervention effects in post-test data, we use the interac­
where Wijk denotes writing skills for individual i in classroom j of stra­ tion between treatment and group indicators to test whether the
tum k; Xi is a vector of covariates including child's score on pretest,

5
S.C. Andersen et al. Learning and Individual Differences 99 (2022) 102211

Table 1 spelling, at the word level there is a substantial unfavorable gap of 21 %


Baseline characteristics and balance between treatment and control group. of a standard deviation for boys and 29 % for immigrants. For sentence-
Control group Treatment group Difference level and text-level writing skills, the gender gap doubles.
The associations between mother's education (six groups) and
Mean SD Mean SD p-value
writing skills differs across the four measures of writing skills: For
Child characteristics spelling and wNAP there is a general pattern that the higher education of
Boy 0.51 0.51 0.96
Age in 2013, years 8.12 0.36 8.12 0.35 0.84
mothers the higher skills the students have. For narrative discourse the
Immigrant 0.23 0.19 0.41 main gap is between students of mothers with only basic education or
Mother's education vocational training versus the remainder. For narrative genre, all levels
Vocational training 0.24 0.24 0.94 of education but vocational training are significantly positively associ­
High school only 0.07 0.09 0.40
ated with skills, but the pattern is mixed. For the composite measure
Associate's degree 0.04 0.04 0.47
Bachelor's degree 0.24 0.24 0.94 results are similar to wNAP. Tables S6-S10 show that when the back­
Master's or PhD degree 0.20 0.22 0.64 ground variables are added step-by-step in bivariate regressions, the
Writing measures (2013) coefficients for immigrant status and the education indicators are larger,
Word level: Spelling − 0.40 0.91 − 0.32 0.90 0.31 while the coefficients for the gender indicator are unchanged compared
Sentence level: wNAP − 0.46 0.89 − 0.31 0.85 0.16
Text level I: Discourse − 0.44 0.93 − 0.20 0.95 0.01
to the multivariate regression. Thus, it appears that the immigrant status
Text level II: Genre − 0.38 1.09 − 0.22 1.05 0.16 is strongly associated with education. The socio-demographic variables
Observations 773 814 1587 explain more of the variation in the wNAP measure (R2 = 17 %) than in
Note: All characteristics but child age and writing measures are from 2011.
the spelling measure (R2 = 7 %), which may be because the spelling
Standard errors are clustered at classroom level. Information on mother's edu­ measure is based on a single item with three values, whereas wNAP is
cation was missing for 64 individuals, whereas gender and immigration status based on 12 items ranging from 0 to 36.
was missing for less than five (exact number concealed to comply with data
protection rules).
3.2. Effects of a reading intervention on writing skills

Fig. 2. The strip used for generating written narratives.


Note: The word in the speech balloon, “MOR”, is Danish for “MOM”.

intervention (partly) closes the gap in writing skills between these Table 3 indicates that the reading intervention did not improve
groups of students. In all analyses, standard errors (SEs) are clustered at spelling skills. As found previously, it did improve writing skills at the
the classroom level to account for the hierarchical structure of the sentence level (wNAP)4 by 33 %. Furthermore, there were significant
dataset (children within classrooms). positive effects on text level writing skills, specifically in discourse and
Because of the exploratory purpose of this study, analyses are not genre skills which increased by 29 and 37 % of a standard deviation,
adjusted for multiple comparisons between subgroups. To account for respectively, as well as effects on the composite writing score which
missing information on parents, we use the missing-indicator approach, increased by 35 % of a standard deviation. Thus, a shared-reading
which (1) replaces missing values on each pretreatment covariate with a intervention can have effects on improved writing skills, and the ef­
fixed value, typically the mean, and (2) includes variables that indicates fects are selective to specific writing outcomes and not others. Table S11
whether participants have missing values on the covariate. This missing- in the Online Supplementary Material shows results of all the interaction
indicator approach gives unbiased estimates for randomized trials where models (subgroup analyses) with covariates included. These sensitivity
the distribution of covariates is balanced across treatment groups analyses show that the results are robust but slightly weaker (no longer
(Groenwold et al., 2012; see also Gomila & Clark, 2020). significant for discourse).

3. Results 3.3. Differential socio-demographic intervention effects

In this section we present the empirical analyses needed to answer The third question was whether the intervention affected the writing
the three research questions. In the first subsection, we examine whether skills of children with low-educated mothers more than it affected those
socio-demographic characteristics predict writing skills, and in the with high-educated mothers, and boys more than girls. Table 3 shows
second subsection, we investigate whether a parent reading intervention that students of low-educated (i.e. less than bachelor's level) benefitted
has effects on children's writing skills and whether those effects differ
across socio-demographic groups.
4
Results for wNAP are presented earlier by Andersen & Nielsen (2016), but
included here for comparison. Some variables have missing information. Minor
3.1. Socio-demographic predictors of writing skills changes in the inclusion of these covariates means that results presented in this
paper are slightly different from results presented in Andersen & Nielsen
Table 2 shows predictors of the four measures of writing skills. For (2016).

6
S.C. Andersen et al. Learning and Individual Differences 99 (2022) 102211

Table 2 the most from the treatment in terms of writing skills at sentence level
Socio-demographic predictors of writing skills (baseline).+ (wNAP) as well as both discourse and genre at the text level. The pattern
Word Sentence Text level Text level Composite of the coefficients is unambiguous and the gaps are sizeable (ranging
level: level: I: II: Genre factor score from 4 percentage points for spelling to 28 for genre), but only one of the
Spelling wNAP Discourse interaction effects are statistically significant. There is no difference
(1) (2) (3) (4) (5) across immigrant status. The treatment effects are equally large and
Boy − 0.214** − 0.431** − 0.379** − 0.419** − 0.460**
statistically significant for Danish and immigrant children. Table 3 also
(0.0514) (0.0459) (0.0514) (0.0576) (0.0480) shows that boys benefited greatly from the intervention and effects are
Immigrant − 0.294** − 0.319** − 0.340** − 0.273* − 0.362** as high as 40, 42 and 59 % for wNAP, Discourse and Genre. The gender
(Danish is (0.0891) (0.0745) (0.0756) (0.132) (0.0896) gap for Genre is statistically significant, while the others are not. The
reference)
findings thereby demonstrated that for some of the writing outcomes the
intervention particularly benefitted children of low-educated mothers
Mother's education level and boys, even though we note that ceiling effects in the genre measure
Compulsory (ref.) (ref.) (ref.) (ref.) (ref.)
education
may contribute to this difference.
Vocational 0.0406 0.120 0.122 0.176 0.160
training (0.0788) (0.0808) (0.0983) (0.131) (0.0992) 4. Discussion
High school 0.138 0.275** 0.337** 0.278* 0.354**
only (0.123) (0.0898) (0.122) (0.127) (0.101)
Associate's 0.235 0.404** 0.426** 0.629** 0.482**
The current study adds to the literature by showing that a parent-
degree (0.161) (0.140) (0.140) (0.178) (0.153) focused shared book reading intervention with second-graders can
Bachelor's 0.205* 0.414** 0.392** 0.474** 0.452** have spill-over effects on the children's growth in sentence-level and
degree (0.0870) (0.0827) (0.0904) (0.131) (0.0900) text-level writing skills over the academic year. Further, we found sig­
Master's or 0.405** 0.602** 0.468** 0.677** 0.629**
nificant differences in children's writing skills on all four of the writing
PhD (0.0861) (0.0868) (0.0902) (0.132) (0.0951)
degree measures in the beginning of second grade across immigration status
R-squared 0.070 0.167 0.117 0.115 0.168 and maternal education. Importantly, the intervention effects on writing
tended to be stronger for children from low maternal education house­
Note: Table shows coefficients from OLS regressions. Dependent variables are
standardized measures of writing skills (fall 2013). Indicators for missing values holds, thus reducing some of the early achievement gaps evident in
are included. N = 1274. Standard errors are clustered at classroom level. writing skills. Furthermore, we found a significant disadvantage for boys
**
p < 0.01. in writing skills, and again, the intervention effects tended to be stronger
*
p < 0.05. for boys than girls. We discuss our findings and highlight the theoretical
+
p < 0.1. and practical importance of these findings below.
The finding that a parent-focused shared-reading intervention with
second graders can have effects on writing skills is novel and important.
Theoretically, the overall pattern of results is consistent with a

Table 3
Effect of reading intervention on writing skills.
Sample Word level: Sentence level: Text level I: Text level II: Composite

Spelling wNAP Discourse Genre Factor score

(1) (2) (3) (4) (5)

All children (N = 1143) 0.11 0.33** 0.29** 0.37** 0.35**


(0.084) (0.11) (0.1) (0.11) (0.11)

Socio-demographic subgroups
Danish background (N = 942) 0.11 0.28* 0.25* 0.34** 0.30*
(0.087) (0.12) (0.11) (0.099) (0.11)
Immigrant background (N = 201) − 0.079 0.30* 0.25+ 0.35 0.31*
(0.17) (0.13) (0.15) (0.22) (0.15)
Mother high education (N = 534) 0.083 0.26* 0.26* 0.24* 0.28*
(0.092) (0.13) (0.12) (0.12) (0.13)
Mother low education (N = 569) 0.12 0.39** 0.33** 0.51** 0.41**
(0.12) (0.11) (0.13) (0.13) (0.12)
Girl (N = 578) 0.059 0.31* 0.21 0.20* 0.29*
(0.1) (0.13) (0.12) (0.088) (0.12)
Boy (N = 565) 0.18 0.40** 0.42** 0.59** 0.46**
(0.11) (0.12) (0.11) (0.16) (0.13)

Interaction significance tests


Danish_background X treatment 0.19 − 0.019 − 0.0024 − 0.013 − 0.0063
(N = 1143) (0.18) (0.14) (0.16) (0.21) (0.15)
Mother_high_education X treatment − 0.039 − 0.13 − 0.079 − 0.28+ − 0.13
(N = 1103) (0.13) (0.12) (0.14) (0.14) (0.12)
Girl X treatment − 0.12 − 0.094 − 0.21 − 0.38** − 0.17
(N = 1143) (0.12) (0.11) (0.13) (0.14) (0.12)

Note: Table shows treatment effects of reading intervention on writing skills. Dependent variables are standardized measures of writing skills (spring 2014). OLS
regressions include a pre-test measure of the relevant writing skill (fall 2013). Standard errors are clustered at classroom level.
**
p < 0.01.
*
p < 0.05.
+
p < 0.1.

7
S.C. Andersen et al. Learning and Individual Differences 99 (2022) 102211

bidirectional account of reading and writing development (e.g., Sha­ in more shared reading with girls than boys, and that encouraging
nahan, 2016), and builds on previous findings from school-based parents to read with boys resulted in a greater change in the home lit­
reading interventions that have effects on writing skills (e.g., Graham eracy environment for boys than girls. This hypothesis is supported by
et al., 2018), by showing similar effects in a parent-based shared-reading literature showing that parents and teachers report that during early
intervention. Specifically, the primary focus of the current intervention childhood girls show more interest and enjoyment in literacy activities
was comprehension and the significant effects on writing were on sen­ including reading than boys do (Baroody & Diamond, 2013), girls tend
tence and text level writing quality (not spelling). Thus, for the current to have better self-perceptions of themselves as readers than boys (e.g.,
findings, the increase in shared reading experience itself, may have had Lynch, 2002), and girls consistently out-perform boys on standardized
a direct effect on writing skills through providing children with measures of reading achievement (see Lietz, 2006 for a meta-analysis).
increased opportunity to engage in the cognitive process of extracting While we cannot be certain that the gender effects are a result of envi­
and constructing meaning through interaction with written language. ronmental rather than biological differences, environmental effects of
More specifically, the emphasis on comprehension encouraged a focus shared reading on reading (and now writing) achievement are clear.
on meta-knowledge (e.g., Fitzgerald & Shanahan, 2000), and on dis­ Therefore, parent-focused interventions might be particularly important
cussing what they were reading, potentially leading children to formu­ for parents of boys, who may be less interested in or confident about
late sentences to explain their understanding of text, a practice that may reading at home compared to their female peers. The extent to which
have supported children's ability to draw on procedural knowledge to children enjoy reading is found to explain a significant amount of the
write brief stories as in the context of our writing assessment. gender difference in literacy achievement (Chiu & McBride-Chang,
The null finding on spelling skills is slightly more perplexing since 2006), therefore encouraging parents to read with boys might boost
the reading intervention did affect the students' decoding skills (blinded their enjoyment and skills, ultimately reducing gender gaps in reading
for review) and shared knowledge theory would predict that if there was and writing. We encourage further research that tests out some of these
an effect on decoding there might also be a spill-over effect on spelling. potential mechanisms of our education and gender findings.
However, the results in the meta-analysis by Graham et al. (2018) imply There are some limitations to the study worth mentioning, several of
that an effect on spelling would have been more likely if the intervention which also pertain to the measures used. First, while students and par­
itself had a more multi-faceted focus and a focus on phonological ents were reading both fiction and non-fiction at home during the
awareness or phonics (Graham et al., 2018). Thus, while a strength of intervention, both the writing prompt test and the text-level measures of
the current study is that we went beyond global measures of writing to writing skills were assessments of narrative genre writing. Indeed, in the
examine effects at different levels, one limitation is that we cannot early grades, most writing activities in school focus on narrative writing
determine why there was an effect on decoding in the reading analysis rather than expository writing (Cutler & Graham, 2008), thus we chose
(blinded for review), but not on spelling in this writing analysis. It may measures the students were familiar with and could be successful on but
be that the nature of the treatment in its focus on text level skills and would also elicit variation. However, the genre measure in particular
comprehension provided enough practice with text to improve decoding showed ceiling effects and did not capture variation in the sample as
skills, yet not to carry over to improving spelling which could arguably well as the other writing measures, suggesting that these eight-year-olds,
require more explicit emphasis on phonics (e.g., Graham et al., 2018). It especially girls and children of high-educated mothers, had already
may also be that the spelling measure was not necessarily sensitive to the mastered the necessary components of narrative genre. As a conse­
intervention. Recall, this was not a spelling test per-se, but a spelling quence, this may be part of the explanation for why the reading inter­
accuracy measure derived from the words children chose to spell on vention reduced the gender and the high/low-education gap in writing
their own in their writing, which may have introduced some noise into as measured by narrative genre, as boys and the children of low-
the measure, as children could avoid using words they do not know how educated mothers started out lower on this measure. Thus, this partic­
to spell. ular result must therefore be interpreted with caution. Further, re­
There are also clear practical implications from this study worth lationships between reading and writing may be stronger within versus
discussing. The fact that the parent-focused reading intervention not across genre, and other genres are indeed found to elicit more sophis­
only showed main effects on writing skills, but reduced gaps in writing ticated syntax (Beers & Nagy, 2011). Future work could include
skills by maternal education and gender is noteworthy. Indeed, in this different types of writing prompts and examine relations between
sample we replicated previous research showing gaps in writing skills reading and writing within versus across genre. Another potential lim­
across socioeconomic status and gender (e.g., Kim et al., 2015; Reilly itation is that we only included measures of writing composition at
et al., 2019). Previous research has identified the home literacy envi­ different levels and did not include measures of other aspects of writing,
ronment as one potential source of these gaps (e.g., Baker & Milligan, such as handwriting, which are also developing during this time.
2016; Golinkoff et al., 2019). The current parent-focused intervention Future studies should be designed to specifically test for different
that encouraged parents to have their children read with them on a effects across subgroups. We note that even if attrition did not differ
regular basis, may have potentially changed the home environment across treatment and control groups, more immigrants and children of
more for lower-educated families and for families with boys resulting in low-educated mothers were lost to follow-up. To improve the exami­
larger intervention effects. We can speculate as to why this might be the nation of moderators, and with respect to children's reading levels,
case. In terms of education level, it is plausible that the families with future research should aim at reducing drop-out, collect more data on
lower education, on average, were not aware that shared reading with the implementation of the intervention, and aim at reducing ceiling
children could be so influential in their children's skills and that effects in the writing measures.
providing them with this parenting knowledge had a stronger influence While we did not collect a measure of treatment dosage, or how
on their home literacy activities than for higher-educated parents who much shared book-reading the families actually engaged in, we did
might already be engaging in shared reading more regularly. Indeed, observe logbook usage in treated classrooms. If we take this as a proxy
parenting knowledge of child development is found to mediate effects for treatment dosage, it appears that treatment dosage was significantly
between parental education and the home language and literacy envi­ higher in families with boys, who also show higher treatment effects.
ronment (Rowe, 2008; Rowe et al., 2016). If this mechanism is correct, Thus, it could be that the greater effect of the intervention for boys was
our results emphasize the importance of parent-focused interventions in due to differences in treatment, however we do not have full data to
addition to classroom interventions, as even beyond early childhood, examine treatment effects.
shared reading with parents can positively impact reading and writing Despite these important limitations, the results have several practical
development. implications. First, practitioners should help parents understand that
The finding about gender implies that parents might typically engage reading together at home helps promote their children's reading and

8
S.C. Andersen et al. Learning and Individual Differences 99 (2022) 102211

writing skills. Conveying this information to lower-educated parents and year in primary school children. Contemporary Educational Psychology, 55, 129–138.
https://doi.org/10.1016/j.cedpsych.2018.09.005
to parents of boys in particular and providing them with opportunities to
Andersen, S. C., & Nielsen, H. S. (2016). Reading intervention with a growth mindset
read together at home can help reduce income and gender-based approach improves children’s skills. Proceedings of the National Academy of Sciences,
achievement gaps in literacy skills broadly speaking. These findings 113(43), 12111–12113. https://doi.org/10.1073/pnas.1607946113
should also be relevant for promoting summer reading programs, as Baker, M., & Milligan, K. (2016). Boy-girl differences in parental investments: Evidence
from three countries. Journal of Human Capital, 10(4), 399–441.
children are found to lose writing skills during the summer, in addition Baroody, A. E., & Diamond, K. E. (2013). Measures of preschool children's interest and
to reading skills (Meyer et al., 2020), and the current findings suggest engagement in literacy activities: Examining gender differences and construct
that summer reading interventions (e.g., Kim & Quinn, 2013) could also dimensions. Early Childhood Research Quarterly, 28(2), 291–301.
Beers, S. F., & Nagy, W. E. (2011). Writing development in four genres from grades three
have effects on children's writing skills. Finally, key components of the to seven: Syntactic complexity and genre differentiation. Reading and Writing, 24(2),
current intervention included providing parents with information about 183–202.
how to engage in shared reading with their children, providing lower- Beuchert, L. V., & Nandrup, A. B. (2018). The Danish national tests at a glance.
DanishEconomic Journal, 1.
educated parents with the means to access age-appropriate books for Chatterji, M. (2006). Reading achievement gaps, correlates, and moderators of early
this purpose, and encouraging a growth-mindset in parents towards reading achievement: Evidence from the Early Childhood Longitudinal Study (ECLS)
their children's literacy. This approach is cost-effective and may be a kindergarten to first grade sample. Journal of Educational Psychology, 98(3), 489.
Chiu, M. M., & McBride-Chang, C. (2006). Gender, context, and reading: A comparison of
nice complement to school-based interventions. Future research should students in 43 countries. Scientific Studies of Reading, 10(4), 331–362.
examine the longer-term effects of parent-focused shared reading Chmielewski, A. K. (2019). The global increase in the socioeconomic achievement gap,
interventions. 1964 to 2015. American Sociological Review, 84(3), 517–544.
Cutler, L., & Graham, S. (2008). Primary grade writing instruction: A national survey.
Journal of Educational Psychology, 100(4), 907–919.
5. Conclusions Dweck, C. S. (2006). Mindset: The new psychology of success. New York: Random House.
Entwisle, D. R., Alexander, K. L., & Olson, L. S. (2007). Early schooling: The handicap of
being poor and male. Sociology of Education, 80(2), 114–138.
The current findings that a parent-focused reading intervention with Fitzgerald, J., & Shanahan, T. (2000). Reading and writing relations and their
second graders also has effects on growth in writing for the students over development. Educational Psychologist, 35(1), 39–50.
the course of the year is new and has several important theoretical and Golinkoff, R., Hoff, E., Rowe, M. L., Tamis-LeMonda, C., & Hirsh-Pasek, K. (2019).
Language matters: Denying the existence of the 30-million-word gap has serious
practical implications. Theoretically, we add causal evidence to a
consequences. Child Development, 90(3), 985–992.
growing literature showing relations between reading and writing Gomila, R., & Clark, C. S. (2020). Missing data in experiments: Challenges and solutions.
development. More specifically, our nuanced measures of writing Psychological Methods. https://doi.org/10.1037/met0000361. Advance online
publication (October 12, 2020).
enabled us to see that this particular intervention had effects, not on the
Gough, P. B., & Tunmer, W. E. (1986). Decoding, reading, and reading disability.
word level of writing (spelling), but on the sentence and text level Remedial and Special Education, 7(1), 6–10.
measures of writing, suggesting that for a reading intervention to have Graham, S., Kiuhara, S. A., & MacKay, M. (2020). The effects of writing on learning in
spill-over effects on spelling as well there may need to be more of a focus science, social studies, and mathematics: A meta-analysis. Review of Educational
Research, 90(2), 179–226.
on phonology in the intervention. More practically, these results expand Graham, S., Liu, X., Bartlett, B., Ng, C., Harris, K. R., Aitken, A.Talukdar, J., … (2018).
our horizon of successful interventions beyond the classroom environ­ Reading for writing: A meta-analysis of the impact of reading interventions on
ment (e.g., Graham et al., 2018) to the homes of primary school chil­ writing. Review of Educational Research, 88(2), 243–284.
Groenwold, R. H. H., White, I. R., Donders, A. R. T., Carpenter, J. R., Altman, D. G., &
dren. This is important, because often parent-focused interventions are Moons, K. G. M. (2012). Missing covariate data in clinical research: When and when
limited to the preschool years or to summer only. Finally, we show that a not to use the missing-indicator method for analysis. Canadian Medical Association
parent-based reading intervention can reduce education- and gender- Journal, 184(11), 1265–1269.
Justice, L. M., Bowles, R., Pence, K., & Gosse, C. (2010). A scalable tool for assessing
gaps in writing skills that are evident early on in schooling. This is children's language abilities within a narrative context: The NAP (Narrative
important because these gaps tend to grow over time, not shrink, Assessment Protocol). Early Childhood Research Quarterly, 25(2), 218–234.
therefore policies and programs that involve parents early and often Kim, J. S., Guryan, J., White, T. G., Quinn, D. M., Capotosto, L., & Kingston, H. C. (2016).
Delayed effects of a low-cost and large-scale summer Reading intervention on
show promise as a profound way to improve not only reading but
elementary school children's reading comprehension. Journal of Research on
writing skills for all children. Educational Effectiveness, 9(sup1), 1–22. https://doi.org/10.1080/
19345747.2016.1164780
Kim, J. S., & Quinn, D. M. (2013). The effects of summer reading on low-income
Declaration of competing interest children's literacy achievement from kindergarten to grade 8: A meta-analysis of
classroom and home interventions. Review of Educational Research, 83(3), 386–431.
Kim, Y. S., Puranik, C., & Otaiba, S. A. (2015). Developmental trajectories of writing
The authors declare no conflicting interests. skills in first grade: Examining the effects of SES and language and/or speech
impairments. The Elementary School Journal, 115(4), 593–613.
Knudson, R. E. (1991). Effects of instructional strategies, grade, and sex on students'
Acknowledgements
persuasive writing. The Journal of Experimental Education, 59(2), 141–152.
Labov, W., & Waletzky, J. (1967). Narrative analysis. In J. Helm (Ed.), Essays on the verbal
This project was financially supported by TrygFonden. TrygFonden and visual arts. Seattle: University of Washington Press.
Lee, J., & Schallert, D. L. (2016). Exploring the reading-writing connection: A yearlong
had no involvement in the decisions relating to the research reported
classroom-based experimental study of middle school students developing literacy in
here. a new language. Reading Research Quarterly, 51(2), 143–164. https://doi.org/
10.1002/rrq.132
Libraries of the Future. (2017). Children’s reading – A quantitative investigation of
Appendix A. Supplementary material children&#x0027;s reading and media use after school. [In Danish: Børns Læsning 2017
– en kvantitativ undersøgelse af børns læse- og medievaner i fritiden]. The Danish Think
Supplementary material to this article can be found online at https Tank – Libraries of the Future. Copenhagen, Den: Tænketanken Fremtidens
Biblioteker.
://doi.org/10.1016/j.lindif.2022.102211.
Lietz, P. (2006). A meta-analysis of gender differences in reading achievement at the
secondary school level. Studies in Educational Evaluation, 32(4), 317–344.
References Lynch, J. (2002). Parents' self-efficacy beliefs, parents' gender, children's reader self-
perceptions, reading achievement and gender. Journal of Research in Reading, 25(1),
54–67.
Abbott, R. D., Berninger, V. W., & Fayol, M. (2010). Longitudinal relationships of levels
Martin, J. R., & Rose, D. (2008). Genre relations: Mapping culture. London: Equinox.
of language in writing and between writing and reading in grades 1 to 7. Journal of
Meyer, F., Yao, E. S., & Meissel, K. (2020). The summer learning effect in writing in New
Educational Psychology, 102(2), 281–298. https://doi.org/10.1037/a0019318
Zealand. Reading and Writing, 33, 1183–1210. https://doi.org/10.1007/s11145-019-
Ahmed, Y., Wagner, R. K., & Lopez, D. (2014). Developmental relations between reading
10003-6
and writing at the word, sentence, and text levels: A latent change score analysis.
Noble, C., Sala, G., Peter, M., Lingwood, J., Rowland, C., Gobet, F., & Pine, J. (2019). The
Journal of Educational Psychology, 106(2), 419.
impact of shared book reading on children's language skills: A meta-analysis.
Andersen, S. C., Christensen, M. V., Nielsen, H. S., Thomasen, M. K., Østerbye, T., &
Educational Research Review, 28, Article 100290.
Rowe, M. L. (2018). How reading and writing support each other across a school

9
S.C. Andersen et al. Learning and Individual Differences 99 (2022) 102211

OECD. (2019). PISA 2018 results (Volume II): Where all students can succeed, PISA. Paris: Rowe, M. L., Denmark, N., Jones Harden, B., & Stapleton, L. (2016). The role of parent
OECD Publishing. https://doi.org/10.1787/b5fd1b8f-en education and parenting knowledge in children's language and literacy skills among
Philippakos, Z. A., & MacArthur, C. A. (2016). The effects of giving feedback on the White, Black, and Latino American families. Infant and Child Development, 25,
persuasive writing of fourth- and fifth-grade students. Reading Research Quarterly, 51 198–220.
(4), 419–433. https://doi.org/10.1002/rrq.149 Rowe, M. L., & Leech, K. A. (2019). A parent intervention with a growth mindset
Proctor, C. P., Daley, S. G., Xu, Y., Graham, S., Li, Z., & Hall, T. E. (2020). Shared approach improves children's early gesture and vocabulary development.
knowledge between reading and writing among middle school adolescent readers. Developmental Science, 22(4), Article e12792.
The Elementary School Journal, 120(3), 507–527. Sénéchal, M., & LeFevre, J. A. (2002). Parental involvement in the development of
Rangvid, B. S. (2010). Source country differences in test score gaps: Evidence from children's reading skill: A five-year longitudinal study. Child Development, 73(2),
Denmark. Education Economics, 18(3), 269–295. 445–460.
Reardon, S. F. (2013). The widening income achievement gap. Educational Leadership, 70 Shanahan, T. (2016). Relationships between reading and writing development. In
(8), 10–16. C. MacArthur, S. Graham, & J. Fitzgerald (Eds.), Handbook of writing research (2nd
Reilly, D., Neumann, D. L., & Andrews, G. (2019). Gender differences in reading and ed., pp. 194–207). New York, NY: Guilford.
writing achievement: Evidence from the National Assessment of Educational U.S. Department of Education, Institute of Education Sciences, National Center for
Progress (NAEP). American Psychologist, 74(4), 445. Education Statistics, National Assessment of Educational Progress (NAEP). (2011).
Rowe, M. L. (2008). Child-directed speech: Relation to socioeconomic status, knowledge Writing assessment.
of child development, and child vocabulary skill. Journal of Child Language, 35, Zevenbergen, A. A., & Whitehurst, G. J. (2003). Dialogic reading: A shared picture book
185–205. reading intervention for preschoolers. In On reading books to children: Parents and
teachers (pp. 177–200).

10

You might also like