A Randomized Controlled Trial of Headsprout on the Reading Outcomes in Children With Autism Using Parents as Facilitators

Abstract

Children with autism spectrum disorder (ASD) are considered an at-risk population for reading delays and challenges. In recent years, there has been emerging support for the computer-assisted instruction (CAI) Headsprout with respect to reading outcomes in children with ASD. CAI, often used within classrooms, is designed using automated and carefully sequenced instruction. A randomized controlled trial was used to explore the implementation of Headsprout by parents, including a treatment package (i.e., behavioral skills training, an online support group, and a consultative model) within the home environment, with their children with ASD. A between-groups design was used to evaluate the effects of Headsprout on the reading outcomes and print motivation of a sample of 26 children with ASD. Thirteen children were in the experimental (Headsprout) group, and thirteen were in the control group receiving treatment as usual. Supplementary online supports were provided to parents based on specific learning problems identified during the intervention. Results demonstrated that participants who received Headsprout showed greater gains in reading rate, word reading, non-word reading, and target sounds and words within Headsprout than the control group. The results of print motivation demonstrated that only Headsprout-specific print material resulted in an increase in assessed preference at posttests.

Reading can be considered a pivotal educational skill for children, as it leads to the exposure to new experiences and opportunities and may be considered the starting point for a child’s transition to independent and self-managed learning. Although many neurotypical children acquire reading skills with relative ease, some populations, namely those with autism spectrum disorder (ASD), demonstrate deficits in both basic and advanced reading skills (Åsberg et al., 2008; McIntyre et al., 2017; Nally et al., 2018). The ability to read is important for children with ASD as it can lead to the future acquisition of functional skills, access to general education (Cullen et al., 2014), and leisure opportunities, such as through following directions via text and accessing media platforms. However, many children with ASD may require additional reading support and specific targeted interventions.

There has been increasing recognition of the benefits of using computer-assisted instruction (CAI) in education, with emerging research showing positive outcomes for children with disabilities, including ASD (Basil & Reyes, 2003; Pennington, 2010; Williams et al., 2002). The use of CAI may provide many benefits for this population, as it (a) is relatively easy to implement; (b) has inbuilt explicit instruction, thus reducing instructional errors; (c) has automated feedback so that following a child’s response, correction can be provided or reinforcement delivered immediately; and (d) has the potential to provide instruction that can adapt to each child’s needs. It may be argued that the extent of external, supplementary, and high-quality reading instruction is currently underresearched.

Although academic instruction typically occurs in the classroom, parents can support improved learning outcomes (Pindiprolu & Forbush, 2009; Sénéchal, 2006). The inclusion of parents as “educational partners” in supplementary intervention has proven to be an important component for achieving successful reading outcomes (Dever & Burts, 2002; Hoover-Dempsey & Sandler, 1997; Jongsma, 2001). To date, however, there is little information regarding the details of parent-facilitated reading interventions for children with ASD. There are potential benefits of home reading instruction, as it provides more opportunities for practice outside of school, thus supporting better reading outcomes. Parents are typically motivated to establish reading skills in their children, and it is a cost-effective means of instruction. Consequently, it may be necessary to investigate the greater utility of CAI for children with ASD and to assess its use outside of classrooms, as well as within.

In recent years, there has been an upsurge of investigations on the use of Headsprout (Rigney et al., 2019) CAI and its effect on reading achievement in children with ASD. Headsprout is an umbrella title for two levels of an internet-based computerized reading curriculum: Headsprout Early Reading (HER) and Headsprout Reading Comprehension (HRC). Headsprout embeds contingent reinforcement, scaffolding (i.e., building upon the child’s previously mastered reading targets), guided practice, repetition, and systematic and sequenced instruction. The program also incorporates the five essential areas of reading instruction outlined by the National Reading Panel (2000; i.e., phonemic awareness, phonics, comprehension, vocabulary, and fluency). It automatically records student responses emitted, time spent on episodes, percentage of correct responses, and a reading activity record outlining the specific skills accessed by the child. The aim is for each child to work independently on the episodes, which last on average 20 min each.

There are currently 13 peer-reviewed studies examining the effects of Headsprout with varying populations. Five of these are with neurotypical children deemed “at risk” for reading difficulties (Huffstetter et al., 2010; Pindiprolu & Forbush, 2009; Storey et al., 2017, 2019; Tyler, Hughes, Beverley, & Hastings, 2015a), one with children with intellectual disability (Tyler, Hughes, Wilson, et al., 2015b), two with children with attention-deficit/hyperactivity disorder (Clarfield & Stoner, 2005; Cullen et al., 2014), and five with children with ASD (Grindle et al., 2013; Grindle et al., 2018; Plavnick et al., 2014; Plavnick et al., 2016; Whitcomb et al., 2011). Participant numbers for the five ASD studies combined were 18. Although these studies contribute to the efficacy of the program, there is also a value in conducting studies with more participant numbers to promote generality and to examine Headsprout on a larger scale. The research with the ASD population to date has demonstrated promising outcomes, with most using single-subject research designs. Many of these authors have called for larger scale future research on Headsprout (e.g., Grindle et al., 2013; Grindle et al., 2018; Rigney et al., 2019).

With regard to teaching reading skills, CAI should be presented in an engaging format and enriched with positive reinforcement and feedback, which are central features of Headsprout. Positive experiences with reading instruction may subsequently establish sounds, words, print, and book-based materials as reinforcers for children—namely, print motivation (i.e., preference for printed books or stimuli). Print motivation is the child’s interest in and enjoyment of books and reading (Lance et al., 2010). Research has provided strong support for associations between print motivation and reading achievement (Guthrie & Wigfield, 2005; Pintrich, 2003; Taboada et al., 2009). However, to date, no research has investigated print motivation following exposure to reading interventions nor investigated print motivation in relation to children with ASD.

Grindle et al. (2013) found that four participants were able to complete 80 HER episodes, with the provision of additional and individualized teaching strategies. These provisions included verbally prompting participants to attend to the screen, manipulating motivating operations for challenging behavior due to participants’ demonstration of escape and avoidance behaviors, and using discrete-trial teaching (DTT) for specific learning problems. Extrinsic reinforcement was also delivered throughout the intervention for following directions, attending, and completing episodes.

The current study employed a randomized controlled trial (RCT) to evaluate the effects of Headsprout in comparison to a control group, which received treatment as usual. This research aimed to investigate the outcomes of the reading Headsprout CAI and a treatment package implemented by parents in their respective homes. A consultative model was employed, consisting of a collaborative process involving family. Procedures described by Grindle et al. (2013) were made readily available in response to each child’s needs, with the intervention delivered by parents to a relatively large sample of children with ASD.

The current research extends the work of Grindle et al. (2013) by (a) examining the effects of CAI (Headsprout) and a treatment package (i.e., behavioral skills training, online support group, and consultative model) on reading measures in comparison to a control group; (b) assessing potential gains on dependent measures within groups; (c) assessing whether exposure to Headsprout increases participants’ interests in reading material (print motivation); and (d) examining parents’ perceptions of the acceptability and effectiveness of Headsprout, as well as participants’ perceptions of the desirability of Headsprout.

Method

Sample and Participants

Ethical approval was obtained for this study, including parental consent and child assent (where possible). Forty-one participants were recruited from an online advertisement and screened for inclusion. Participants were included for intervention if they presented with (a) a diagnosis of ASD that met the criteria outlined in the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 2013); (b) a reading ability below age-normative scores as tested by the Neale Analysis of Reading Ability, second edition (NARA-II; Neale, 1999); (c) language skills sufficient to participate (i.e., vocal verbal skills consisting of two-word imitation) as tested by the oral expression subtest of the Wechsler Individual Achievement Test, second edition (WIAT-II; Wechsler, 2005); (d) the ability to independently navigate a cursor using a mouse (i.e., move the mouse in all directions and accurately click on various targets); and (e) the ability to respond to 10 different two-step listener directions (e.g., “Stand up and touch your nose”), based on a specific list developed by the first author, Nally, which were verbally administered to each participant. This assessment was included to ensure participants had sufficient listener skills to follow the automated instructions of Headsprout (see Grindle et al., 2013; Grindle et al., 2018). Finally, participants were assessed on the Headsprout levels, HER and HRC. These were “placement” tests composed of short passages for participants to read that indicated the level of entry in the program.

Forty-one parents of children with ASD expressed an interest in participating; however, nine children did not meet the inclusion criteria. Of the remaining 32 children who met the inclusion criteria, four parents failed to complete the target intensity of the study due to scheduling difficulties in the household, and two children (control group) were lost due to attrition. The final sample included 26 children (13 in the experimental group and 13 in the control group): 22 males and 4 females (1 female in the experimental group and 3 females in the control group) ranging in age from 4:07 to 12 years, eight months (M = 8:02, SD = 2.06). The mean language score, as measured by the Clinical Evaluation of Language Fundamentals, fourth edition (CELF-4; Semel et al., 2004), was 59.9 (SD = 24.0), and the mean abbreviated battery IQ (ABIQ) score, as measured by the Stanford–Binet Intelligence Scales, fifth edition (SB-5; Roid, 2003), was 64.0 (SD = 24.0). Participants in the experimental group had a mean age of 7 years, 5 months, a mean language score of 54.1 (SD = 20.3), and a mean ABIQ score of 64.3 (SD = 24.2). Participants in the control group had a mean age of 8 years, 8 months, a mean language score of 66.2 (SD = 26.9), and a mean ABIQ score of 63.7 (SD = 16.9). Further information on participants’ mean ages, ABIQ scores, language scores, and scores of ASD symptomatology severity is provided in Table 1.

Fig. 1.

Mean Figures (Unadjusted) for the Dependent Variables of the Multivariate Analyses of Covariance. Note. HER = Headsprout Early Reading.

Table 1.

Mean Age, Language Scores, ASD Severity, and IQ of Groups

	Both groups (n = 26)		Experimental (n = 13)		Control (n = 13)
	M (SD)	Range	M (SD)	Range	M (SD)	Range
Age	8.02 (2.06)	4.7–12.8	7.05 (1.9)	4.70–10.02	8.08 (2.0)	5.70–12.8
Language (40–145)	59.9 (24.0)	40–103	54.1 (20.3)	40–98	66.2 (26.9)	40–103
ASD severity (0–39)	21.2 (5.75)	13–34	21.3 (5.8)	15–33	21.0 (5.8)	13–34
ABIQ (40–160)	64.0 (20.8)	47–124	64.3 (24.2)	47–124	63.7 (16.9)	47–97

Note. ASD = autism spectrum disorder; ABIQ = abbreviated battery IQ.

Differences in Age, Language, Social Communication Questionnaire Scores, and ABIQ

Differences between the experimental and control groups were analyzed for the following: differences in age, language, and ABIQ (to assure equal variances between groups ahead of intervention). Independent-samples t tests were conducted to determine whether there were significant differences in age, language, ABIQ, and ASD symptomatology between the experimental and control groups. There was no significant difference in age for the experimental (M = 7.5, SD = 1.90) and control groups (M = 8.8, SD = 2.07), t(24) = −1.68, p = 0.10, indicating that equal variances in age between groups could be assumed. There were no significant differences in ABIQ scores for the experimental (M = 64.03, SD = 22.2) and control groups (M = 63.7, SD = 16.9), t(22) = .067, p = .38. There were also no statistical differences found in language between the experimental (M = 54.1, SD = 20.3) and control groups (M = 66.2, SD = 16.9), t(23) = −1.27, p = 0.66. Finally, there were no significant differences found in ASD severity between the experimental (M = 21.3, SD = 5.86) and control groups (M = 21.0, SD = 5.88), t(24) = .134, p = .761. This indicated that equal variances in ABIQ, language, and Social Communication Questionnaire (SCQ) scores between groups were assumed.

Setting

Participants resided in urban and rural communities across seven different counties in Ireland. Sessions were conducted within the participants’ homes. Participants were seated at a computer or laptop at a desk, in a quiet area of the home (usually the sitting room or dining room). No additional reading instruction was implemented within the home for either group. The control group did not receive any type of CAI in the home. Participants attended 20 different schools, with students from each school being counterbalanced across each group to minimize the effect of confounding variables. Throughout the study, parents were asked to inform the researchers if any changes occurred to the typical reading instruction at school.

Design and Participant Assignment

This study used an RCT to evaluate the effects of Headsprout CAI on the reading outcomes and print motivation of children with ASD. Stratified random sampling was used to match parent–child dyads across groups according to the level of entry in the reading intervention (i.e., HER or HRC). Once entry levels were identified, participants were randomly allocated to either the experimental group or the control group as identified by the Headsprout placement test, as it was aimed to have an even distribution of participants beginning the program at these two different levels: the experimental (n = 13) group or control (n = 13) group. Table 2 shows details of participants assigned to HER and HRC groups, as well as information on participant ages, and scores on language, ABIQ and ASD severity.

Table 2.

Age, Level on Computer-Assisted Instruction, Language, IQ, and ASD Severity Scores for Both Groups

Experimental group					Control group
Age	Level	Language	ABIQ	ASD severity	Age	Level	Language	ABIQ	ASD severity
4.4	1	90	124	20	5.7	1	62	52	13
9.3	1	40	47	22	9.11	1	40	48	22
5.2	1	40	47	18	11.2	1	50	67	22
7.4	1	44	106	15	7.4	1	40	61	19
5.2	1	40	47	19	10.2	1	40	17	23
9.4	1	40	47	26	12.8	1	40	47	16
8.0	1	40	47	26	7.9	1	56	52	25
6.10	1	60	70	29	7.1	1	40	73	22
9.2	1	100	80	20	8.5	1	98	91	27
6.9	1	42	55	18	7.0	1	50	52	17
10.7	2	40	47	28	10.3	2	91	67	23
6.8	2	111	79	94	6.9	2	103	103	16
8.9	2	88	68	76	9.4	2	100	100	13

Note. ASD = autism spectrum disorder; ABIQ = abbreviated battery IQ.

Descriptive Measures

Language

The CELF-4 (Semel et al., 2004) is an individually administered test designed to assess receptive and expressive language for individuals from 5 years to 21 years of age. The test–retest reliability scores range from .71 to .86 for subtests and .88 to .92 for composite scores based on the standardization population.

Cognitive Functioning

The ABIQ (Roid, 2003) scale of the SB-5 was used and is suitable for individuals between 2 years, to 85 years of age. The ABIQ consists of two subtests: nonverbal (object series/matrices) and verbal (vocabulary).

ASD Symptomatology

The SCQ (Rutter et al., 2003) was used to measure the degree of severity of ASD symptomatology. The SCQ total score yields an optimal cutoff score of 15 for predicting a diagnosis of ASD; however, scores can also be used to compare overall levels of ASD symptomatology.

Dependent Measures

Two standardized reading measures were used to assess four reading components: (a) word reading (WIAT-II), (b) non-word decoding (WIAT-II), (c) reading rate (NARA-II), and (d) reading comprehension (NARA-II). Direct measures were taken for Headsprout target words and sounds, and a paired-stimulus preference assessment was used to identify a ranking of print-based materials as potential reinforcers. As Headsprout target words and sounds are indexed based on episodes, placement assessments were conducted for both experimental and control participants to allocate which items in the HER direct measures would be administered to participants. All dependent measures were taken by the first author at pre- and postexposure to Headsprout. A second investigator scored the results of the measures. This second investigator was a master’s-level psychology graduate, blind to the aims of the study.

Word Reading and Non-word Decoding

Word reading and non-word decoding were assessed using the WIAT-II (Wechsler, 2005). This measure is suitable for individuals aged 6 years to 16 years, 11 months. The word reading subtest requires participants to name letters of the alphabet, identify and generate rhyming words, identify initial sounds in words, match sounds with letters and letter blends, and read aloud from a graded word list. Non-word decoding assesses the ability to apply phonetic decoding skills to non-word. The participant is asked to read aloud a list of nonsense words designed to mimic the phonetic structure of the English language.

Reading Comprehension and Reading Rate

The NARA-II (Neale, 1999) was used to assess reading comprehension and reading rate. The NARA-II provides standardized scores and reading-age estimates and is a widely used attainment and diagnostic measure that can be used for individuals from 6 years to 12 years, 11 months. Participants are required to read aloud a series of short passages and then answer a set of open-ended questions after each passage. It provides reading ages and standardized scores for two parallel sets of passages based on a fully representative UK sample (Neale, 1999).

Headsprout Words

HER comes with additional resources called Sprout Words, which are phonetically regular words that can be used for fluency building. These printable flash cards are composed of 42 targets per card and were used to assess participants’ acquisition of print words taught explicitly throughout the episodes. These were flash cards available to download from the Headsprout website. Only the Headsprout target words that corresponded to episodes that the participants would be exposed to during intervention were administered at baseline. Headsprout words were printed and laminated and then individually presented to participants on paper copies, along with the instruction “I am going to set the timer for 60 seconds. I want you to tell me as many words as you can. Ready? Go!” A stopwatch timer was then started for 1 min. This measure provided an estimate of the fluency of decoding Headsprout target words.

Headsprout Sounds

Similar to the Headsprout target words, fluency-building stimuli were used to assess approximately 42 Headsprout sounds (letter–sound correspondences), which also included some letter combinations. These sounds corresponded to specific episodes; therefore, the selection of assessments was individualized to each participant’s starting point. The testing procedure was identical to that used for Headsprout words, apart from the stimuli used. This measure provided a fluency estimate for Headsprout sounds.

Print Motivation

A paired-stimulus preference assessment (Fisher et al., 1992) was individually administered to participants to assess print motivation. Six items were included in the assessment: four preferred items (typically toys and games) selected by the participant, a book of their choice, and a Sprout Story printed in full color and presented in a book format. Sprout Stories and the book of participants’ choice were the target stimuli to assess participants’ levels of print motivation. Items were presented in twos, and paired trials were repeated until every item had been presented with one another, allowing for the creation of a ranked hierarchy of each participant’s preferred items.

General Procedure

Pretests

Assessments were administered to participants in both experimental and control groups and lasted approximately 60 min. Assessments were delivered by the first author in the participant’s home, in a quiet room with a table and chair and at least one parent present. Breaks lasting approximately 5 min were given as needed to avoid fatigue in which the participants could leave the table and have free access to preferred items. No additional corrective feedback was provided for responses during pretest sessions.

Parent Training

At the onset of the intervention, parents of children within the experimental group received training by the first author, individually within their homes, on the implementation of Headsprout. The training was composed of behavioral skills training (BST; e.g., Sarokoff & Sturmey, 2004), following a model-lead-test instructional format (Dundon et al., 2013). Training consisted of a verbal introduction to the background of Headsprout and its research, followed by an overview of the behavior-change procedures (e.g., reinforcement, scaffolding, multiple exemplars) of the program. The trainer modeled and led parents through each element of the Headsprout episode (i.e., leading them through logging in, accessing the episodes, prompting the “speak-out-loud” activities, and intermittently providing praise for on-task behavior). The vast majority of the instruction was aimed to be completed by the children as independently as possible (as it is possible to navigate through the program by correctly responding to instruction and stimuli on-screen); therefore, a large emphasis was on participants’ attending to key tasks, such as the provision of prompts for children to read out loud and remain on task. Parents were encouraged to verbally prompt their children to look at the screen (i.e., remain on task). The “test” element of the BST involved the trainer observing the parent facilitating the first episode. If corrective feedback was required, the trainer repeated the same test episode or moved to the next episode. Parents were required to score 100% in all steps of the researcher-developed fidelity-check protocol in order to independently implement the program with their children. The duration of training was approximately 40 min for each participating parent.

In addition to the training provided, a protocol manual was developed by the first author, which consisted of a three-page booklet based on the recommendations and key areas of learning difficulties outlined by Grindle et al. (2013). Parents could refer to this manual throughout the intervention, as it provided an overview of those areas that others reported difficulties with and the possible strategies needed to ameliorate difficulties as identified by Grindle et al. (2013).

Intervention

Participants were exposed to Headsprout for an average of 15–25 min (this was the typical range of the duration of episodes), three times per week, for 10 weeks. The duration of the intervention for some participants took 11 or 12 weeks due to illnesses (n = 2), family holidays (n = 2), or other unforeseen personal events (n = 1). However, all participants had similar exposure to the intervention despite interruptions. Instructional time took place after school hours or on weekends depending on family availability. Participants responded to the stimuli presented on-screen as independently as possible, with parents present nearby to assist as necessary. Parents were not required to take data on any part of the delivery, but only to note learning issues or progression failures, if any, on episodes. It should be noted that this CAI does not allow children to progress to the next episode until they reach a mastery criterion of approximately 80%. Additionally, if the children did not reach the mastery criterion, the episode would be practiced up to three times until they reached it. If following the third practice, children did not meet the criterion, parents alerted the first author and additional support was provided. Upon meeting the mastery criterion, participants marked off the episode using a progress map (a visual representation of episodes). For the duration of the intervention, all participants (both experimental and control groups) received their typical reading instruction in their respective educational settings. Typical reading instruction generally included a phonics-based curriculum and instruction on sight words. This instruction occurred daily, 5 days per week, for approximately 20 min. No participants used Headsprout at school.

Additional Treatment Package Details

Seven participants required additional supports during the intervention to support their progression through the episodes. The episodes that required additional support varied across participants. The decision to implement additional teaching strategies was made if participants did not master an episode in 1 week after a minimum of three attempts. The first author was present in each home and gave feedback throughout these additional procedures, which were based on the principles and procedures of applied behavior analysis (ABA). Five participants from the HER group and two participants in the HRC group required four main adaptations. Table 3 provides an overview of these learning difficulties and the additional supports required.

Table 4.

Additional Supports Provided to Participants and the Frequency of Sessions

Participant	HER			HRC
	DTT	TE	D	R+
1
2			✓ (3)
3	✓ (2)
4	✓ (2)
5
6	✓ (1)
7
8
9
10				✓ (1)
11
12				✓ (1)
13		✓ (1)

Note. HER = Headsprout Early Reading; HRC = Headsprout Reading Comprehension; DTT = discrete-trial training; TE = token economy systems; D = desensitization-of-stimuli procedure; R⁺ = access to reinforcement following completion.

Table 3.

Examples of Difficulties on HER and HRC That Were Facilitated Through Additional Teaching

Level	Target skill	Difficulty area	Remediation strategy
HER	Negation	Participants were required to click on a word or sound if it matched a spoken sound or, alternatively, click on the arrow if it did not (i.e., “Click on the ‘ch.’ If it is not ‘ch,’ click on the arrow.”).	As per Grindle et al.’s (2013) protocol, participants were taught this skill independent of the program via the use of mastered tacts. Next, instructions and material were modeled more closely to those used in HER.
HER	Attention to instruction	HER episodes increased in duration with progression. Participants were required to remain on task (look at the screen, listen to instruction, and make responses accordingly). One participant had difficulty maintaining attention even with verbal prompts to look at the screen or listen.	A token economy system was put in place to reinforce on-task behaviors and also allow for breaks with each exchange. A variable-interval 10-s schedule was necessary to maintain attention throughout the study (i.e., the schedule was not thinned).
HER	Desensitization of characters	Particular characters on the program were fear-eliciting stimuli for two participants (siblings). Atypical characters included aliens from outer space, some presenting with three eyes or one eye.	Coloring was identified as a preferred activity for participants. Participants were instructed to color the pictures, which were paired with edibles, social praise, and tacting the character’s name. Next, the screen was presented with the same character name tacted, along with gestures to color.
HRC	Motivation	HRC passages could be quite lengthy and required much attention. Participants began to display escape behaviors (refusals, avoidance) contingent on instruction to begin HRC.	As per Grindle et al. (2018), the Premack principle was introduced for these participants; following completion of an episode, they were allowed access to a preferred activity.

Note. HER = Headsprout Early Reading; HRC = Headsprout Reading Comprehension.

Online Support Platform

In addition to the direct contact with the first author for technical and learning support, an online parent support group was launched. This group provided an online platform for parents to discuss the progress and general technical issues in the use of Headsprout. Descriptive data were taken on problems presented on the forum. The support platform was launched on a popular social media site with customized privacy settings (i.e., nonmembers were unable to join, comment, or view the page without authorization). A total of 10 parents joined the group. The first author managed and facilitated questions and answers to the group. There were on average of 10 posts (either posted publicly or privately messaged) per member throughout the course of 10 weeks.

HER Additional Supports

Two participants required support on negation tasks (beginning on Episode 4). The supports were based on tactics employed by Grindle et al. (2013). One participant required support to sustain attention throughout the program, and a further two participants required a desensitization procedure with specific elements of the episodes. The list of strategies included DTT, token economy systems, and a desensitization-of-stimuli procedure (see Table 3).

Table 5.

Adjusted Mean Posttest Scores for Reading Dependent Measures

Dependent variable	Experimental group			Control group
	N	M	SE	N	M	SE
Word reading	13	80.7	2.35	13	73.6	2.35
Non-word reading	13	76.2	1.41	13	73.1	1.25
Comprehension	11	75.5	2.44	11	71.4	2.16
Reading rate	11	89.1	3.30	11	86.16	2.92
HER words	10	23.7	1.22	10	7.42	1.08

Note. HER = Headsprout Early Reading.

Table 6.

Examples of Parental Perceptions of Headsprout

	Comment
Parent	“I could not praise this program enough. My son has struggled with his reading (in particular, phonics) since junior infants (now 1st class), and having only completed 20-plus episodes, myself, his teacher, special needs assistant, and resource teachers have seen a massive improvement. My son still has a long way to go, but if the improvements so far are anything to go by (after just 20 episodes), I am really looking forward to seeing where he is at when he fully completes the program. My son has become much more confident in attempting to read aloud in class and is trying his best to sound out anything he gets stuck on, thanks to the skills learned with the program. The only slight criticism I would have of the program is the American accents, as he is copying them, but it’s a very minor thing and would not stop me using Headsprout.”
Parent	“I have seen a definite interest in books; J is now reading everything in her environment and paying more close attention to print in books, besides only looking at the pictures. I will keep it going as long as we have the license. She loves it, loved all the animations and would squeeze my hand during the cartoons. We intend to do more instruction through the computer. J did need prompts to speak out loud.”
Parent	“I never thought he would be able to read, but now he can read independently. It’s been a lifesaver. I would use it again.”
Participant	“Sometimes it’s hard to know what question they are talking about.”
Participant	“You get to play games before you start. I like it!”
Participant	“I liked when the episodes were finished.”

DTT

The first author visited the participants’ homes and, alongside parents, implemented DTT. First, participants were taught to follow receptive instructions with negation using familiar objects or mastered concepts. Examples included the following: “Show me the one that is NOT the horse” versus “Show me the one that IS the horse.” Initially, these instructions were kept short and succinct (e.g., “NOT blue,” wherein the participant was required to touch the red card). When the participant could receptively discriminate these instructions (90% correct responding once), they were tested again on the relevant episode. If they still could not follow the computer directions, they received additional DTT, which more closely resembled the computerized instruction (e.g., “Click on the fish. If it is NOT the fish, click on the arrow.”). During all trials, an arrow figure was placed on the table in front of the child (pointing in the same direction as the arrow on the computer screen) with a picture of an animal to the left of the arrow. The participant was taught first to touch the picture of the animal that always corresponded to the animal named in the instruction (e.g., a picture of a horse would be placed next to the arrow, and they were required to touch it).

Token Economy

One participant required extra support to maintain attention throughout episodes. This adaptation included the addition of a token economy system throughout episodes. This participant had a history of using a token economy at school and home; therefore, little additional training on its implementation was required. Tokens were delivered as reinforcement contingent on sitting appropriately, looking at the screen, and making responses on a variable interval of 10 s, irrespective of correct reading responses to the program instructions. The participant exchanged the tokens after receiving 10 check marks on the token mat and was allowed “free choice” access, which included a 2-min break away from the screen to play with toys or activities of his choice. The participant’s parent continued this support for the remainder of the study.

Desensitization Procedure

Two participants (siblings) required a desensitization procedure for the characters of the program: There were “alien” characters in the program that presented with atypical features (i.e., three eyes). Contingent on the appearance of these characters, the child elicited a fear response (i.e., crying, covering their eyes, and running away from the computer area). Within the preference assessment conducted at pretests, coloring activities were ranked as top reinforcers for these participants. Headsprout readers were downloaded from the website in black and white (stories that included the target characters). Social praise and edibles were delivered each time the participant colored the characters or part of the characters. Parents were also instructed to refer to the atypical features of the alien and to deliver phrases such as “That’s very funny” and “What color will we color his third eye?” thus exposing the participant to the target features and allowing them to be accessed “safely.” Access to other coloring materials was temporarily withheld during this conditioning procedure, which lasted 2 days, following which time there were zero instances of the previous fear response.

HRC Additional Supports

The animations and automated feedback such as “Great” were insufficient to motivate two participants on this level. These participants were found to engage in avoidance/escape behaviors (running away when it was time for Headsprout and clicking out of the program). Parents reported that this behavior was typical for these participants for similar demanding tasks. Backup reinforcers were identified for these participants, and access to these reinforcers was administered following episode completion in the absence of avoidant or escape behaviors, irrespective of reading performance. For those who needed additional supports, there were between one and three sessions required. See Table 4 for further information.

Fidelity Checks

A procedural fidelity checklist was developed by the first author using instructions from the Headsprout teacher’s manual, which included essential or critical elements (e.g., redirection to the task, appropriate reinforcement for speaking out loud and completing episodes independently). Fidelity checks were conducted remotely via video calls at least once every 2nd week by the first author with each parent–child dyad in the experimental group. During these video calls, parents positioned the camera phone so that it was possible to observe participants engaging with the reading program. Parents were provided with direct feedback regarding their monitoring of on-task behavior and prompting, as well as further advice where necessary. A total of 60 video calls were made to the experimental group across the 10-week intervention. Each parent–child dyad received a minimum of five fidelity checks. These excluded the home visits for additional training. These checks accounted for 17% of participants’ sessions, with fidelity accuracy scores of 89%. Fidelity was calculated by adding the number of correctly implemented steps by the parent, dividing this by the number of steps in total, and multiplying by 100.

Posttests

On the completion of the 10-week intervention, participants were tested again at their homes using the same dependent variables as in the pretests. Different toys were permitted in the paired-stimulus preference assessment to allow for a change of preferences over time.

Social Validity

Following the completion of posttesting, a questionnaire was administered to parents to assess the acceptability of the reading intervention. The questionnaire was adapted from the Behavior Rating Scale (Elliot & Treuting, 1991) and entailed 15 items, ranging from strongly disagree to strongly agree for scoring. Questions focused on parents’ perceptions of the program, its utility, how it fitted into the home environment, its perceived effectiveness, and parents’ likelihood of reusing the program. There was an open space provided for any additional comments. Examples of items are “The intervention was effective in improving my child’s reading,” “My child was motivated to use this intervention,” and “I would be willing to use this in the future.” A 10-item questionnaire was also devised for child participants, which included seven statements, whereby they were required to respond “Yes,” “No,” or “No response.” The questions were read aloud to child participants by the first author and recorded on the questionnaire. Questions focused on their enjoyment of Headsprout, whether they would like to continue its use, and whether they found reading easier than they did before. Examples of questionnaire items are “I enjoyed using Headsprout,” “I can read more words than before,” and “I want to keep using Headsprout.” Three open-ended questions asked participants about their favorite and least favorite part of Headsprout.

Results

During the 10-week intervention, participants in the experimental group completed a mean of 30 episodes (M = 30.1, SD = 10.4) and an average of 12 hr 6 min (M = 12.6, SD = 0.80) of exposure to Headsprout. Differences in reading outcomes between the groups were assessed for the overall efficacy of the intervention, individual analyses of reading subcomponents, and differences in preferences of print material, from pre- to posttests.

Comparison of Groups

To evaluate the effects of Headsprout, two types of analyses were undertaken. A multivariate analysis of covariance (MANCOVA) was conducted to compare the differences between the experimental and control groups on all reading dependent variables. Participants’ pretests scores were used as the covariates to control for preexisting differences for each group. Post hoc analyses were also conducted for each variable using paired-samples t tests to investigate the statistical differences in scores from pre- and posttests for both the experimental and control groups. In addition to using participant’s pretest scores as the covariate in the analyses, independent-samples t tests were conducted to assess preexisting differences in groups for all dependent variables.

Preliminary analyses were conducted to make sure no assumptions of the test were violated. The results indicated that the variables (reading rate, reading comprehension, non-word reading, HER target words) were not significant, indicating equal variances between groups at pretest. However, there was a statistically significant result for Headsprout sounds between the experimental group (M = .50, SD = 1.58) and the control group (M = 6.60, SD = 8.80), t(18) = .000, p < 0.5, indicating that equal variances at pretest could not be assumed between the two groups for target HER sounds. HER sounds were not individually investigated further within the MANCOVA. From the variables that had not violated the assumption, there were significant predicator relationships between covariate reading pretests and the scores on word reading, F(22, 6), λ = .002, η² = .964, as well as non-word reading, F(36, 6), λ = .001, η² = .978. A significant difference was found on reading rate, F(7.3, 6), λ = .022, η² = 98, and finally for HER target words, F(29, 6), λ = .001, η² = .972. The results of the MANCOVA for the combined effects of each group on reading subcomponents (word reading, non-word reading, comprehension, rate, and HER target sounds and words) were statistically significant, F(53, 6), λ = .000, η² = .015. Table 5 summarizes the adjusted mean performances for both groups on each dependent measure at pre- and posttest. The adjusted mean scores use participants’ pretests as the covariate to help reduce bias in group scores. Unadjusted means of the same dependent variables of the MANCOVA analysis are summarized in Fig. 1.

Within-Subject Analysis

Paired-samples t tests were conducted to evaluate dependent measures from pretest to posttest in each group. There were no statistically significant findings for the control group. Reported in what follows are the significant findings for t-test analyses for the experimental group. Results between the experimental and control groups are also discussed relative to print motivation.

Experimental Group

Statistically significant increases were found for tests of word reading, HER sounds and words, and preferences for Headsprout stories. There was a statistically significant increase in scores on the word reading subtest for the experimental group from pretest (M = 85.4, SD = 25.6) to posttest (M = 89.5, SD = 27.6), t(12) = −2.25, p < .05. The eta-squared statistic (.29) indicated a large effect size. There was a statistically significant increase in the measure of HER sounds from pretest (M = .05, SD = 1.58) to posttest (M = 15.80, SD = 11.1), t(9) = −4.485, p < .05. The eta-squared statistic (.74) indicated a large effect size. A statistically significant increase was also found in target HER words from pretests (M = 4.40, SD = 2.62) to posttests (M = 21.20, SD = 2.51), t(9) = −13.03, p = .000. The eta-squared statistic (.94) again indicated a large effect size. Finally, there was a statistically significant increase in preference for HER stories from pretest (M = 7.06, SD = 19.2) to posttest (M = 21.20, SD = 19.2), t(12) = −2.944, p < .05. The eta-squared statistic (.41) indicated a large effect size.

Print Motivation

As preferences for print material were unrelated to the dependent variables in the MANCOVA, a one-way analysis of covariance was used to determine whether there were significant differences between the adjusted means of the experimental and control groups. Following adjustments for pretests on preference, there was a statistically significant difference for Sprout Stories between the experimental and control groups, F(1, 22) = 7.6, p = 0.11, η² = .258. After adjusting for pretest scores on preference for books, there was not a statistically significant difference in posttest scores on book preference between the groups, F(1, 22) = 2.81, p = .153, η² = .090.

Social Validity

Ten parents and five participants completed the questionnaire. From those, results indicated that 100% of parents agreed that Headsprout was an effective reading intervention for their children and would recommend its use to other parents. All parents agreed that they liked the procedures used in Headsprout and, following exposure, reported it had provided lasting improvements in their child’s reading. Ninety percent of parents agreed that their child was motivated to use the program. Ten percent of parents agreed that there were negative side effects from using the program (HRC participant). All but one parent disclosed that they would like to continue to use the program following the cessation of the study; this parent was part of the HRC level of the intervention. The majority (80%) of children indicated that they enjoyed using the program and would like to continue to use it following the cessation of the study. Finally, 80% of child participants reported they found reading easier following the intervention. See Table 6 for some details of the comments.

Discussion

The present study demonstrated that exposure to Headsprout, supplementary teaching, and parental supports resulted in the overall improvement of reading performances between the experimental group and control group. More specifically, participants who received Headsprout showed greater gains in word reading, non-word reading, HER sounds, and HER words. The current study also demonstrated that Headsprout can be successfully implemented by parents of children with ASD, in the home environment, with the addition of adaptations and learning support. The results of this study are encouraging for future parent-facilitated instruction for this population.

Consistent with the findings of Whitcomb et al. (2011), the current study indicated that participants showed improvement in HER target words following exposure to the CAI, additional teaching, and parental supports. Therefore, parents (of children with similar profiles to the participants in this study) and educators may anticipate that, at the very least, this CAI will result in these specific outcomes. The significant findings in word reading and non-word reading suggest that this CAI is effective in the generalization of decoding strategies employed. The significant improvement in reading rate suggests that this CAI may be effective in increasing the fluency of reading connect text. The results are promising in terms of the future provision of explicit and systematic applications of the phonetic approach to teaching reading, such as the systems used in Headsprout with the ASD population. Although there were no significant gains in reading comprehension, only three participants of the experimental group were exposed to the level targeting comprehension skills; therefore, further research is warranted to examine the use of HRC with individuals with ASD. Within the current study, it is possible that the standardized reading comprehension measure used was not sensitive enough to detect smaller changes in this skill. At the time of the intervention, there were no direct comprehension measures available from the publishers of Headsprout similar to HER words and sounds. In addition, participants did not complete all of the available episodes of the HRC level. There were additional learning opportunities for answering complex, inferential questions in the latter episodes. Perhaps additional intervention time, allowing for full exposure to all of the episodes, would have impacted results. Therefore, gaps within the research persist in relation to the specific comprehension outcomes of HRC for children with ASD. Furthermore, in relation to the within-subject analysis, significant outcomes were found only for the experimental CAI group on many of the target dependent variables—namely, Headsprout sounds and words. This provides some internal validity for the CAI, as it seems to teach the targets (sounds and words) that it aims to teach.

The current study also measured whether exposure to Headsprout would increase print motivation (i.e., preference toward print material), as assessed in the paired-stimulus preference assessment. The results demonstrated that only the print material specific to the program (i.e., Sprout Stories) had an increase in preference scores at posttest. This may be a result of the learning histories with the characters, as participants regularly encountered them throughout episodes. Furthermore, the new abilities to successfully decode these books may have contributed to this result, whereas the other books selected by participants may have included words and pictures unfamiliar to the participants.

This study sought to extend the work of Grindle et al. (2013) by addressing some of the limitations outlined by the authors. Specifically, the present study targeted the absence of a control group, group size, and narrow reading subtests assessed. Just over half of the participants in the experimental group required additional strategies. These strategies considered each participants’ specific learning difficulty and transitory interests (Grindle et al., 2013). Although accommodations were established by the trainer in the present study, future research on Headsprout may target the development of a “bank” of strategies or additional instructions necessary to overcome potential learning barriers to access this program. Such a bank would be valuable and should be made available to educators of children with ASD and other disabilities. Continued research in this area will increase awareness of common learning difficulties that may arise in children with ASD, in addition to tactics that may be implemented to remedy these difficulties. It is apparent that children with ASD may require additional supports; therefore, parents or educators may not take up a curriculum that has been developed for a neurotypical population and expect seamless learning. Educators must consult the research base to assess what the common learning difficulties are in association with the CAI. Similarly, they must also be willing to address learning difficulties that are unique to each child, as per the current study.

Practical Considerations for Parents Implementing CAI

One consideration when having parents independently facilitate an intervention is the practical challenges with administering the intervention. Training parents to implement additional strategies (e.g., DTT) for their children to progress past a learning problem or certain episode was time intensive. To progress as swiftly as possible, it was necessary to have the first author present in all additional teaching sessions until participants mastered target skills. Despite this, involving parents as much as possible in the entire process, as well as in the administration of the intervention itself, was valuable, as information on such detailed supports is rarely passed on from school to home. Parents were most frequently provided with feedback suggesting they allow their child sufficient time to respond independently (i.e., 3–4 s) before administering a prompt, thereby allowing the child to respond to the CAI. Results from the present study are consistent with those of Pindiprolu and Forbush (2009), showing that parents can become active agents in their children’s learning, without sole dependence on their educational setting to teach academic skills or to resolve learning issues.

The findings from the current study may also lend themselves to some general and practical advice for parents wishing to implement Headsprout with their child with ASD in the home environment. As there were some technical issues reported throughout this study, it is recommended that there is a provision of internet access with sufficient bandwidth and speed to support quality delivery of automated instructions and reduce lag times for loading. As a lack of motivation to complete episodes was sometimes evident, parents should limit access to Headsprout to three or four times per week to ensure children maintain an interest in order to complete episodes. The use of visual stimuli was also found to be effective, as it provided the child with a visual representation of what potential reinforcer or preferred item would be available following the completion of an episode. Finally, parents should also allow their child to respond as independently as possible to the CAI, but remain close by to prompt speaking out loud, when necessary, and to provide praise throughout and extrinsic reinforcement following the completion of episodes.

Limitations

A key limitation of the study was that the control group was not receiving any additional reading intervention; instead, they were receiving their regular reading instruction as usual. Having a comparison of other CAI (similar to Storey et al., 2019) would have provided further insight into the extent of significant reading outcomes. Although this study employed an RCT, which is often considered the “gold standard” in group designs, the principal investigator was not blind to the random allocation of participants to their groups. Furthermore, the fidelity of treatment checks was lower than optimum levels, due to time constraints. The recruitment of additional research team members and assistants would be advised given the number of participants and the frequency of episodes and intervention sessions, as well as for assisting with the shortfall of blinding in the RCT to reduce possible performance bias. As demonstrated in previous research (e.g., Grindle et al., 2013; Plavnick et al., 2014; Plavnick et al., 2016), children with ASD need individualized support for reading skills, most notably comprehension skills. Although the current study offered individualized support, there were no significant gains in reading comprehension. As outlined previously, this may have been due to the small sample size of participants on the HRC level and the specific standardized comprehension measures used in this study. Further investigation on Level 2, HRC, that focuses on a larger sample size and direct comprehension measures that target the outcomes of HRC is warranted, along with a reading comprehension measure that is more sensitive to changes.

Conclusion

The present study is one of the first to evaluate a treatment package using Headsprout within an RCT design with children with ASD, as well as the inclusion of an assessment of print motivation. Findings indicated that Headsprout CAI not only is applicable to use in home environments facilitated by parents and their support but also can be conducted in conjunction with additional principles based on ABA to overcome barriers to learning. Together with the additional applications of ABA tactics, it is evident that when reading development is treated as a science, with consistency and intensity, far-reaching, socially significant outcomes can be achieved.

The presence of technology is becoming more and more of a stable entity in the lives of children, but continuing work is needed to ensure this entity develops in line with the research base. The contribution of CAI to the acquisition of early reading skills and the remediation of reading difficulties in children with ASD, as observed in this study, is very encouraging. Having clear guidelines on the specific curricula, such as Headsprout, that can supplement reading acquisition and support emerging reading development is a substantial contribution. The findings of this study indicate that it is necessary to explore and invest more time and effort into expanding learning opportunities (environments and materials) for children with ASD. The present investigation provides a promising insight into the reading supports that can be carried out by parents, to encourage more involvement, to bridge the gap between school- and home-based instruction, or to provide supplemental reading instruction more generally. The potential for parents to facilitate effective, research-based instruction in addition to instruction in schools or during summer breaks is important for supporting children with ASD who may present with reading difficulties.

Authors’ contributions

All authors contributed to the study’s conception and design. Material preparation, data collection, and analysis were performed by Amanda Nally. The first draft of the manuscript was written by Amanda Nally, and all authors contributed to the development of the final version of the manuscript. All authors read and approved the final manuscript.

Data availability

The data that support the findings of this study are stored in the National University of Ireland, Galway, repository. The data are not publicly available due to privacy restrictions (e.g., data contain information that could compromise research participant privacy/consent).

Declarations

Conflicts of interest/Competing interest

The authors declare that they have no conflict of interest.

Ethics approval

Approval was obtained from the ethics committee of the National University of Ireland, Galway. The procedures used in this study adhere to the tenets of the Declaration of Helsinki.

Consent to participate

Informed consent to participate in this study was obtained from parents.

Consent for publication

Additional informed consent was obtained from all individual participants for whom identifying information is included in this article.

Code availability

Not applicable.