Introduction

Computational Thinking

Although, there has been a substantial rise in scientific interest in computational thinking (CT) since 2013 (Ezeamuzie & Leung, 2021; Hsu et al., 2018; Ilic et al., 2018), CT is not a new concept. In fact, within computer sciences, the concept dates back to the 1950s and 1960s. The term CT was first used by Papert in the 1980s (Nicoletti & Suemasu, 2021; Tang et al., 2020) and is most frequently associated with Wing´s seminal article on the subject in 2006 (Ezeamuzie & Leung, 2021; Kakavas & Ugolini, 2019). Wing (2006) referred to CT as “a universally applicable attitude and skill set everyone, not just computer scientists, would be eager to learn and use” (p. 33). She defined it as the “thought processes involved in formulating a problem and expressing its solution(s) in such a way that a computer—human or machine—can effectively carry out” (Wing, 2017, p. 8) and considered it to be just as important as reading, writing and arithmetic (Wing, 2006). Papert followed a more constructionist approach in fostering CT in children by emphasizing that social and affective involvement is as crucial as technical content (Lodi & Martini, 2021). Although there is no universal consensus in the literature on how CT and its key components are to be defined (Fagerlund et al., 2021; Ilic et al., 2018; Kakavas & Ugolini, 2019; Lockwood & Mooney, 2018; Tang et al., 2020; Zhang & Nouri, 2019), there is general agreement that the practice centers on the concepts involved in problem solving, particularly as applied by computer scientists (Ezeamuzie & Leung, 2021). Most frequently, the framework of Brennan and Resnick (2012) is used to conceptualize CT skills (Kakavas & Ugolini, 2019; Ogegbo & Ramnarain, 2021; Tang et al., 2020; Zhang & Nouri, 2019) and covers three key dimensions: computational concepts (sequences, loops, parallelism, events, conditionals, operators and data), computational practices (iteration, testing and debugging, reusing and remixing, abstracting and modularizing) and computational perspectives (expressing, connecting and questioning). Moreover, most of the (systematic) reviews on the topic refer to CT as a combination of skills, such as problem decomposition, abstraction, algorithmic thinking (Ezeamuzie & Leung, 2021; Kakavas & Ugolini, 2019; Zhang & Nouri, 2019), pattern matching (Grover & Pea, 2013; Lockwood & Mooney, 2018), sequencing (Ezeamuzie & Leung, 2021), identification of control structures, testing and verification, and parallel thinking (Kakavas & Ugolini, 2019; Zhang & Nouri, 2019).

CT is seen as an essential 21st century skill (Haseski et al., 2018; Lockwood & Mooney, 2018; Rich et al., 2019; Zhang & Nouri, 2019) that should be taught to students of all ages to enable them to acquire the skills needed for various applications in daily life (Hsu et al., 2018; Lockwood & Mooney, 2018; Wing, 2006). In recent years, many reviews have reported on CT and its important role in primary education (Bakala et al., 2021; Bati, 2021; Fagerlund et al., 2021; Kakavas & Ugolini, 2019; Rich et al., 2019; Zhang & Nouri, 2019), in secondary education (Lockwood & Mooney, 2018), and in K-12 education as a whole (Grover & Pea, 2013; Lye & Koh, 2014; Merino-Armero et al., 2021; Sun et al., 2021a, b). The development of CT has been examined at all levels, from early childhood to university-level education (Ezeamuzie & Leung, 2021; Ilic et al., 2018). However, the main focus of research has been on primary education (Bati, 2021; Ilic et al., 2018; Kakavas & Ugolini, 2019; Tang et al., 2020).

There is now growing global interest in implementing CT in classroom education and in integrating CT into the K-12 curriculum (Bakala et al., 2021; Hsu et al., 2018; Lockwood & Mooney, 2018; Ogegbo & Ramnarain, 2021; Zhang & Nouri, 2019). However, considerable diversity exists across regions and countries concerning what this means in practice (Ezeamuzie & Leung, 2021; Hsu et al., 2018; Ilic et al., 2018; Lockwood & Mooney, 2018; Taslibeyaz et al., 2020). The existing differences in educational systems and cultures, as well as the absence of a universally recognized definition of CT do not make matters any easier (Hsu et al., 2018; Lockwood & Mooney, 2018; Zhang & Nouri, 2019). It is generally recognized that CT is an important cognitive ability which is applicable to problem solving in many areas (Ezeamuzie & Leung, 2021; Hsu et al., 2018; Lodi & Martini, 2021), and that the methods and concepts of computer science can be helpful in developing such an ability (Buitrago Flórez et al., 2017; Hsu et al., 2018; Lodi & Martini, 2021). While CT depends to a large extent on concepts applied in programming (Ezeamuzie & Leung, 2021; Kakavas & Ugolini, 2019; Zhang & Nouri, 2019), the transferal of such concepts to other learning domains is far more extensive (Buitrago Flórez et al., 2017; Ezeamuzie & Leung, 2021; Merino-Armero et al., 2021).

CT is applicable to numerous aspects of daily life (Tsortanidou et al., 2021) and is also relevant in social problem solving. D’Zurilla et al. (2004) refer to social problem solving as “the self-directed cognitive-behavioral process by which an individual, couple, or group attempts to identify or discover effective solutions for a specific problem encountered in everyday living” (p. 12). As CT-related activities may facilitate collaborative problem-solving, they also serve to promote communication and the sharing of ideas and knowledge (Bakala et al., 2021; Herro et al., 2021; Tsortanidou et al., 2021). It is therefore no surprise that CT has been shown to be linked to an increase in social interactions and the development of social-emotional skills.

Interventions to foster CT skills in K-12 students are most often implemented in the field of programming education (Hsu et al., 2018; Taslibeyaz et al., 2020) as this is considered the most effective means of promoting student CT-related skills (Merino-Armero et al., 2021; Sun et al., 2021a, b). The strategies employed mainly focus on plugged activities based on different types of programming tasks (Kakavas & Ugolini, 2019; Moreno-León et al., 2018). Plugged activities concern visual programming environments (arrow-based or block-based visual environments, e.g., Scratch, Alice, Kodu), textual programming languages (e.g., Java, Python, Logo), and technologies connected to the physical world, for example, in the form of programmable toys, robots or boards (Buitrago Flórez et al., 2017; Ching et al., 2018; Kanika et al., 2020; Moreno-León et al., 2018). Research reveals that the most common approaches to student CT-skill acquisition concern project-based, problem-based, collaborative and game-based learning (Hsu et al., 2018; Kanika et al., 2020). Unplugged activities relating to CT, i.e. those not linked to any digital devices, are used and studied less frequently (Kakavas & Ugolini, 2019; Moreno-León et al., 2018).

Students with ASD and ADHD

Although CT is considered an essential skill for learners of all ages and should be taught to everyone (Ching et al., 2018; Wing, 2008), little is known about teaching CT to students with disabilities (Gribble et al., 2020; Koushik & Kane, 2019; Lechelt et al., 2018; Prado et al., 2021). As CT promotes problem-solving skills and is applicable in many different contexts, it is particularly valuable for such students. In particular, for children with autism spectrum disorder (ASD) and/or attention deficit/hyperactivity disorder (ADHD), CT could have a beneficial impact on the development of social skills. ASD is a neurodevelopmental disorder characterized by deficiencies in social communication, interaction and social relationships that can be observed in different contexts. It may also be characterized by restricted and repetitive patterns of behaviour, adherence to specific routines, and to interests exhibiting abnormal intensity or focus. Children and adolescents with ASD exhibit hyper- or hypo-reactivity to sensory input, or a special interest in sensory impressions. It is referred to as a 'spectrum disorder' since it manifests itself differently depending on its severity (American Psychiatric Association (APA), 2013). With regard to comorbidity, it should be noted that ASD and ADHD often co-occur (Antshel & Russo, 2019; Dellapiazza et al., 2021). Thus, individuals diagnosed with ASD are frequently affected by ADHD (Rong et al., 2021). The core symptoms of ADHD are inattention, hyperactivity and impulsivity. These result in specific behavioral characteristics such as restlessness, fidgetiness, difficulties concentrating and an inability to focus attention (APA, 2013). Due to their impulsive behaviour children with ADHD have difficulties developing and maintaining social relationships and thus often experience peer neglect and social rejection (Klicpera et al., 2019; APA, 2013). However, difficulties in social communication and relationships are more a consequence of the restlessness and impulsivity associated with ADHD and are not considered to be a key component of the disorder (Klicpera et al., 2019; Mahendiran et al., 2019).

Both ASD and ADHD are among the most commonly diagnosed neurodevelopmental disorders in childhood (Harkins et al., 2021). Although research shows that ASD and ADHD are mainly diagnosed in males (APA, 2013; Mahendiran et al., 2019), there is growing evidence that manifestations of ASD in females are not adequately reflected in current diagnostic procedures (Ratto et al., 2018). This might be due to a camouflaging effect since females seem to be better at masking socio-communicative impairments than their male peers (Cook et al., 2018; Milner et al., 2019; Tubío‑Fungueiriño et al., 2021; Ratto et al., 2018). Furthermore, girls and women with ASD report that their experiences are different from those of boys and men with ASD in areas such as getting a diagnosis, friendships with peers, or fitting in and being catered for in autism programs which are often mainly targeting boys and their interests (Cook et al., 2018; Cridland et al., 2014; Milner et al., 2019). Similar findings are evident in research on ADHD, as the disorder often manifests itself differently in girls (e.g., higher level of inattention) and gender differences are often not identified in the diagnostic process. As with ASD, this could be explained by coping strategies girls apply to hide their ADHD symptoms (Klefsjö et al., 2021).

ASD and ADHD have several behavioral symptoms in common and are associated with impaired communication, poor social relationships, and the inability to recognize social cues. This then leads to problems at home, at school, and in other social contexts (Cervantes et al., 2013; Harkins et al., 2021; Salley et al., 2015). As CT entails an interactive, well-structured, step-by-step approach, and fosters self-management in problem solving, CT-skills can be helpful in promoting the inclusion of children diagnosed with ASD or ADHD.

Numerous (systematic) reviews and meta-analyses exist concerning the learning (e.g., Hsu et al., 2018; Zhang & Nouri, 2019), teaching (e.g., Hsu et al., 2018; Lockwood & Mooney, 2018) and assessment (Tang et al., 2020) of CT in educational contexts. There are also several studies on the use of computer-based interventions in fostering specific skills such as social emotional skills (e.g., Tang et al., 2019) or fostering STEM skills (e.g., Ehsan et al., 2018) in children with ASD and/or ADHD. However, very few studies have investigated how CT and/or programming may explicitly be used to help students with ASD and/or ADHD or how CT is related to other areas (e.g., the development of social-emotional skills) among such students.

Research Gap and Research Questions

To address this gap in the research, we conducted a review of existing studies in order to make the nature and impact of possible interventions more visible. Our review is based on the following research questions:

  1. 1.

    How are programming skills and/or CT promoted in children (age 6–15) with ASD and/or ADHD?

  2. 2.

    Which other skills may be fostered simultaneously among such children and what is the role of social and emotional skills in such a context?

  3. 3.

    What are the respective challenges and benefits associated with the various interventions?

Method

General Approach

Since the theoretical and empirical work on this topic is very limited and lacks methodological uniformity, we had to drop our initial idea of performing a systematic literature review on the effect of teaching CT and/or programming skills to students with ASD and/or ADHD.

In addressing the research questions we thus oriented ourselves towards scoping studies (Arksey & O’Malley, 2005; Levac et al., 2010) and scoping reviews (Munn et al., 2018; Peters et al., 2015). According to Arksey and O’Malley (2005), “the scoping study method is guided by a requirement to identify all relevant literature regardless of study design” (p. 22). Instead of providing a clear answer to a distinct question, as in a systematic review, the scoping review allows for a broader approach to the topic at hand (Munn et al., 2018; Sucharew & Macaluso, 2019).

Several guiding frameworks are now available describing how scoping studies and scoping reviews may be conducted (e.g., Arksey & O’Malley, 2005; Levac et al., 2010; Peters et al., 2015). We roughly followed the approach proposed by Arksey and O’Malley (2005) whereby scoping studies are conducted in five (often iteratively rotating) stages: Stage 1: identifying research question(s), Stage 2: identifying relevant studies, Stage 3: study selection, Stage 4: charting, Stage 5: collating, summarizing, reporting.

Literature Research Process

Search Terms

Although CT is an overarching category and programming represents just one way of promoting and revealing CT-related skills (Grover & Pea, 2013; Lockwood & Mooney, 2018), we use both terms interchangeably and thus speak of programming skills and/or CT. Furthermore, we use the terms coding and programming synonymously when talking about solving computational problems, even though coding is considered a sub-task of programming (Sun et al., 2021a, b; Zhang & Nouri, 2019).

In accordance with our above research questions (Stage 1), we determined three sets of search terms, each representing one of three crucial components in our research interest (Stage 2):

  1. 1.

    disorder/disability (ASD and/or ADHD and related symptoms),

  2. 2.

    computational thinking and/or programming, and

  3. 3.

    educational context.

These search terms were linked using Boolean operators (AND between the sets and OR within the sets of terms) and, if necessary, the search strings (see Table 1) were adapted to the specific requirements of the different databases.

Table 1 Search strings used
Full size table

Databases

The literature search was conducted in two steps in the period from September to December 2020. First, we conducted a search of the relevant publishers' databases in the areas of computer science and education, namely ACM Digital Library, IEEE Xplore, Springer Link, Science Direct, and Taylor and Francis Online. Second, we used Web of Science and Google Scholar to identify any possible additional articles which may have been overlooked in the previous step.

Inclusion/Exclusion Criteria

Based on our specific research interest, we defined a comprehensive set of inclusion criteria (Stage 3) to determine whether an article was relevant for our research questions. These inclusion criteria reflected the three components of the search strings, and concerned aspects related to methodology and publication-type. The criteria for inclusion of papers are presented in Table 2.

Table 2 Criteria for inclusion of papers
Full size table

Article Selection

The process of literature screening is shown in PRISMA flow chart (see Fig. 1).

Fig. 1
figure 1

PRISMA flow chart of the article selection process

Full size image

The database search yielded 1731 articles that fulfilled the first two inclusion criteria (relevant time period and language). Duplicates were removed and the remaining articles were screened with respect to whether they fulfilled the third and, potentially, also the fourth criterion. This left us with 256 articles. The abstracts of these were then checked in more detail to determine whether they complied with criterion 4 and whether the population/sample of the study matched our research interest (criterion 5). At the end of the selection process, 21 articles were identified that fulfilled all five criteria and these were then used in the analysis (Table 3).

Table 3 Databases used for included articles
Full size table

The key information/data from the selected papers was then extracted and structured according to the research questions (Stage 4). The relevant results are presented in the following section (Stage 5).

Results and Discussion

As a first step, we focused on specific study characteristics, i.e., the publication venue, location of the study, study design, materials and methods used to evaluate the effectiveness and feasibility of the approaches implemented in teaching programming and CT related skills, and on information about the sample (size, age of participants, gender and type of disorder of the students in the study). Since prior knowledge can play a significant role in the development of programming skills, we also checked whether the study participants had any previous experience with programming (or CT related skills) before participating in the intervention. We then systemized information on the interventions themselves (programs used, duration of the study, instructor(s) and setting).

Following this general description of the study and intervention characteristics, we then addressed the main research questions.

Characteristics of Selected Papers

An overview of the studies was gained by collating the information gathered in terms of several basic characteristics. The results of this collation are presented in Tables 4 and 5.

Table 4 Study characteristics: author(s), year of publication, publication venue, location of study, study design, sample, material and methods, and intervention/teaching approach
Full size table
Table 5 Intervention characteristics: author(s), year of publication, programs used, duration of the study, instructor(s) and setting
Full size table

Publication Venue and Location of Study

The 21 studies included in this review were published either as articles in scientific journals (n = 13) or as full or short papers in conference proceedings (n = 8). Most studies were conducted in Anglo-American countries (n = 15). However, as only those studies published in English were included here, no statement can be made on how many studies are otherwise available on the topic.

Study Type

Most of the studies are based on case studies (n = 6) or pilot studies (n = 7). We also identified one grounded theory study, some single case studies (n = 2), a combination of pilot and case studies (n = 2) as well as a combination of pilot and single-case studies (n = 1). Sometimes, the study design was not explicitly reported (n = 2) and had to be inferred by the authors.

None of the studies included in the present review had a control group. The source of the effects reported thus often remains unclear. Furthermore, the articles usually described studies in which students with ASD and/or ADHD were accompanied during the study period (up to 2 years), but not beyond it. None of the studies reported on a post-intervention follow-up after more than six months. Only Sakka et al. (2018) reported a follow-up six months after intervention and emphasized that intervention effects were still observable. As CT and social and emotional skills are becoming increasingly important in professional and daily life, studies that consider the long-term impact of interventions would be particularly valuable when attempting to gain further insight into this relatively unexplored field of research. In their systematic reviews, both Kakavas and Ugolini (2019) and Lockwood and Mooney (2018) point out that long-term studies on the cultivation of CT-skills are more or less non-existent. Our research also points to the need for more long-term analysis in this area.

Materials and Methods

Pre-and post-tests (n = 5) and multiple skill probes (n = 3) were conducted. The majority of the studies used interviews and observation (either directly or in the form of video analyses). As many papers did not specify the instruments used (e.g., observation grids) it is not possible to ascertain how systematic the analysis was.

Systematic empirical research in the field of teaching CT and related concepts to students with ASD and ADHD is clearly lacking. In accordance with the systematic literature reviews by Kakavas and Ugolini (2019), Lockwood and Mooney (2018) and Tang et al. (2020), we also stress the need for a standardized and systematic approach to reporting on and assessing CT. Without such an approach, reliable, objective and valid measurement of intervention effects becomes impossible. Tang et al. (2020), for example, pointed out that CT assessments are not applied equally in formal and informal educational settings, and that the former type of assessment dominates.

Intervention/ Teaching Approach

The studies included in the present review applied a number of, often quite specific, approaches (e.g., explicit instruction, project-based learning, game-based learning, discovery learning, video prompting, puppet master technique). In many cases (n = 9), the papers did not explicitly mention any underlying intervention approach or method. Thus, it remained unclear upon which learning theory the implemented intervention was based or even whether there was any underlying concept at all.

Duration and Instructor(s)

The respective durations of the interventions varied considerably, ranging from a single session lasting between 30–40 min (n = 1) to interventions lasting approximately two years (n = 2). Three studies reported upon interventions lasting about one week, three on interventions of one month, and four on interventions up to two or three months. Four studies covered a period of more than three months, but never exceeding two years. This is consistent with the finding of Kakavas and Ugolini (2019) that most studies on CT lasted up to one week or varied between one and six months. Even though most papers covered the whole duration of the program (n = 17), some papers (n = 4) gave no detailed information on the frequency and/or intensity of the interventions themselves.

Many studies found the relatively low number of participants to be quite challenging. While some authors (Albo-Canals et al., 2018; Yuen et al., 2014) argued that this was due to the short duration (sometimes a few consecutive days) of the intervention, others (Eiselt & Carter, 2018) reported that there was a decrease in the number of participants over time due to its relatively long duration (e.g., withdrawal due to moving). Whatever the case, a more flexible time arrangement, one based more closely on student needs, would appear to be more appropriate.

The types of instructor(s) most frequently identified were researchers (n = 4), teachers (n = 4), clinical staff (n = 2) and instructors from the STEM education center (n = 1). There were also various combinations: researcher and teacher (n = 3), clinical staff and robotics and sound specialist (n = 1), two college students with two high school students (n = 1). Sometimes the background of the person carrying out the intervention was not specified (n = 5). Whatever the case, the importance of systematic instructor training prior to intervention is clear, and this also needs to be considered in further studies.

Programs Used

Different programs and materials were used to teach programming to students with ASD and ADHD. Scratch (n = 8), Lego Mindstorms (n = 4) and Ozobot (n = 3) were mentioned most frequently. These programs provide a playful environment that is highly conducive to student learning. Some programs are also suitable for younger children. KIBO robot and Lego Mindstorms, for example, have already been used with six-year-old children (Albo-Canals et al., 2018; Lindsay & Hounsell, 2016; Lindsay et al., 2019).

Setting

Interventions frequently took place in special education schools (n = 6), both within (n = 3) and outside the classroom (n = 3). Some studies reported on programming activities in general education schools (n = 4) or in a medical context (hospital (n = 2), therapy (n = 1)). Some interventions also took place as after-school activities or in camps (n = 4). In some cases (n = 4) it was not possible to find information on the setting. Additionally, some interventions took place in a one-on-one setting (e.g., Knight et al., 2019a, b). Thus, how such interventions might be implemented in a group setting remains unclear.

Sample

Most samples (n = 18) were rather small (from 1–12 participants). Just three studies comprised a sample with more than twelve participants (45, 41 and 33 participants).

As indicated by several authors (e.g., Bossavit & Parsons, 2017; Gribble et al., 2020; Ratcliff & Anderson, 2011; Snodgrass et al., 2016), the explanatory power of such sample results is quite limited. Even where a much larger sample size is employed results may still not be generalizable for students with ASD as the disabilities among the children concerned can still be very heterogeneous (Lindsay & Hounsell, 2016). A further problem when attempting to compare impacts on ASD and ADHD students is the general lack of mixed groups (students with ASD and/or ADHD and students with other or without disabilities). Only nine studies reported on mixed groups, one of them, however, offering a separate group for children with ASD (1st group with eight participants without disabilities, 2nd group with four participants diagnosed with ASD). The other eight studies covered very heterogenous samples consisting of students with ASD and/or ADHD and students with other and/or without disabilities. The participant numbers ranged from two to 45 participants. In three cases there was just one student with ASD in the sample (for further information concerning the sample size and group composition see Table 4). Twelve studies reported on exclusively working with students with ADHD/ASD. The samples were rather small ranging from one participant in individual settings to a maximum of twelve participants. However, eight of these studies included no more than five participants.

Furthermore, some studies (e.g., Elshahawy et al., 2020; Knight et al., 2019a; Wright et al., 2019) referred to certain requirements the students had to fulfill to be included in the sample (e.g., sufficient motor, cognitive and communication skills). As students with ASD (and ADHD) do not form a homogeneous group but show various manifestations of the disorder, such findings are not generalizable to all students with ASD and/or ADHD.

Our own findings highlight that research focusing on teaching programming to students with ASD and ADHD has so far only addressed small samples. Additionally, there is a need for mixed-group studies using a larger sample in order to better represent ASD and/or ADHD learner diversity.

The participants were aged between six and 18 years. Most of the groups were mixed-age groups. Four of the studies did not report on the participants’ age.

Although girls participated in several of the studies examined here (n = 9), most participants were male. While ASD is more commonly diagnosed among males (APA, 2013; Lindsay et al., 2019), it is also possible the gender imbalance is related to the fact that women are currently underrepresented in the STEM field (Bati, 2021; Buitrago Flórez et al., 2017). Nevertheless, there is some evidence suggesting that the process of learning programming is different for girls and boys. According to Ratcliff and Anderson (2011), boys generated more commands and explored more than girls did, and used fewer reasoning abilities. In contrast, girls were more purposeful when creating specific procedures and were able to plausibly explain why they wanted to take certain steps. Although Bati (2021) found that girls and boys (without disabilities) performed similarly in CT and programming activities in early childhood education, the findings need to be augmented by more specific studies.

Of the 21 studies included in the present review, 18 reported on a group of students with ASD, whereas only two studies dealt with participants with ADHD. One single-case study consisted of a participant diagnosed with ASD and ADHD.

Concerning the participants’ prior experiences with computer programming, one third of the articles (n = 7) indicated that the participants already had at least some experience with computer programming before the study, while some other studies (n = 6) reported that the participants had no prior experience at all. Eight articles did not provide any information on this matter. One study reported a sample that already had programming experience. However, it should be noted that several authors suggested, albeit not always explicitly, that most participants with ASD and ADHD had already shown some interest in coding and/or robotics before the study. In accordance with Yuen et al. (2014), we conclude that it cannot be determined whether the results would be replicable if participants had not shown any previous interest in coding and/or robotics.

Summarizing all these findings, it is obvious that there is a dearth of (a) inclusive interventions with mixed groups, (b) interventions that equally include girls and boys and consider their different needs, (c) studies with larger samples, (d) studies with control groups allowing for reliable analysis of intervention effects, (e) studies presenting follow-up data and (f) studies that properly report on sample characteristics.

Results Concerning the Research Questions

Table 6 offers an overview on the reported findings concerning the research questions.

Table 6 Key findings on research questions
Full size table

Fostering Programming Skills

Even though it partly remained unclear how the effects of the interventions were measured, the majority of studies (n = 14) found an increase in students’ programming skills after the intervention and indicated that the participants progressively acquired more advanced CT skills. In contrast, one third of the articles (n = 7) mentioned no effect on CT skills, because this was not systematically investigated in the study.

Several studies reported that students learned skills such as the ability to operate with robots (e.g., calibrating a robot, drawing tracks) and gained greater understanding of fundamental programming concepts (e.g., reading and writing scripts, using sprites, loops, conditionals and variables). Some studies also reported that students learned to (independently) apply foundational coding skills to a novel, untaught code and gained the ability to plan, implement and check self-generated novel sets of codes. Other skills gained were enhanced problem-solving abilities such as logical thinking, synchronization and abstraction, as well as problem decomposition. As CT is applicable to various aspects of daily life (Tsortanidou et al., 2021) this finding is considered particularly valuable. Furthermore, some follow-up studies (Knight et al., 2019a, b; Sakka et al., 2018; Wright et al., 2019) revealed that students were able to maintain the skills acquired.

In most cases (n = 15), the studies did not explicitly mention which specific factors had a beneficial effect on the development of CT skills. However, it should be emphasized that approaches which entailed a structured and supportive learning process were particularly beneficial.

Other Skills Affected

Numerous studies reported on improvements in skills other than those related to programming and/or CT. Some studies (n = 3) found an increase in enthusiasm for (STEM) disciplines and that students exhibited a more positive view of computing careers following the intervention. In this regard, Sun et al. (2021a) indicate that “students' STEM learning attitude can positively predict their CT skills” (p. 356). One study reported an improvement in oral expression skills. The improvement of language-related skills can be especially relevant for children with ASD since this area is often affected in the diagnoses (APA, 2013). Several studies mentioned, and one study explicitly demonstrated, that student ability to focus attention and concentrate increased during coding activities. This finding is clearly relevant for interventions designed to help students with ADHD and/or ASD. In a similar vein, some of the studies (n = 2) also found that student perseverance increased when the latter were trying to complete a task.

Social-Emotional Skills

More than half of the studies (n = 12) reported that programming and CT activities had a positive impact on the social-emotional skills of students with ASD and ADHD. Some studies (n = 3) stressed that the computer appeared to act as a mediator for such children when engaged in social settings. In line with this, Gribble et al. (2020) reported in a single-case-study, that the students´ communication and interaction with others (peers and teacher) improved when working on a computer, regardless of the type of computing devices used (laptop vs. desktop) and classroom context (classroom vs. computer lab). In general, the findings indicate an improvement in pro-social behavior and a subsequent increase in (verbal) interaction and collaboration. Wright et al. (2019) reported that the students willingly taught their peers how to code once they had learned how to do it themselves. Sakka et al. (2018) found that such effects were still present six months after the intervention. While two studies reported that students preferred working with adults to working with peers, several other studies (n = 6) underlined that social interaction entailing communication and collaboration (e.g., helping each other) between peers was successfully stimulated through programming activities. One study observed an increased ability to offer positive and constructive comments, and an increased ability to receive feedback. These results indicate that programming can be used as a facilitator when reaching out to children with ADHD and/or ASD. The findings of the present scoping review reveal that social-emotional competences improve even when this is not the explicit aim of the intervention. It thus seems all the more likely that even greater benefits will arise when such competences are deliberately pursued.

Challenges Reported

Most of the authors of the studies investigated here did not go into any detail regarding the problems or challenges faced when implementing the various interventions (n = 13). Some mentioned that offering students a lot of freedom in their decision-making was quite challenging in some cases. In this regard Bossavit and Parsons (2017) emphasized that students with ASD may require a more rigid and planned environment in order to successfully engage in collaborative learning. Moreover, Eiselt and Carter (2018) indicated that students with ASD found achieving a specific target quite challenging as they tended to co-opt a given programming task and modify settings whenever they wanted. In line with this, Ratcliff and Anderson (2011) emphasized the importance of accompanying students in their learning process and of giving them clear instructions in order for them to fully understand the underlying concept and logic of the programming exercises (otherwise they just did things at random). Thus, a more structured and systematic teaching approach turned out to be more appropriate. In addition, Pilkington and Gelderblom (2010) reported difficulties concerning the introduction of new concepts and stated that “when the code was fairly simple, the learners were able to reproduce the code without difficulty. However, when more complex structures were demonstrated, learners resorted to copying the code from the teacher’s computer, showing little understanding of the programming structure” (p. 81). Consequently, the authors suggested dividing the phase of concept introduction into two separate sub-phases so that task practice and theory could be treated separately.

One study reported that students did not manage to complete the task in time. Another study reported on language issues when the software was written in a language completely unknown to most of the participants. In addition, Albo-Canals et al. (2018), who conducted a study with a duration of four consecutive days (consisting of two sessions a day), stated that not every child included in the sample was able to participate in every session due to mental and behavioral challenges associated with ASD. This was perceived as problematic since it resulted in a lack of data and thus interfered with the significance of the results. Accordingly, some authors (e.g., Bossavit & Parsons, 2017; Elshahawy et al., 2020; Lindsay & Hounsell, 2016; Snodgrass et al., 2016) emphasized that students needed varying amounts of support in order to strengthen their participation, problem-solving and programming skills. Lindsay and Hounsell (2016) reported on the adaptations required concerning the educational, cognitive, learning, physical and social needs of the participants. In line with this, Snodgrass et al. (2016) indicated that besides content-specific support relating to CT, student-specific support (e.g., giving verbal explanations, incentives to stay on task, model a task) was particularly valuable when attempting to increase student participation in CT activities.

These findings highlight the necessity of considering and addressing the students’ individual needs. Consequently, interventions should be designed and carried out in a way that corresponds to the diversity of learners with ASD and ADHD and need to take account of the specific challenges children with these disorders may experience. In addition, a certain degree of flexibility seems essential in order to meet the multifarious needs present in such a heterogeneous group of students.

Benefits Reported

As was the case with challenges, the benefits associated with the studies were also rarely mentioned explicitly (n = 12). Nevertheless, all of the studies mentioned in one way or another how conducive programming activities were to CT-related abilities and/or social skills. According to Lindsay et al. (2019), group-based programs (instead of a one-to-one format), had a particularly beneficial effect on communication and teamwork skills. Eiselt and Carter (2018) also stressed the importance of structured social groups prior to the intervention, as they provided a good opportunity for students to practice social skills (e.g., interaction and mutual understanding/empathy). Moreover, Munoz et al. (2016) identified the use of some type of collaboration scheme (students who completed their tasks were allowed to help their peers) as being particularly relevant when fostering participants' social skills. Bossavit and Parsons (2017) highlighted the fact that game-design learning allowed for flexibility in supporting individual preferences and thus helped learners become more independent and autonomous. Knight et al. (2019a) referred to the advantages gained when programming a robot. For example, such a “learning by doing” approach provides students with a direct and natural reward when a robot responds to instructions. Another beneficial approach described by Sakka et al. (2018) is related to the “puppet master technique”, where the students take on the role of a puppet master while programming a robot. Since children with ASD often face difficulties in direct interaction with others, such a setting provides safety for the students as the robot serves as a companion and “is protecting the programmer from a direct interference with other human beings” (Sakka et al., 2018, p. 12).

As mentioned above, some studies refer to the motivational effects observed among students engaged in programming exercises. Albo-Canals et al. (2018) indicated that teachers used robotics sessions as a reward for motivating the children to perform better in the regular classroom activities. Dorsey and Howard (2011) reported that the participants had a very high interest in telling their stories in virtual worlds. This led to their engagement, focus and social interaction being at an all-time high during game design and storytelling activities (compared to sessions merely consisting of robotic programming). Also, the presence of the participants´ parents during the intervention was described as being crucial in maintaining a familiar environment and thus in enhancing student learning (Lindsay & Hounsell, 2016; Munoz et al., 2018).

Conclusion

Our results showed that students with ASD and/or ADHD progressively acquired programming and/or CT skills that persisted beyond the intervention period and point to the beneficial long-term effects of fostering such skills among this group of learners. It also became apparent that the promotion of CT-related skills is particularly effective in enhancing social-emotional competences even when this is not explicitly intended. This probably derives from the fact that CT helps develop problem-solving ability in numerous contexts and (social) situations in everyday life. As students with ASD and ADHD often have difficulties in developing and maintaining social relationships, such an effect is particularly relevant. Although the studies examined rarely addressed the challenges and benefits related to the specific interventions, our results highlight that the following factors are important for success: (a) a planned and well-structured environment, (b) clear and systematic guidance and instruction, (c) individualized support, (d) flexibility in terms of implementation, (e) a positive learning experience involving tasks designed to trigger students´ curiosity, and (f) an opportunity for collaborative learning.

In line with Eiselt and Carter (2018), we thus conclude that an approach designed to foster CT and social skills is of particular benefit for children with ASD and/or ADHD. Interventions which focus on both these areas simultaneously, or which foster the development of social-emotional skills through programming are likely to provide a useful contribution to this field of research.

To gain more insight into what might work, future studies need to cover the following points: (a) the need for larger sample sizes, (b) properly reporting on sample characteristics, (c) sufficient elaboration of method and study design, (d) using a control group to investigate changes in skills, (e) reporting on follow-up data, (f) clear specification of the intervention goal and its setting, (g) shedding light on inclusive interventions with mixed groups, and (h) equal inclusion of girls and boys while also considering their specific needs.