Background
The American Academy of Pediatrics (AAP) recommends literacy promotion in primary care beginning as soon as possible after birth [
1]. Literacy is a major social and public health issue, the cost of low achievement estimated at over $350 billion per year in the United States, and over $1.2 trillion worldwide [
2]. As of 2015, 64% of US 4th graders scored below proficient in reading, largely unchanged from prior reports [
3], and lower still for children from minority and low-socioeconomic status (SES) households [
3,
4]. While 5 to 14% of reading difficulties have an organic cause (e.g. dyslexia) [
5], the majority are environmentally based and largely preventable, a consequence of inadequate resources, motivation and/or stimulation required to learn to read [
2]. Many children arrive at school at a substantial disadvantage in readiness, unlikely to catch up with peers as academic demands accelerate [
4]. Thus, early screening and intervention offer large potential savings in terms of productivity and health [
6‐
8].
Parents are considered to be a child’s “first and most important teachers [
9].” Cognitive stimulation in the home, exemplified by shared reading [
10‐
12], greatly influences educational and health outcomes [
13,
14]. Literacy promotion programs based in pediatric clinics (notably Reach Out and Read [
15]), preschool [
16], and home visitation [
17] share a goal of enhancing home literacy environment, a composite of quantitative and qualitative factors [
1,
18]. Quantitative factors typically include number of children’s books in the home, frequency of shared reading (e.g. days per week, minutes per day), approximate age when shared reading was initiated, and variety/type of books read [
19,
20]. Qualitative factors typically include parent and child interest in and enjoyment of reading, and verbal and social-emotional interactivity during shared reading. Given the ease of assessment via parental report, quantitative factors are most often screened and addressed [
21‐
23]. However, qualitative factors such as verbal interactivity during shared reading may be even more important and have significant impact [
14,
24], though are often overlooked.
Dialogic reading is a method of shared (usually parent-child) reading developed to promote reciprocal dialogue between a caregiver and child during story sharing [
25], advocated by Reach Out and Read and the AAP [
1,
15]. The acronym PEER is used to reflect the dialogic process [
26], as follows: 1) Prompt the child to say something about the story, 2) Evaluate what the child says, 3) Expand on what the child says, and 4) Repeat and reinforce associations. Similarly, the acronym CROWD is used to reflect evocative caregiver prompts: 1) Completion (of a sentence), 2) Recall earlier aspects of the story, 3) Open-ended questions, 4)
Wh- questions, and 5) Distancing (relate the story to the child’s experience). Behavioral evidence suggests that this qualitative aspect of shared reading may confer moderate to large cognitive and social-emotional benefits beginning in infancy [
27], especially for children from low-SES backgrounds [
28,
29]. However, there is currently no validated measure of dialogic reading or shared reading quality currently available that is feasible for clinical use.
The purpose of this study was to develop and pilot test a brief caregiver report measure of shared reading quality (DialogPR) based on a dialogic reading conceptual model. Our eight-item measure was reviewed by experts in measure design and child development, pilot tested for clarity, and then administered as an exploratory aim in 2 unrelated, MRI-based studies involving healthy, preschool-age children and their mothers: one comprised exclusively of low-SES (
n = 22) and the other of largely higher-SES (
n = 27) dyads. The validated StimQ-P measure of cognitive stimulation in the home [
30,
31], which includes a subscale of home reading practices, was administered as an external standard. Psychometric analyses, including modern-theory Rasch modeling, were performed. Our hypothesis was that the DialogPR would be feasible to administer, reliable, and valid in this combined sample, attesting to the value of a cohesive conceptual model of dialogic reading, warranting further investigation.
Discussion
Literacy is a major predictor of educational, occupational and health outcomes [
2]. While causality has not definitively been proven [
39], important drivers of reading difficulties include deficient or absent reading role models in the home and consequently impaired abilities, attitudes and routines [
40,
41]. Substantial resources are devoted to initiatives to enhance home literacy environment [
14,
15,
42], efficacy and improvement dependent on the ability to collect data efficiently and reliably. Research to date has largely relied on aspects of home literacy environment that are straightforward to assess via parental report, such as access to books and reading frequency, potentially neglecting critical behaviors such as verbal interactivity. These “dialogic” qualities may be particularly at-risk in parents lacking confidence or experience with shared reading, such as those from low-SES households [
43], and can influence reading outcomes [
24,
28].
The DialogPR instrument performed remarkably well in preliminary psychometric analysis. With only 8 items and brief administration time, this performance suggests potential value in clinical and research settings, though more expansive validation studies are needed. We attribute this strong performance to an evidence-based conceptual model of dialogic reading [
26], which guided item development. In general, DialogPR items showed low to moderate inter-item correlation and good response variability, suggesting that each contributed uniquely to the overall score, and that parents could identify their own, shared reading behavior in a range of options. The five items corresponding to specific CROWD prompts (items 3–7) each performed well, with moderate correlations between them suggesting cohesion as reading behaviors. Interestingly, items 3 and 4 were less strongly correlated with the other CROWD items than with each other, which we suspect may be attributable to completion and recall prompts seeming more abstract or unnatural to some parents at this age, compared to the other types of questioning. This may also be why items 3 and 4 were the behaviors parents reported least often. By contrast, more straightforward items 5 and 6 (open-ended and
wh- questions) were more strongly correlated with each other than with the other CROWD items, yet were among the easiest to endorse. Items 1 and 8, reflecting discussion before and after reading, respectively, and not core PEER/CROWD components, were also among the most highly correlated item pairs, yet discussion before reading was reported far less frequently.
The item performing the weakest (though still respectably), item 2, concerns frequency of pauses to answer a child’s questions, intended to reflect dialogic evaluation and expansion (Es in PEER). Possibly due to its general wording, all but 7 parents responded that they “always” or “usually” do so, the other 7 responding “sometimes,” and none responding “never.” It is intriguing that the only significant inter-item correlations involving item 2 were with items 5 (open-ended questions) and 6 (“wh-” questions). While speculative, this finding seems intuitive and suggests that this sample of mothers associated these types of relatively straightforward prompts with further dialogue. Variability in this item may have been higher if it instead asked how often specific types of responses (e.g. evaluations and expansions) were made. Such refinement in future versions of DialogPR to more explicitly assess evaluation and expansion during shared reading, may be worthwhile.
Initial evidence suggests that the most frequently endorsed items for these parents, reflecting the most common or least “difficult” shared reading behaviors, were item 2 (frequency of pauses during reading to respond to a child’s questions or comments, discussed above), item 8 (discussion after reading), item 6 (“
wh-” questions), and item 5 (open-ended questions). The least commonly endorsed, or most “difficult” shared reading behaviors were item 4 (recall prompts), item 1 (discussion before reading), and item 3 (completion prompts). Overall, this item performance seems highly intuitive, the more straightforward, or “natural” behaviors reported most often, with the possible exception of discussion before reading to generate interest. Interestingly, this finding also concurs with shared reading observations conducted for a separate MRI study involving our low-SES population (
n = 22) [
32] where
wh- prompts and open-ended questions were used almost exclusively.
Variability in item responses (with the marginal exception of item 2), and even slight skew towards lower scores, suggests that the DialogPR was not overly influenced by social desirability bias, a universal concern in parental report measures [
44]. While both DialogPR and StimQ READ scores were not technically normally distributed (Shapiro-Wilk test,
p = 0.04), strong correlation between DialogPR and this validated instrument is also reassuring in this respect. Interestingly, DialogPR total scores were negatively correlated with household income, though not with child gender or maternal education. StimQ-P READ scores were similarly negatively correlated with income, and also with maternal education. While speculative, these paradoxical findings may be attributable to a more nuanced type of reporting bias, where mothers from low-SES backgrounds may be more likely to over-report desirable reading behaviors, particularly ones that are more straightforward (in this sample, notably ‘
wh-‘questions and responding to the child), while mothers from higher-SES backgrounds may be more critical of their reading behaviors. Comparison between DialogPR scores and direct observation of shared reading in the home would be useful to quantify potential reporting effects, though this may be difficult in practice with a large sample.
This study has several important strengths. The DialogPR was developed and refined referencing an evidence-based conceptual model of dialogic reading, which afforded clarity in item content and organization. Despite a small sample size, it exhibited remarkably strong psychometric properties using advanced analytic modeling techniques. DialogPR scores were highly correlated with a validated parental report measure of reading behaviors in the home (StimQ-P READ). Administration time was brief with no concerns reported in two samples of parents from diverse backgrounds, including low-SES families who are in greatest need of effective assessment and intervention. The DialogPR addresses an important evidence gap and need for an efficient assessment of shared reading quality via caregiver report, with potential research and clinical application in programs advocating dialogic reading such as Reach Out and Read.
Our study also has a number of limitations that should be noted. Our sample population was a combination from 2 smaller studies, which can also be viewed as a strength, and efficient, innovative use of resources. The first exclusively involved 4-year-old girls from low-SES households by design, while the second was diverse in age (3 to 5) and gender from largely higher-SES employee families at an academic medical center. Together, this provided a remarkably diverse sample, showing that DialogPR may be generalizable, though a larger, inherently diverse sample would help affirm this. Girls are marginally over-represented, though it is reasonable to assume that shared reading quality should not be overly dependent on a child’s gender, especially in the preschool age range. For budgetary reasons, we were not able to conduct follow-up visits, and thus unable to assess test-retest reliability. We were unable to compare directly these findings to direct observation, and it is unclear whether reported behaviors reflect a long-term pattern, especially in oft-chaotic home environments. Most importantly, while this study offers a respectable first step, our relatively small sample size is inadequate for definitive validation, and future research with a larger sample will be needed to corroborate the findings here, including test-retest reliability and, ideally, concurrent validity with observational data. Further refinement is needed, perhaps including more explicit means to assess parental responses to the child (i.e. evaluations and expansions) during shared reading, in a parsimonious way. Overall, at this preliminary stage, the DialogPR offers a conceptually- and psychometrically-sound step toward improved, efficient insight into dialogic reading and shared reading quality in the home, important catalysts for healthy cognitive- and social-emotional development.