Early Childhood Robotics: Children’s Beliefs and Objective Capabilities to Read and Write Programs

ABSTRACT

This longitudinal study of robot programming in early childhood (ROPEC) was performed based on summative and formative assessments of the robotics program in kindergarten and year one of elementary school. The study aims to broaden our knowledge about children’s understanding of programming, their confidence in ability to read and write programs, and their real capabilities of reading and explaining programs. In this study, we used a sample of 114 children (age 5–7 years) participating in the robotics program. Quantitative and qualitative data were collected from participant interviews. We added to the usual surveys a real evaluation by experts of children’s capability to read and explain the code. Unique emphasis was put on being able to assess objective vs. subjective aspects of a ROPEC participant. Our findings revealed significant differences between objective beliefs and real capabilities of children that should be dealt with in any robot programming; however, the findings were very positive. The results of this study provide crucial evidence that participation in ROPEC contributes in reducing the gender gap in science and engineering. Girls are not less interested or capable than boys in reading, writing, or explaining the code and in important aspects are even better and more eager.

KEYWORDS:

• Computational thinking
• computing education
• early childhood education
• gender differences
• programming
• robotics in education

Our children are surrounded by smart technologies and robots from birth. In the future, robots and Internet of Things devices will be omnipresent and numerous – thousands or even millions per person. Therefore, we should teach our children how to use these technologies, how to communicate with smart devices, and how to build and program robots.

Communication with robots includes the basic understanding of what a robot does, how it does it, and how to change its behavior. This requires at least some minimal basic understanding of programming. In the 21st century, being able to program will be as important as being able to read. Programming is a new literacy; it empowers the user and endows one with new ways of communicating, thinking, and expressing ideas (Bers et al., 2019; Kafai & Burke, 2014). There is an ongoing academic discussion concerning whether programming is a new literacy. Some argue that programming should not be taught as a compulsory subject (e.g., Tamatea, 2019). Some claim that they do not want to expand the notion of 21st century literacy beyond spoken and written language, and others claim that learning to program is excruciatingly frustrating and difficult. However, many educators have come to realize the importance of teaching how to program and agree that programming is a new literacy that should be taught (Wing, 2006). Programming is already a subject in many schools all over the world (Balanskat & Engelhardt, 2015; Hsu et al., 2019; Mannila et al., 2014). Numerous national educational programs are focusing on science, technology, engineering, and mathematics (STEM) literacy and making programming and computational thinking a priority for education (Manches & Plowman, 2017; Pretz, 2014). Some important questions being asked are: How early should we start teaching programming? Is it possible to teach programming in early childhood? How do we do it in an enjoyable and entertaining way?

Educational robotics

One of the means for integrating programming into the early childhood curriculum is early childhood robotics. Robotics is the branch of technology that deals with the design, construction, operation, and application of robots. The term “educational robotics” defined by Angel-Fernandez and Vincze (2018) refers to “a field of study that aims to improve student’s learning experiences through the creation and implementation of activities, technologies, and artifacts related to robots.” “Early childhood robotics” is a term related to implementation early childhood age-appropriate activities in “educational robotics.” Robotics allows for an enjoyable and playful way to learn technology, science, and computational thinking. Educational robotics (ER) and early childhood robotics are perceived by children as an exciting learning environment and fun activity (Zviel-Girshin & Rosenberg, 2021; Eck et al., 2014; Eguchi, 2014; Jung & Won, 2018; Sullivan, 2008). Playing with the robot can improve a child’s learning abilities (Connolly et al., 2012; Johnson et al., 2019; Prensky, 2001; Vogt et al., 2018). Several studies have shown that ER can improve children’s attitudes toward technology and science education (Zviel-Girshin et al., 2020; Benitti, 2012; Cejka et al., 2006; Eguchi, 2016; Sharma et al., 2019). ER is rich with opportunities to integrate not only STEM but also many other disciplines, including literacy, numeracy, social studies, music, and art (Durães, 2015; Goldenberg & Carter, 2021; Grover & Pea, 2018; Jung & Won, 2018; Macrides et al., 2021). Children who participate in ER programs are required to explore and think creatively in order to reach a solution (Bers et al., 2013; Israel-Fishelson & Hershkovitz, 2022). In addition to inspiring curiosity and creativity, ER gives participants the opportunity to solve problems and implement ideas with technology, to practice important 21st century skills and find ways to work together, to express themselves using technological tools, and to think critically and innovatively (Eguchi & Uribe, 2017; Noh & Lee, 2020). Essential 21st century skills include collaborative problem-solving, teamwork, communication, critical thinking, creativity, and imagination (Dede, 2010; Jung & Won, 2018; Kampourakis, 2013).

Aim of the study and research questions

This longitudinal study of robot programming in early childhood (ROPEC) was performed based on a Ministry of Education-supported pilot that added ROPEC to the curriculum of kindergarten and year one of primary school in Israel. Since 2016, thousands of students were observed and some specifically interviewed for this research. We intentionally strayed from the usual surveys to use this unique opportunity to gain insight into what is really going on in ROPEC programs. Many studies in ER investigate students’ computational thinking (CT) abilities. In our study, we didn’t want to use assessment tools designed for specific educational programming environments to measure students’ CT abilities but rather to investigate students’ beliefs and feelings about their ability to read and explain programs. Unique emphasis was put on being able to assess objective vs. subjective aspects of ROPEC students.

In this study, we aim to answer several research questions related to programming – from a child’s confidence in their ability to write and read a new program to their ability to read and understand code. The present study aims to broaden our knowledge about children’s understanding of programming, their confidence in ability to deal with programs, and their real capabilities. Specifically, we address the following research questions:

RQ1.

Do students of different ages or gender feel comfortable with their ability to write a new program?

RQ2.

Do students of different ages or gender feel comfortable with their ability to read and explain the code of their final project?

RQ3.

Could kindergarten and 1st-grade students read and explain the code in their final project? Are there any age- or gender-related differences in their abilities?

This article is structured as follows. In the next section, a theoretical framework is given. After that, a brief description of our robotics program and its implementation in kindergarten and elementary schools is presented. We then describe our methodology: research method, participants, procedure, and data analysis. The article proceeds with a summary and discussion of the relevant results and concludes with a suggestion for implementation of this program and its methods in early childhood education and future directions. At the end, some limitations of this study are given.

Theoretical framework

Several studies about teaching computational thinking and coding in early childhood have been conducted in recent years. Computational thinking and coding are closely related, as computational thinking serves as the foundational thought process behind coding. They are interconnected and often taught together, especially in computer science education. According to Wing (2011), “Computational thinking is the thought processes involved in formulating problems and their solutions so that the solutions are represented in a form that can be effectively carried out by an information-processing agent.” Stewart et al. (2021) found that educational robotics has expanded into curriculum beyond traditional STEM fields and can also be used to foster computational thinking skills. Resnick (2013) claimed that “in the process of learning to code, people learn many other things; they are not just learning to code, they are coding to learn. In addition to learning mathematical and computational ideas (e.g., variables and conditionals), they are learning strategies for solving problems, designing projects, and communicating ideas.” These skills are useful not just for computer scientists but for everyone, regardless of age, background, interests, or occupation. Shein (2014) argued that “not everyone needs coding skills but learning how to think like a programmer can be useful in many disciplines.” The discussion about CT was reopened by Wing (2006), claiming that computational thinking is a fundamental skill for everyone, not just for computer scientists. According to Wing, “to reading, writing, and arithmetic, we should add computational thinking to every child’s analytical ability.” Batı (2022) showed that age was an important factor in learning computational thinking in early childhood and that girls and boys performed similarly in programming and computational thinking. Yang et al. (2023) explored the impact of story-inspired programming on preschool children’s CT and found that storytelling as a culturally responsive approach, whether using robots or tablets as a programming tool, can provide young children with more inclusive and sustainable CT learning experiences. The main conclusion of Papadakis’s literature review of coding apps (Papadakis, 2021) was that coding apps positively affect the development of children’s CT skills and none of the apps ultimately support the development of computational fluency.

Numerous review studies have shown that computer programming through robotics can be a promising tool for the integration of technology and engineering in early childhood STEM education (Anwar et al., 2019; Çetin & Demircan, 2020; Ioannou & Makridou, 2018; Jung & Won, 2018; Tselegkaridis & Sapounidis, 2022). A majority of the studies in these reviews examined ER in entire K-12 settings and showed that empirical studies investigating programming in early childhood are scarce (Bakala et al., 2021; Macrides et al., 2021). The systematic review performed by Bakala et al. (2021) showed that computational thinking is still quite a new concept, not particularly explored in early childhood education (ECE) using robots. Also Su et al. (2023) in a scoping review of studies on coding curriculum in ECE claimed that coding curriculum design and use in ECE settings, as well as its effectiveness, is under studied. Additional studies found that young children age 4–7 years can create and program basic robotics projects (Bers, 2010; Cejka et al., 2006; Zviel-Girshin, 2020). Shoop et al. (2016) showed that robotics is viewed as an interdisciplinary pathway to integrate and practice computational thinking.

Numerous studies about children’s opinions and preferences regarding a tangible vs. a graphical interface during programming activities showed that the tangible interface was characterized as easier to use by younger children (5–6 years old) who were less experienced with computers (Sapounidis & Demetriadis, 2013; Sapounidis et al., 2015). Additional studies showed that screen-free or unplugged robot programming is a developmentally appropriate tool to enhance children’s STEM-related cognitive abilities, such as CT and sequencing ability in ECE (Çetin & Demircan, 2020; Macrides et al., 2021).

Bers and the DevTech Research Group at Tufts University conducted several studies to address questions related to ECE and coding. The concept of “coding as a playground” and “coding as another language” allowed children at a very young age to code robots through fun, play, and creativity. Coding as a playground, together with the Positive Technological Development Framework, confirmed that it is possible to start teaching coding and computational thinking in early childhood, even for children 3 years old (Bers, 2017, 2019; Bers et al., 2013, 2019). Yang et al. (2022) demonstrated the positive benefits of robot programming to early childhood development in terms of CT and sequencing ability, compared to a traditional curriculum activity such as block play in ECE.

Description of the ROPEC program

The program

The authors have developed a novel approach to teaching robotics starting at especially young age. It has been running very successfully since 2016. It became a Ministry of Education and local authorities’ novel initiative for robotics as a springboard to enhance technological thinking and learning values in early childhood. Under the auspices of the program, the curriculum in this pilot (later to become the nationwide model) for K-1 (kindergarten and 1st grade) was changed, adding the study of technology and robotics as compulsory component. Initially, 4 kindergarten and 12 first-grade classes in 3 schools from diverse socioeconomic and ethnic backgrounds were involved. After two years, the program was extended to 5 kindergarten and 20 first-grade classes in six different schools.

The uniqueness of the program, in addition to its scale and academic approach, is in the Vygotsky scaffolding (Vygotsky, 1978) being provided; it used our novel model of the students’ dialog with their natural surroundings and caregivers (e.g., teacher and parents), who all were instructed and in continuous communication with experts. In this program, the classroom teachers play the important role of robotics instructor. Before joining the program, teachers participated in special training workshops: one for those working in kindergartens and the other for those working in elementary schools. The workshop duration was 25 hours. Program managers assisted and supported the classroom teachers in their robotics instructor role. This support included weekly teaching materials, immediately answering questions by phone, e-mail, or chat application. Novel teachers participating in the program received an additional one-hour weekly meeting with a more experienced teacher or an external help.

The program was funded by the regional council after approval by the Israeli Ministry of Education. The Ministry’s Science Supervisor granted it a special official permit that is required for underage surveys and for programs conducting research.

The main objectives of the program are to: integrate an engaging matter robotics into science and technology education; create a confident personality; improve technology and science education; introduce students to the technology-infused world in which we live (Ioannou & Makridou, 2018; Manches & Plowman, 2017); and enhance students’ self-confidence, self-efficacy, and belief in their own abilities (Benitti, 2012). The program also aims to help children develop essential 21st century skills, such as collaborative problem-solving, teamwork, communication, critical thinking, creativity, and imagination (Kampourakis et al., 2013).

Kindergarten robotics

In order to introduce robotics and technology to kindergarten students, a special lesson was added to the curriculum once a week. The main equipment used for this lesson was the LEGO® Education WeDo kit, which came with an easy-to-use programming environment that could be installed on desktop computers and tablets. This construction kit has more than 150 elements, including an intelligent brick, motor, tilt sensor, motion sensor, and LEGO USB smart hub. Each kindergarten created a dedicated “robotics area” equipped with tables, various electronic and robotics kits, tablets, and computers with the LEGO® Education programming environment installed (Figure 1).

Figure 1. Robotics play area in the kindergarten (here with the lego robotics equipment).

The lesson was taught by a local kindergarten teacher, who received training before and during the school year. An additional specially trained kindergarten teacher, called the expert, also visited each kindergarten once a week to assist the local teacher with ER. At the beginning of the program, the expert’s role was to reinforce the local instructor and give support in case of technical or pedagogical difficulties. Later, during the school year, the expert’s presence allowed division of the students into smaller groups for individual tutoring and education. Working in small groups enabled the teachers to stimulate the students’ curiosity and encourage them to think, provide different solutions, and explain their choices. It also opens an opportunity for the students to practice oral communication and use technical terms related to the subject. Overall, the program aimed to create a friendly learning community that encouraged innovation and problem-solving skills.

Elementary school robotics

A two-hour robotics lesson was added to the 1st-grade curriculum in elementary schools. During the robotics lesson, the class was divided into two groups that took turns attending the lesson in the regular homeroom classroom with their regular teacher and the science classroom with a science teacher or external instructor. Each group was later divided into smaller teams (2–4 members) to work on collaborative problem-solving assignments. Both the homeroom and the science class had a robotics area with tables, a variety of electronic and robotics kits, and several computers on which the LEGO WeDo 2.0 programming environment was installed. The main equipment was the LEGO® Education WeDo 2.0 kit specially designed for elementary schools and accompanying materials that included an eLearning program to help teachers with the kit and curriculum. The school computers were used for instructional booklets, designing predefined models, and programming the robot. Programming is an important part of ER, and it is an essential part of all WeDo 2.0 projects. Writing code gives life to the models that the students create and teaches them basic principles of computational thinking.

The homeroom general education teacher taught the robotics class. Each lesson employed a mediated learning approach that included both direct instruction and open-ended, student-directed inquiry. The direct instruction included short lectures or multimedia demonstrations on various subjects, while the open-ended inquiry had students working in teams to solve programming and design challenges, encouraging them to give oral explanations and predict outcomes. Some content was based on accompanying materials of the LEGO® Education kit, and some were specially designed units. Some challenges were well-defined while others were intentionally left open-ended, leaving room for creativity, imagination, and inventive thinking.

Programming environment

At the beginning of the program, some concerns about teaching children to program or to write code, before or in parallel with literacy and numeracy, were raised. Participants in this program were 5–7 years old; the majority of them either could not read or had just started to read and write. An additional concern was that their mother tongue was Hebrew, which uses a non-Latin alphabet. Hebrew, like Arabic and Syriac, is written from right to left. The text on a single page is written from right to left, and the whole book itself is written and read from right to left. This creates an additional difficulty in many programming environments where programs are written from left to right.

In both the school and kindergarten settings, the LEGO Education programming environment was used. This icon/block-based visual programming language lets users create programs by manipulating program elements graphically rather than by specifying them textually. It does not require writing or choosing graphical blocks with text, like Scratch or mBlocks. Therefore, in our case it was the most appropriate learning environment.

This programming environment is specially designed for young children. They write their code by choosing a correct command symbol from a palette that shows all the programming blocks, written like an icon. Then they drag and drop blocks to compose a program. A program is a list of blocks combined in some order (Figure 2). Program execution is immediate. The robot, connected to the environment, immediately executes the written program. This physical artifact helps to connect between the abstract activity of programming to concrete observation of the execution of the written activity. The result of running the program, also known as program execution, and understanding if this robot solves the problem is immediate. In the case of an incorrect solution, they can fix the code and re-run the program. This process encourages constant collaboration around the programming task at hand and discussion of different possibilities of the written solution.

Figure 2. (a) A programming environment screen and (b) children working with the environment.

Final project

Six to eight weeks before the end of the school year, participants started to work on a capstone project of their choosing. This part of the program adds storytelling or/and story-inspired robot programming responsive pedagogy approach to ROPEC (Yang et al., 2023). During the preceding months, all were engaged in a dialogue about engineering problems that robots could solve. The story defining the problem to be solved by the participants was changed every year. One year, it was about the Moon travel and helping Moon settlers. Another year, it was smart robotic transportation and self-driving transportation. Later, it was about robots assisting animals in trouble. Teams of 2–4 participants were challenged to analyze this problem from an engineering point of view (i.e., environmental or technological) and then to construct and program a creative solution for it using robots. The robots, along with descriptive posters, were presented at the annual exhibition at the school or kindergarten and at a Robotics Day event at the Science Center. The teams made presentations to other students, family members, teachers, and stakeholders.

Methodology

In this section, we discuss how the study was implemented, the research group structure and activities, analytical framework, the participants, the procedure, and data analysis.

Research group

A special multidisciplinary group of researchers planned, supported, and examined the program. The research group included experienced researchers and doctorate students: six researchers from the field of education, three psychologists, two researchers in the field of engineering, and two researchers in the field of management. Special emphasis was given to the researchers in the field of early childhood. Different group members examined different aspects of the program: educational, linguistic, scientific, engineering, managerial, psychological, and more. During the first year of the program, the research group only observed the program, visited kindergartens, and schools periodically and talked with the educational teams. Those preliminary qualitative observations helped the researchers build a set of interviews, surveys, and activities for different aspects of the program. During the study, researchers constantly talked about research activities, built a detailed research plan, performed quantitative and qualitative analysis, and discussed the results and future research activities and questions. The analytical framework of this research is shown at Figure 3.

Figure 3. Analytical framework.

Participants

In this study, we used a sample of 114 students participating in a robotics program. The participants were from two kindergartens and two elementary schools. The kindergartens and schools were selected according to availability of research group members, teachers, and professional video expert (all familiar to the children). In the kindergartens, the participants, age 5–6, were chosen by the program manager or kindergarten teacher. One of the criteria for choosing children was good verbal communication skills. At schools, 6- to 7-year-old 1st-graders were selected by the program manager, the school principal, and the relevant teachers. For the one-on-one interviews, there were 51 (45%) kindergarten students (23 boys and 16 girls) and 63 (55%) 1st-grade students (22 boys and 24 girls). The proportion of boys was slightly higher than girls, with 61 boys and 53 girls (53.5% and 46.5%). All the children willingly volunteered to talk to the researchers, to present their robotic model, to explain what it does, and to answer the researchers’ questions.

Procedure and data analysis method

The Science Supervisor in the Israel Ministry of Education granted the program a special official permit required for underage surveys and programs for conducting research. All participants were informed about the study, what it is about, and how it would be conducted. Parental consent for their child’s participation in research was received at the beginning of each year. In addition, prior to the interview, children were asked to give their consent for video-recording of the interview. The confidentiality and anonymity requirements to protect the participants’ privacy were rigorously followed. Therefore, all material has been anonymized during the analysis and publication of the research results. The collected material is only used for research purposes.

The one-on-one interviews were conducted in a quiet room. The use of interviews is very suitable to ascertain one’s opinions and beliefs. The interview had two parts. In a quantitative part of the interview, data was collected from the children in the form of a predefined survey; in a qualitative part of the interview, the children answered an open-ended list of questions. The majority of closed-ended interview questions had only dichotomous yes/no format of the answers. Coombes et al. (2021) in a systematic literature review showed that children less than 7 years old think dichotomously and need two response options only. Mellor and Moore (2014) argue that a dichotomous yes/no format is one of the best formats for young children and if a Likert scale is used, then word-based response formats should be used. In kindergarten, the interviews were conducted in a special robotics area inside the kindergarten building where the robotics lessons took place. All the adults participating in this interview were familiar to the children. In addition to researchers, a kindergarten teacher was present. At schools, the interviews took place in a special small classroom located near the regular class. All the adults were familiar to the 1st-graders and also a homeroom teacher was present. During the quantitative part of the interview, each participant was asked the same questions (with minor adaptations to kindergarten and school). The research assistant – with whom the participants were familiar because of several visits during the school year to their school or kindergarten – helped them read the questions or statements and the choices of answers and, for those participants too young to read or write, recorded their responses. Each participant brought their capstone project robotic model to the interview (at schools, they also brought a descriptive poster prepared by the team). The robot model was physically present in the room (Figure 4).

Figure 4. (a) An example of a capstone robotic model and (b) its code.

In the context of early childhood robotics, interview protocols aim to explore how children perceive and make sense of robotics, programming concepts, and their own experiences with robotic models and programming. The focus is on understanding the children’s beliefs and thought processes related to these subjects. In a conversation-like structured second part of the interview, each participant presented his team robotic model, explained what the model did, and answered questions from an open-ended list of questions. The interview protocols consisted of a series of open-ended questions that prompt children to talk about their robotic models, programming experiences, and beliefs. The questions were designed to elicit detailed responses and encourage the children to reflect on their own knowledge and understanding. Here are examples of some of the interview protocol’s questions: Can you show us your robotic model and explain what it can do? How did you learn to program your robot? What do you understand about programming? What do you like most about using your robot? Is there anything challenging about it? Can you tell us about a specific task or problem you solved with your robot? How did you approach it? Here is a code of your final project. Can you read it and explain what it does? Can you explain some specific part of your code (chosen by the research assistant) and what it does? Do you think your robot can do anything else or be improved? Why or why not? If you could create a new program for your robot, what would it be? All interviews were video recorded. Qualitative analysis included in-depth examination of audio and video recordings and analysis of their transcription. The video recordings were kept for two years in separated external hard disks. All the video recordings were transcribed and used in the research. All the transcriptions and the obtained data were anonymized before the analysis. The videos were deleted after two years, according to the requirement of the Science Supervisor in the Israel Ministry of Education.

To answer the third research question, each participant was asked to read and explain the code of their final capstone project, on which the participants worked in teams of 2–4 members for the period of 6–8 weeks. Then, participants were asked to show the final project model and tell a story of how the robot/program worked. Later, each participant was asked to tell what the functions of the different program blocks were, and which codes or parts of it do what. To answer this last question, each participant was given a printed-in-color version of the code, the code was shown on the computer screen, and the robot model was physically present at the room near the child (Figure 4).

All the questions were formulated by ECE experts and psychologists. For each question, an additional simpler description or version of the question or statement was given. For example, when we wanted the participant to give a definition of “robotics,” instead of asking, “Can you give us a definition of robotics?,” we asked the child to give an answer the following question: “If a new child comes to kindergarten/school how will you explain to him what do you do during robotics lesson?” Prior to the interviews, all the questions were shown to the teachers to obtain their comments and thereby formulate a better version of the question. The questions were asked by an adult familiar to the students. Before telling a story of how the robot program worked and explaining the code of the project, each participant was asked several general questions to estimate the level of students’ communication skills (receptive and expressive language) because one of the criteria for choosing the kindergarten student was ability to communicate. This ability is important since Marinus et al. (2018) found a positive association between cognitive compiling of syntax in natural language and programming ability of 3- to 6-year-old children.

The story-inspired robot programming or story-telling part of the interview was analyzed by a team of experts, two researchers with backgrounds in engineering education and education. They evaluated the participants’ confidence level in reading and explaining the code. A score for understanding was defined in a Likert scale from 3 to 1 (i.e., 3 - could read and explain the code, 2 - could partially read and explain the code, 1 - does not understand what was written). In 84% of the cases, the raters agreed one with the other (96 of 114 interviews). An inter-rater reliability analysis using the Cohen’s kappa statistic was performed to determine consistency among raters. Later, in cases of disagreement, researchers re-watched the video recording and came to a common agreement.

IBM SPSS Statistics 28 software was applied for quantitative data analysis. The nonparametric Fisher’s exact test and the Chi-square test for independence were used. For the qualitative data, the participants’ responses to the open-ended questions part of the interview were subjected to content analysis. In addition to giving a grade for reading and explanation of the code, our team of experts was asked to perform the content analysis and to identify the problems in participants’ reading or explaining the program. For this part of the analysis, an additional expert in the field of computer science education was added to the team of two researchers with backgrounds in engineering education and education. A deductive coding approach was used for categories that emerged from the analysis.

Results

The results of the quantitative stage of the research are followed by the results of the qualitative stage. Cronbach’s alpha coefficient showing the internal consistency reliability was 0.765, indicating an acceptable level of reliability.

The aim of the first research question was to check the children’s belief in themselves. Psychologists refer to the concept of self-belief as self-efficacy and define it as a belief in one’s capability to accomplish certain goals (Bandura, 2010). To check confidence in ability to write a new program, each participant was asked to give their level of agreement to the statement: I feel that I can program a new behavior of the robot. The children were given dichotomous yes/no format. The results are shown in Table 1.

Table 1. Frequency of responses to the statement “I feel that I can program a new behavior of the robot” (n = 114).

Response	Boys	Girls	Kindergarten	1st grade
Yes	48	47	40	55
No	13	6	11	8

The Fisher’s exact test revealed that the difference in percentage of responses of boys and girls (p = .209) was not significant. The same test revealed no significant differences in percentage of responses of the kindergartners and 1st-graders (p = .219).

The aim of the second research question was to check the children’s confidence in their ability to read and explain the code of their final project. Each participant was asked to give their level of agreement to the statement: I feel comfortable to read and explain the code of our final project. The results are shown in Table 2. Once again, the children were given a dichotomous yes/no format. During this question, the printed version of the code and the robot were present in the interview room.

Table 2. Frequency of responses to the statement “I feel comfortable to read and explain the code of our final project.”

Response	Boys	Girls	Kindergarten	1st grade
Yes	53	42	40	55
No	8	11	11	8

The participating girls were slightly less confident than the boys. However, the Fisher’s exact test revealed that the difference in percentage of responses of boys and girls (p = .205) was not significant.

More participating 1st-graders strongly agreed with the statement than did kindergartners; however, this difference was still not significant according to the Fisher’s exact test that revealed no significant differences in percentage of responses of the kindergartners and 1st-graders (p = .196).

The aim of the third research question was to check if the children’s confidence in their ability to read and explain the code of the final project matched the objective’s results. Each participant was asked to read and explain the code of their final project (the capstone projects were performed by teams of 2–4 members). Then, participants were asked to show the final project model and to tell a story of how the robot/program worked. For this part of the interview, the robot model was physically present at the room near the child (Figure 4) and each participant was given a printed-in-color version of the code and the code was also available on the computer screen. Later, each participant was asked to tell what the functions of the different program blocks were, and which codes or parts of it do what.

As mentioned before, this part of the interview was analyzed by two researchers with backgrounds in engineering education and education who gave a number between 1 and 3 (i.e., 3 - could read and explain the code, 2 - could partially read and explain the code, 1 - does not understand what was written) to identify the participants’ confidence level in reading and explaining the code. The inter-rater reliability for the raters of the explanations given by the participants was found to be Cohen’s kappa = 0.747 (p < .001), indicating according to Fleiss et al. (2003), good consistency between the raters.

According to Table 3, girls’ performance exceeds boys’ performance in reading and explaining the code. Chi-square test for independence revealed dependence between the child’s gender and the confidence level in reading and explaining the code χ2 (2) = 13.76, p = .001.

Table 3. Frequency of the experts’ mark to the question “can the child read and explain the code in its capstone project?” by gender.

Response	Boys	Girls
Yes	29	43
Partially	23	7
No	9	3

The results (Table 4) of the chi-square test for independence indicate independence between the age of the child (kindergarten, 1st-grade) and the confidence level in reading and explaining the code χ2 (2) = 1.16, p = .36.

Table 4. Frequency of the experts’ mark to the question “can the child read and explain the code in its capstone project?” by age.

Response	Kindergarten	1st grade
Yes	32	40
Partially	16	14
No	3	9

In addition to giving a grade for reading and explanation of the code, our team of experts was asked to perform the content analysis and to identify the problems in participants’ reading or explaining the program. One common mistake was not reading or explaining the code from the poster or computer screen but describing the sequence of events from memory. Some children did not read the program at all or some only partially. They did not describe blocks of commands in their program precisely; instead, they either told the story of what the robot should do or did not describe the behavior of their robotic model at all.

Additional deeper qualitative content analysis of the 114 transcriptions by experts identified common problems in reading and explaining the code. The categories emerged from this analysis and the percentages of participants that expressed them were as followed: did not describe all icons of commands in the program (28%), did not read commands in the correct order (13%), did not describe last icons (11%), did not mention icons in which direction of rotation of motor was set (8%), did not mention icons in which motor power and motor on for specific number of LEGO seconds (6%), sometimes missed the condition or misunderstood the condition (23%), or did not understand that the set of icons was in the loop structure (18%). Some of the transcriptions identify having no problems at all, some having only one problem, and some having more than one problem.

Discussion

In general, very little research exploring gender differences in young children’s robotics and programming abilities exists (Montuori et al., 2022). Therefore, gender- and age-related analysis of the results were deemed as very important. Our goal in this study was to investigate students’ beliefs and feelings about educational robotics and their ability to read and explain programs.

The first research question aimed to check the children’s confidence in ability to write a new program (Table 1). A majority (83.3%) of participants agreed with the statement: I feel that I can program a new behavior of the robot. This result is very promising; it shows that these participants consider themselves capable of writing a new program or coding a new behavior of the robot. Kindergartners were slightly less confident than the 1st-graders; however, this difference was not significant. This distinction was expected, since school children’s competence with written language, in both reading and writing, is higher as they have had one year more of reading and writing practice. An additional interesting finding here is gender equality in this research question. Our findings were not consistent with those of the Master et al. (2021) study in which they showed that children endorse gender-interest stereotypes favoring boys about engineering by 1st grade and about computer science by 3rd grade. In our case (age 5–7 years), we did not find gender-interest stereotypes. However, we did not ask the specific “stereotyped” question about the programming activity, like “girls are much less interested in this activity than boys.” With regard to actual coding ability, Price and Price-Mohr’s (2023) findings strongly support the absence of gender differences in coding ability of middle-class (children age 10–11 years old). Our findings extend their results for an earlier age.

Zdawczyk and Varma (2022), in their study of students’ attitudes and beliefs about learning Scratch and Python in elementary and middle school, did reveal higher self-efficacy for boys compared to girls. However, our study was different in two aspects: our participants were younger and the question in our case was about programming a new behavior of a tangible device and not only writing a code in programming language, like Scratch or Python. It is possible that there is an attitude change with age. For example, Kong et al. (2018) showed that students in later grades (senior primary school) viewed programming as less meaningful and had less programming self-efficacy. Also, boys showed more interest in programming than girls did. Their results related to gender suggested that more effort is needed to attract girls to engage in programming activities, as girls indicated less interest in programming than boys did. According to our study, this motivation of girls is less crucial in early childhood since girls feel comfortable in their ability to write, read, and explain the code.

The second research question aimed to check the children’s confidence in their ability to read and explain the code of their final project (Table 2). Once again, a majority (84.2%) of all participants agreed with the statement: I feel comfortable to read and explain the code of our final project. The girls were slightly less confident than the boys; however, the difference was not significant. More 1st-graders strongly agreed with the statement than did kindergartners, which was expected due to their experience in reading. During the period commonly known as the 5- to 7-year shift, in addition to biological maturation and experiential differences, children develop more sophisticated cognitive skills and capabilities. However, this one-year development difference was still not significant.

Atmatzidou and Demetriadis (2016) explored the topic of level of self-efficacy and confidence between boys and girls. In this study, boys demonstrated a higher level of self-efficacy and confidence toward STEM subjects than the girls. However, participants in their study were two groups of students age 15 and 18. Our findings in early childhood also revealed no gender significant difference.

Del Olmo-Muñoz and coauthors’ research (Del Olmo-Muñoz et al., 2020) showed that gender does not influence the acquisition of CT skills in early primary education but gender influences motivation toward a computational thinking instruction. This finding can explain why in our study girls were slightly less confident than boys in their ability to read and explain the code of the final project.

The objective results (Table 3) show that experts found that the majority (63.2%) of participants could read and explain their code and a quarter of them could do it partially. The most surprising finding of our study is the fact that girls’ performance exceeded boys’ performance in reading and explaining the code. This result is extremely interesting since girls’ confidence in their ability to read and explain the code (RQ2) of the final project was lower than the boys. This shows that girls evaluate their abilities differently from boys and tend to have less positive attitudes about their ability to read and explain the code. Breaz (2019) argued that girls manage to enrich their vocabulary and use the right terms faster than boys and that their communication is much easier than boys, which would explain why girls could better read and explain their final project code. In addition, according to Lynn and Mikk (2009) mean gender differences in quantitative reasoning (such as mathematics and science achievement) are generally small, gender differences in reading achievement are somewhat larger and found universally across all nations which make them less controversial. Overall, according to Lynn and Mikk, girls and women demonstrate better language and reading skills than boys and men. These results can also support our findings. Reilly (2015), in research findings drawn from the Programme for International Student Assessment (PISA), showed that reading achievement of girls (15-year-old students globally) were higher than boys. Maybe this reading achievement starts even earlier, at the early childhood age, and can explain our findings.

Hassenfeld et al. (2020) found that there was evidence for a weak, positive correlation between students’ literacy levels and their programming mastery, as determined by the curricular programming assessments. The positive correlation suggests that there may indeed be underlying constructs that overlap between literacy and programming. Perhaps their findings can support our results of significantly higher grades in reading and explanation of the code for girls. Ardito et al. (2020) found that 6th-grade boys focused more on the operational aspects of building and coding their robots while the girls focused more on literacy and active collaboration. Their findings can also explain why in our study girls had significantly higher grades in reading and explaining code at an earlier age.

Sullivan and Bers (2013), during the TangibleK Robotics Program, found that very few of the gender-related differences were statistically significant. In their study, kindergarten girls did not score significantly higher than boys in any area. During the final project, all tasks and debugging concepts were assessed and boys’ and girls’ mean scores were compared and no significant gender differences were found in the final project. In our study, in the category of reading and explaining the code of the final project, the girls were significantly better than the boys.

Montuori et al. (2022) revealed a strong association between children’s coding abilities and their executive functioning, as well as the existence of gender differences in the maturation of response inhibition and planning skills, but with an advantage for girls. They found that the existence of gender differences favoring girls in response inhibition and planning from as early as 5–6 years of age would cause one to expect an advantage for girls over boys to emerge also in the coding tasks that involve algorithmic thinking. In our study, we did not measure algorithmic thinking; however, the stages of the final projects were identifying the problem and finding the solution. Finding the solution required construction of the robot and implementing some algorithm with a new behavior for the constructed robot. Therefore, their findings also support our results that showed that girls were significantly better than boys in reading and explaining the code.

Angeli and Valanides (2020) identified a statistically significant interaction effect between gender and scaffolding strategy and found that boys, age 5 to 6, benefited more from the individualistic, kinesthetic, spatially oriented, and manipulative-based activities with the laminated cards using robotics devices, while girls benefited more from CT-related collaborative writing. The final capstone project requires teamwork and collaboration writing during a period of time therefore the results of Angeli and Valanides can support our findings.

Price and Price-Mohr (2023) found no gender differences in coding ability of middle-class children and suggested that practitioners should not assume that gender differences exist in the context of coding ability, and they should not adapt their teaching to gender. We support their suggestion and propose choosing the topic of the final project in such a way that it will be attractive for both genders. Also, Price and Price-Mohr (2018) argued that coding can be thought of as a form of literacy and that taking this perspective may lead to benefits for children as both literacy learners and also learners of computer programming.

Analysis of the mistakes showed that one common mistake was not reading or explaining the code from the poster or computer screen but rather describing the sequence of events from memory. The qualitative content analysis of the problems in reading and explaining the code revealed that most common problems were not describing all icons of commands in the program (28%), misunderstanding of the condition or missed conditions (23%), and misunderstanding that the set of icons was in the loop structure (18%). No gender-related mistakes or problems were identified. Our analysis had some similarities with the Veselovská and Mayerová (2015) study in which they aimed to identify in which types of activities pupils most often made mistakes. Sullivan and Bers (2013), in their study during the TangibleK Robotics Program, also identified some of those problems during the debugging, or problem-solving, element of their program.

Su et al. (2022) wrote a review about the influences of gender and socioeconomic status on children’s use of robotics in early childhood education. Several suggestions on how to reduce the gender gap in robotics activities were offered. Most of those suggestions are present in our program: diversifying the classroom pedagogies, controlling the size of class, communicating with parents, and inviting female teachers to guide students in robotics education. The majority of our research findings show that there are no gender-related differences and that girls feel as comfortable as boys about their ability to read, write, and analyze computer programs. Our findings are similar to recent literature review regarding computational thinking and programming in early childhood education; the study of Batı (2022) showed that girls and boys performed similarly in programming and computational thinking. Also, Papadakis et al. (2016) found that preschooler gender does not affect performance in computational and digital skills and that the age of children did not affect their performance in understanding basic programming concepts.

Conclusions and future directions

In the world of education, children need to be prepared with skills that will ensure their competitive level in different fields and especially in the field of science and technology. In this research, we assume that programming is a new literacy. Literacy plays a significant role in reducing gender, race, nationality, and religious inequalities. Literacy skills are crucial capital for someone to develop individually, to live satisfactorily, and to achieve success both in learning and everyday life. Our findings provide significant support for teaching this literacy at the earliest possible stage. The present study aimed to broaden our knowledge about children’s understanding of programming, their confidence in ability to deal with programs, and their real capabilities. The results offer important information about children’s feelings toward implementing new programs, beliefs of self-efficacy, and their reflections toward programming. The results of this study support that participation in robotics projects in early childhood may reduce the gender gap in science and engineering. In the category of objective assessment of reading and explaining the code of their final project, girls overperformed boys. This provides additional evidence that participation in robotics projects in early childhood helps to empower girls and achieves gender equality in science and engineering.

So, how early should we start teaching programming and is it possible to teach programming in early childhood? According to our results, both kindergartners and 1st-graders feel that they can write a new program or code a new behavior of a robot. They also feel comfortable in their ability to read and explain the code of their final project. Our objective test of that ability showed that it is not too early to start teaching programming in early childhood – it is even recommended.

Our findings show that there are significant differences that should be dealt with in any ROPEC program between objective beliefs and real capabilities. On the other hand, the findings are very positive. Even kindergarten is not too early to start ROPEC. Girls are not less interested or capable than boys and are better and more eager in important aspects. Thus, ROPEC is an important instrument in bridging the gender gap. ROPEC is perceived by students with more positive beliefs and mental mobilization, even beyond their age-related capabilities; it has significant positive impact, not only immediately, as scaffolding for their Vygotsky zone of proximal development. It helps to build not only engineering skills and motivation but also a happier, more confident, and capable person.

In future studies, we plan to examine the ability to read and explain the other’s code. Programming activities and tasks should be given to the students to check their ability to write new programs. Team-work and team-programming should be compared with single participant programming.

Limitations

This study has some limitations. It should be taken into consideration that this study was not conducted using a random sample, but rather used a sample of children who agreed to participate in the interviews. Also, the observers who analyzed the answers to the third research question could have some prior beliefs and thus bias the results. However, we feel that having two independent evaluators analyzing the findings can reduce this bias. An additional limitation is usage of only an on-screen environment. Some further exploration of the identified results with other robotics platforms (tangible robotics environments and unplugged/off-screen environments) should be done.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Ruppin Academic Center [33139].