Gender Differences in Computational Thinking Skills among Primary and Secondary School Students: A Systematic Review

Abstract

With the rise of the concept of gender equality and the emphasis on computational thinking, more and more research on computational thinking is being placed in the context of gender differences to promote gender equality. This systematic review examines and analyzes 23 articles that focus on gender differences in computational thinking skills among primary and secondary school students, providing a comprehensive overview of the existing literature while also providing direction and framework for future research. The results show that (1) the current research is imbalanced across regions and age groups, and the research on inherent gender differences in computational thinking skills remains insufficient; (2) the development of a clearer and more specific definition of computational thinking and corresponding assessment instrument is required for a more specific identification of gender differences; (3) under the existing educational environments, gender differences in computational thinking skills among students in different regions exhibit different pattern; and (4) interventional factors contributing to gender effects in interventions have also been identified.

1. Introduction

Women are still an underrepresented group in the fields of science, technology, engineering, and mathematics (STEM). According to reports, women accounted for only 33% of the technology workforce in 2024 [1], only 19% of students who obtained a Bachelor’s degree in Computer Science were women in 2016 [2], and only 25% of computing roles belong to women since 2015 [2].

With the rise of gender equality in most countries around the world, more and more people are striving to narrow the gender gap, especially in the fields of STEM. And since Wing redefined computational thinking (CT) [3], now CT has been regarded as the core of all modern science, technology, engineering, and mathematics disciplines [4] and is emphasized by educators and scholars. Therefore, interventions aimed at addressing gender disparities in computational thinking skills have gained significant attention in recent years. Recognizing the importance of fostering gender equity in STEM fields, educators and scholars hope to develop certain educational interventions on cultivating and improving computational thinking skills among students in order to narrow the potential gender gaps and promote the entry of female students into STEM fields, including computer science, in the future [5].

Some studies suggest that the imbalance in gender in the STEM field is not only due to gender stereotypes [6] but also to differences in abilities and performance [7,8]. While responding to the wave of gender equality, the objective physiological differences between men and women cannot be ignored. Research has shown that the weight of the men’s brain is 10–15% heavier than that of the women’s brain; men’s inferior parietal lobes, which control spatial and mathematical reasoning, are more developed than those of women’s, while women have a more developed left side of the brain, which controls language and writing function [9]. These studies lead to a question: will these physiological gender differences reflect in the acquisition of computational thinking skills?

Regarding these issues, many evidence-based empirical studies have been conducted in recent years. Therefore, this systematic review aims to explore and provide a comprehensive overview of the existing literature on gender differences in computational thinking skills among primary and secondary school students. By examining the research findings and identifying the effective interventions contributing to gender equality in computational thinking skills, this review will provide insights into future research. On this basis, this review proposes the following three specific research questions:

RQ 1: Which geographical regions and age groups are primarily examined in studies on gender differences in primary and secondary students’ computational thinking skills?

RQ 2: Which specific aspects have been examined, and which instruments have been adopted for assessment in studies on gender differences in students’ computational thinking skills?

RQ 3: What significant gender differences in computational thinking skills among primary and secondary school students were reported in these studies?

2. Methodology

2.1. Searching Strategy

As Web of Science (WOS) is a wide-coverage database that covers most research areas, including education, science, social sciences, etc., this systematic review selected it as the source for literature retrieval to ensure the search has included relevant articles. “Computational thinking” and “gender” are used as search terms, and the Boolean operator “AND” is used to search for the title, abstract, and keywords in order to refine the search and exclude irrelevant articles. Due to Wing’s first introduction of the term computational thinking in 2006 [3], the publication time range of literature search has been refined from 2006 to 2023.

As Bradford’s Law points out, most of the significant scientific findings have been reported in journal articles [10], and also considering the necessity of peer review to ensure the quality of articles, this systematic review only includes journal articles—other document types such as proceeding papers are excluded.

2.2. Selection Criteria

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) approach is adopted in the phase of selection of articles in this review [11,12]. According to the research topic and specific research questions, this review has developed specific inclusion and exclusion criteria for study selection. The criteria are shown as follows.

Inclusion criteria:

• Studies related to gender differences in computational thinking skills;
• Studies that clearly indicated the age of participants and the regions of the studies;
• Studies focus on primary and secondary school students;
• Empirical studies that report quantitative/qualitative data on gender differences in computational thinking skills;
• Studies published in peer-reviewed journals.

Exclusion criteria:

• Studies that do not explicitly investigate computational thinking skills or related concepts;
• Studies focus beyond primary and secondary school settings, such as college students and teachers;
• Studies that do not address gender differences in computational thinking skills;
• Non-peer-reviewed publications, such as proceeding papers;
• Studies written in non-English languages.

Figure 1 summarizes the process under the framework of PRISMA.

2.3. Data Extraction and Analysis

Based on the framework of PRISMA and the selection criteria, 197 studies are identified from the database. Then, 123 studies were excluded in the screening phase based on their irrelevant titles. In eligibility phase, the full contents of the 74 articles were carefully examined, resulting in the exclusion of 51 articles. Finally, 23 articles were retained in the review scope (shown in Appendix A).

Then, the information from these retained articles was extracted, including the metadata of the articles (author, publication time, title), research objects (age, region characteristics), instruments, and research topics. In order to extract information more effectively and efficiently, a coding framework has been developed (see

Table 1. The coding framework.

Research Questions	Category	Code
RQ 1	Metadata	Author, year of publication, title
	Research objects	Participants’ age group, geographical regions
RQ 2	Instruments	Instruments to assess computational thinking skills: Bebras, CTt, etc.
RQ 3	Research topics	Research topics: 1—Inherent gender differences in computational thinking skills. 2—Current situation of gender differences in computational thinking skills. 3—Gender differences in the interventions of computational thinking skills.

). The coding for the age of the objects covered from grade 1 to grade 12; the coding for the region of the objects included Turkey, mainland China, Taiwan, China, Korea, America, Spain, Singapore, Indonesia, and Greece. In terms of instruments, the codes included Bebras, CTt, self-designed instruments, TechCheck-2, Scale of Computational Thinking Skill Levels, and Mathematics Problems Solving Test. As for research topics, the codes included inherent gender differences in computational thinking skills, current situation of gender differences in computational thinking skills, and gender differences in the interventions of computational thinking skills.

3. Results

3.1. Geographical Characteristics of the Research

By extracting information from 23 articles within the review scope, this review summarizes the regions in which these studies were conducted (see Figure 2). Turkey and mainland China are the regions where gender differences in CT skills among primary and secondary school students are most studied, and six studies have been conducted, respectively. Two studies were conducted in South Korea, America, Spain, and Taiwan, China, respectively, and only one in each other region (Singapore, Indonesia, Greece, and America). In terms of research topics, the most abundant region is mainland China, and other regions, except Turkey and Korea, only focused on one research topic. The region with the most research on topics 1 and 3 is mainland China; the region with the most research on topic 2 is Turkey.

3.2. Focuses of Age Groups in the Research

This review summarizes the age groups in each study in the following figure (Figure 3) and table (

Table 2. The frequency of each age group in the research.

Grade	Frequency	Grade	Frequency
Grade 6	11	Grade 1	2
Grade 5	7	Grade 2	2
Grade 7	6	Grade 3	2
Grade 9	5	Grade 10	1
Grade 4	3	Grade 12	1
Grade 8	3	Grade 11	0

). The horizontal axis in the figure represents different studies, while the vertical axis represents age groups in each study. In general, the results show that most studies focus on grades 5, 6, and 7, while there is a research gap in higher grades (grade 10, 11, 12) and lower grades (grade 1, 2, 3), and grade 11 has not yet been studied yet.

Investigating age groups according to research topics (see Figure 4), the results show that there are still many research gaps in research topic 1 (inherent gender differences in CT skills). Currently, only one study has been conducted on grades 4, 5, and 6. Research topics 2 and 3 (current situation of gender differences in CT skills and gender differences in the interventions of CT skills) exhibit a pattern similar to the overall pattern: the number of studies focusing on the higher and lower grades is relatively small.

3.3. The Employment of Instruments in the Research

To answer RQ2, this review summarizes the instruments used for the assessment of computational thinking skills in the studies within the review scope (see Figure 5). The results showed that the Bebras test (including the adapted test) was the most frequently used instrument [13], with a total of eight studies using it. Next is the Computational Thinking Test (CTt) developed by Román-González [14], which has been applied in six studies. While five studies developed their own instrument, the rest of the studies adopted TechCheck-2 [15], the Scale of Computational Thinking Skill Levels [16,17], and the Mathematics Problems Solving Test [18].

To answer which specific aspects of computational thinking skills were measured and studied, this review summarizes the dimensions of these instruments in the table below (

Table 3. The assessed dimensions of computational thinking corresponding to the instruments.

Instrument	Dimensions	Instrument	Dimensions
Bebras test [13].	Abstraction; Decomposition; Algorithmic; Evaluation; Generalization.	Computational Thinking Test (CTt) [14].	Basic directions and sequences; Loops—repeat times; Loops—repeat until; If—simple conditional; If/else—complex conditional; While—conditional; Simple functions.
Jiang and Wong’s self-designed instrument [19].	Conditionals; Logical operators; Pattern recognition; Generalization.	Atmatzidou and Demetriadis’s self-designed instrument [20].	Abstraction; Generalization; Algorithm; Modularity; Decomposition.
Wu and Su’s self-designed instrument [21].	Decomposition; Pattern recognition; Abstraction; Algorithm design.	Kim et al.’ self-designed instrument [22].	Abstraction (problem-solving, pattern analysis, algorithm design); Automatization (algorithm implementation, structural programming, debugging).
Kwon et al.’s self-designed instrument [23].	Identifying the meaning of an individual code; Predicting the results of multiple codes listed in sequence; Fixing codes when the intended outcome is not achieved (debugging).	TechCheck-2 [15].	Algorithmic thinking; Modular structure; Control structures; Representation; Software/hardware (identifying technological concepts); Debugging.
Scale of Computational Thinking Skill Levels [16,17].	Creativity; Algorithmic thinking; Collaboration; Critical thinking; Problem-solving.	Mathematics Problems Solving Test [18].	Not applicable.

). The results show that although the dimensions measured by the instruments overlap in some aspects, there are also differences. This is because computational thinking has not been clearly defined, and different scholars have constructed their assessment instruments based on different conceptual frameworks of computational thinking. The selection of instruments reflects that the studies focused on different aspects of skills in computational thinking.

3.4. Gender Differences in Computational Thinking Skills

Existing research has investigated gender differences in computational thinking skills among primary and secondary school students from three perspectives:

• Inherent gender differences in computational thinking skills;
• Current situation of gender differences in computational thinking skills;
• Gender differences in the interventions of computational thinking skills.

The last category can be further divided into the following:

• Gender differences in computational thinking skills exhibited during the CT task process;
• The effects of interventions on computational thinking skills on different genders.

This review summarizes the 23 articles within the review scope from these aspects to answer RQ3.

3.4.1. Inherent Gender Differences in Computational Thinking Skills

Computational thinking, as an innate skill, has been studied by Jiang and Wong [19]. Jiang and Wong excluded the interference of previous education on computational thinking concepts and skills, aiming to explore the inherent gender differences in these innate skills. They designed multiple-choice items and open-ended item sets to evaluate the computational thinking skills of students in grades 4–6 in China from four dimensions: conditionals, logical operators, pattern recognition, and generalization. The results indicate that there is no significant gender effect or significant interaction effect between gender and age in both the overall level of CT skills and the level of each dimension of CT skills [19].

3.4.2. Current Situation of Gender Differences in Computational Thinking Skills

Some studies have not ruled out the influence of students’ prior education/instruction on computational thinking; therefore, they do not reveal the inherent gender differences in computational thinking skills. Instead, these studies reveal the current situation of gender differences in computational thinking skills exhibited by students after exposure to the existing educational environment, which is often controlled by regional factors. Due to the fact that the students involved in each study may not have had the same educational background, the research results exhibit diverse differences. Therefore, this review summarizes the results of each study by region as follows. Unexpectedly, this review found that female students in culturally patriarchal East Asian regions exhibited higher overall levels of computational thinking skills (despite differences in some dimensions of CT skills); in West Asia, most studies indicate that currently, there is no gender difference between male and female students; as for Southeast Asia, male students in Singapore have higher levels of computational thinking skills than female students; while in Indonesia, female students exhibit higher levels of computational thinking skills in the context of solving mathematics problems.

Current Situation in Turkey

Four studies have investigated the current situation of gender differences in the computational thinking skills of Turkish students. These four studies focused on grades 3–4 [24], grades 5–6 [25,26], and grades 6–8 [27], respectively. Except for Küçükaydın and Çite, who used TechCheck-2 to assess students’ computational thinking skill levels [24], the remaining three studies used CTt for assessment [25,26,27]. Three of these studies indicate that there is no gender difference in computational thinking skills among students (grades 3–8) [25,26,27]. The interpretation of Demir-Kaymak et al. on this result is that the students involved in the study had previously received different interventions related to computational thinking [27]. However, these three studies did not report whether male and female students exhibited differences in specific dimensions of computational thinking skills.

However, contrary to the results of these three studies, Polat et al.’s study on students in grades 5–6 showed that male students had significantly higher scores than female students in terms of the overall average level of computational thinking skills. After dividing by age, the average level of CT skills among sixth-grade male students is significantly higher than that of female students, while the average score among fifth-grade male students is slightly higher than that of females, but there is no significant statistical difference. A specific analysis of the sub-dimensions of CT skills shows that male students have significantly higher levels than females in the following sub-dimensions: visual blocks, textual; the maze, with coding nesting; sequencing; completion; loops—repeat until; if—simple conditional; if/else—complex conditional; and the while conditional [26]. This result is inconsistent with Atman’s research [25], which also focused on students from grade 5 to grade 6, but Polat et al. mentioned in the article that their research was conducted in Istanbul, the largest city in Turkey [26]. In this context, students may have been affected differently in their educational experiences.

Current Situation in Mainland China

A study on seventh-grade Chinese students used the Bebras computational thinking test to measure their computational thinking skills [28]. This study indicated that female students have significantly higher computational thinking skills than male students, but their future development in computational thinking skills may be influenced by their more negative attitudes toward programming.

Current Situation in Korea

A study in Korea investigated two aspects of computational thinking skills among students in grades 4–9: abstraction (including problem-solving, pattern analysis, and algorithm design) and automation (including algorithm implementation, structural programming, and debugging) [22]. Their research found that female students in secondary schools performed significantly worse than male students in automation, while they performed significantly better in abstraction, and female students in primary schools exhibited higher levels in both aspects compared to male students.

Current Situation in Taiwan, China

Two studies in Taiwan focus on the differences in computational thinking skills developed by students of different educational backgrounds.

Lee et al.’s study assessed the computational thinking skills among students in grades 5–6 and pointed out the exhibited gender differences in school environments with different resources: in schools with relatively scarce information and communication technology resources, male students exhibited significantly lower levels of computational thinking skills than female students, while no significant gender difference was found in schools with abundant information and communication technology resources [29].

Wu and Su focused on the existing gender differences in computational thinking skills among students of different genders in grades 5–6 in different program learning environments from four aspects: decomposition, pattern recognition, abstraction, and algorithm design [21]. They selected students who have been studied in three different programming environments: students with a background in learning in visual programming environments (VPE group), students with a background in using robots for learning in visual programming environments (VPERobot group), and students who had never studied in visual programming environments (NoVPE group). The results showed that the overall level of computational thinking skills among female students in the NoVPE group was significantly higher than that of male students; in the VPE group, although male students have significantly higher scores in terms of algorithm design skills, female students have a significantly higher overall level of computational thinking skills than male students; as for VPERobot group, no significant gender difference was found.

Current Situation in Singapore

Students in grades 9–10 in Singapore participated in Chan et al.’s study [30], and CTt developed by Román-González was adopted to assess students’ computational thinking skills in this study [14]. Researchers divided students into four groups based on the average and standard deviation of logit value person (LVP): “very high ability of CT”, “high ability of CT”, “moderate ability of CT”, and “low ability of CT”. The distribution of male and female populations in these four groups is shown in

Table 4. The distribution of male and female populations [14].

Very High Ability of CT		High Ability of CT		Moderate Ability of CT		Low Ability of CT
Male	Female	Male	Female	Male	Female	Male	Female
22%	0%	40%	7%	27%	55%	11%	38%

.

The results showed that the majority of male students were in the moderate-to-very-high-ability group, while the majority of female students were in the moderate-to-low-ability group. Male students had higher CT skill levels than female students.

Current Situation in Indonesia

In a study on ninth-grade students in Indonesia [18], researchers used the mathematics problem-solving test to measure students’ computational thinking skills in the context of solving mathematical problems in order to investigate the existing gender differences in computational thinking skills among Indonesian students. Female students were found to have significantly higher CT skill levels in solving mathematical problems than male students, but due to the limitations of the assessment instrument, this study did not report the gender differences in specific aspects of computational thinking skills.

3.4.3. Gender Differences in Computational Thinking Skills Exhibited during CT Task Process

Ardito et al. designed a LEGO robotics program that included three challenges for grade 6 students—Boat Building, Robot Racing, and Dance Off—aiming to develop students’ three core computational thinking skills: building, programming, and problem-solving [31]. Unlike other studies, this study adopted a qualitative approach and collected students’ diaries throughout the entire robotics program, which recorded the whole task process. The analysis of diary content indicates that during the process of the robotics program, male students focus more on operational aspects (building and programming); on the contrary, female students are more concerned about problem-solving and group dynamics regarding collaboration skills.

Atmatzidou and Demetriadis focus on the five core dimensions of computational thinking skills: abstraction, generalization, algorithm, modularity, and decomposition [20]. Their study has designed 11 sessions on the Lego Mindstorms NXT 2.0 educational robotics kit for grade 9 and 12 students and adopted a project-based learning approach. Based on the two intermediate questionnaires filled out by students after the fourth and tenth sessions, this study found that although both male and female students ultimately achieve the same level of computational thinking skills, female students need to spend more time practicing to achieve the same level as male students. This phenomenon is reflected in the overall level of computational thinking skills, abstraction, generalization, and modularity, while there is no significant difference in algorithm and decomposition.

3.4.4. The Effects of Interventions on Computational Thinking Skills on Different Genders

This review summarizes the interventions aiming at cultivating and improving students’ computational thinking skills in the reviewed studies in Table 5. While 60% of the studies reported that no significant gender differences were found in the improvement of CT skills, this review extracted several factors with gender effects in the interventions from the remaining 40% of studies through multiple comparisons:

• Gamified Python programming (text-based coding) with 5E pedagogical model [32]:
  The combination of these two intervention factors has provided a higher improvement of abstraction and decomposition (AD) skills among female students and a higher improvement of pattern recognition and evaluation (PE) skills among male students in grade 6;
• Programming instruction in problem-solving pedagogy [33]:
  Compared to the traditional teacher-centered teaching approach that may slightly reduce students’ computational thinking skill levels, problem-based pedagogical methods effectively improve fifth-grade female students’ critical thinking, algorithmic thinking, and problem-solving skills, which is consistent with Ardito et al.’s study: female students are more focused on problem-solving during the task process [31].
• Traditional teacher-centered programming instructional model [33]:
  Traditional teacher-centered programming instruction results in a slight decrease in girls’ scores, suggesting the necessity of conducting teaching reform and designing new gender-fair interventions;
• Algorithm for motions and Tospaa unplugged coding game [34]:
  Although some of the unplugged activities have been found to have no gender effects in other studies, these two unplugged activities have been identified as more effective in improving the computational thinking skills of male students in grade 6;
• mBot robot programmed by Scratch and Python with discovery learning [35]:
  The difference between mBot and other educational robots that have been proven to have no gender effect is that it is programmed by Scratch and Python, which may result in a higher improvement of CT skills among male students in grade 7.

Table 5. Description of the interventions.

No.	Age Group	Duration	Tool Type	Description	Effect
7	Grade 5–6	11 sessions (conducted once a week)	Plugged (Entry and Hamster robot programming)	Teaching philosophy: four-component instructional design (4C/ID), creative problem-solving (CPS) model, a strategy to promote divergent thinking, and collaborative learning. Teaching content: orientation, sequence, iteration, selection, debugging.	No significant gender differences were found in the improvement of CT skills [36].
8	Grade 6	13 weeks	Plugged (gamified Python programming)	Teaching philosophy: 5E pedagogical model was adopted in the lessons. Teaching content: basic Python programming syntax, functions, algorithms, while loops, variables, and other concepts.	Although both genders can achieve the same level of CT skills, the sub-skills of CT exhibit gender differences: the progress in PE skills was significantly higher in males than in females, while the progress in AD skills was significantly higher in females than in males [32].
9	Grade 4–6	5 lessons (40 min each, conducted once a week)	Mix (unplugged and programming activities)	Teaching philosophy: constructionism and embodied cognition theory. Teaching content: learning activities from the Barefoot Computing Project, Code.org, and CS Unplugged.	No significant gender differences were found in the improvement of CT skills [19].
13	Grade 5	14 lessons (90 min each, conducted once a week)	Plugged (Scratch)	Teaching philosophy: Experimental group: problem-based learning (as part of the adapted IGGIA framework). Control group: traditional (teacher-centered) programming instructional model, in which students engaged in passive learning. Teaching content: interdisciplinary problems integrated with CT process and procedures.	Programming instruction embedded with problem-solving pedagogy can significantly improve girls’ CT skills and has a significant positive impact on their critical thinking, algorithmic thinking, and problem-solving skills, while traditional programming instruction results in a slight decrease in girls’ scores [33].
17	Grade 2	5 lessons (45 min each, conducted once a week)	Plugged/unplugged (plugged: Code.org; unplugged: graph paper programming, cups stacking)	Teaching content: a selection of activities extracted from Code.org. Teaching process: Phase 1: one group worked with unplugged activities, and the other worked with plugged activities; Phase 2: both groups worked with plugged activities.	Male students have experienced greater progress, but this difference is not statistically significant [37].
18	Grade 1–2	5 lessons (35–45 min each)	Unplugged (Bee-Bot)	Teaching philosophy: embodied learning (allows students to actively engage in full-body locomotion that simulates robots’ spatial information). Teaching content: a series of path-finding problems.	No significant gender differences were found in the improvement of CT skills [23].
19	Grade 6	unrevealed	Unplugged	Teaching philosophy: students worked in groups or individually, and the instructor worked as facilitator. Teaching content: designed from multiple-choice Bebras tasks.	No significant gender differences were found in the improvement of CT skills [38].
20	Grade 6	6 weeks	Plugged/unplugged (Plugged: Scratch, Code.org; unplugged: algorithm for motions; Tospaa unplugged coding game)	Teaching content: loop structure. Experiment group 1: Code.org. Experiment group 2: unplugged activities. Control group: Scratch.	The unplugged activities are significantly more beneficial for males, while no significant gender differences were found in the improvement of CT skills in the other two groups [34].
21	Grade 3, 7	5 sessions (4 h each)	Plugged (Scratch, Code.org, Code & Go robot, mBot robot)	Teaching philosophy: guided learning and discovery learning. Teaching content: Primary guided learning group: Code & Go robot, with activities from Code.org. Primary discovery learning group: design a guitar on Scratch. Secondary guided learning group: activities from Code.org; students simulated a Pong game in Scratch. Secondary discovery learning group: mBot robot circuit.	No significant gender differences were found in the improvement of CT skills in primary students, while secondary male students experienced significantly greater improvement [35].
22	Grade 7	8 lessons (40 min, conducted once a week)	Plugged/unplugged/mix (Plugged: Code.org; unplugged: graph paper programming, cups stacking, loop game on map, paper exercises)	Teaching philosophy: 5E pedagogical model and 4P learning theory. Plugged group: 8 plugged activities. Unplugged group: 8 unplugged activities. Plugged-first group: 4 plugged activities + 4 unplugged activities. Unplugged-first group: 4 unplugged activities + 4 plugged activities.	No significant gender differences were found in the improvement in different programming approaches [39].
23	Grade 1	8 lessons (40 min, conducted once a week)	Plugged/unplugged/mix	Plugged group: 8 plugged activities. Unplugged group: 8 unplugged activities. Plugged-first group: 4 plugged activities + 4 unplugged activities. Unplugged-first group: 4 unplugged activities + 4 plugged activities.	No significant gender differences were found in the improvement in different programming approaches [40].

Table 5. Description of the interventions.

No.	Age Group	Duration	Tool Type	Description	Effect
7	Grade 5–6	11 sessions (conducted once a week)	Plugged (Entry and Hamster robot programming)	Teaching philosophy: four-component instructional design (4C/ID), creative problem-solving (CPS) model, a strategy to promote divergent thinking, and collaborative learning. Teaching content: orientation, sequence, iteration, selection, debugging.	No significant gender differences were found in the improvement of CT skills [36].
8	Grade 6	13 weeks	Plugged (gamified Python programming)	Teaching philosophy: 5E pedagogical model was adopted in the lessons. Teaching content: basic Python programming syntax, functions, algorithms, while loops, variables, and other concepts.	Although both genders can achieve the same level of CT skills, the sub-skills of CT exhibit gender differences: the progress in PE skills was significantly higher in males than in females, while the progress in AD skills was significantly higher in females than in males [32].
9	Grade 4–6	5 lessons (40 min each, conducted once a week)	Mix (unplugged and programming activities)	Teaching philosophy: constructionism and embodied cognition theory. Teaching content: learning activities from the Barefoot Computing Project, Code.org, and CS Unplugged.	No significant gender differences were found in the improvement of CT skills [19].
13	Grade 5	14 lessons (90 min each, conducted once a week)	Plugged (Scratch)	Teaching philosophy: Experimental group: problem-based learning (as part of the adapted IGGIA framework). Control group: traditional (teacher-centered) programming instructional model, in which students engaged in passive learning. Teaching content: interdisciplinary problems integrated with CT process and procedures.	Programming instruction embedded with problem-solving pedagogy can significantly improve girls’ CT skills and has a significant positive impact on their critical thinking, algorithmic thinking, and problem-solving skills, while traditional programming instruction results in a slight decrease in girls’ scores [33].
17	Grade 2	5 lessons (45 min each, conducted once a week)	Plugged/unplugged (plugged: Code.org; unplugged: graph paper programming, cups stacking)	Teaching content: a selection of activities extracted from Code.org. Teaching process: Phase 1: one group worked with unplugged activities, and the other worked with plugged activities; Phase 2: both groups worked with plugged activities.	Male students have experienced greater progress, but this difference is not statistically significant [37].
18	Grade 1–2	5 lessons (35–45 min each)	Unplugged (Bee-Bot)	Teaching philosophy: embodied learning (allows students to actively engage in full-body locomotion that simulates robots’ spatial information). Teaching content: a series of path-finding problems.	No significant gender differences were found in the improvement of CT skills [23].
19	Grade 6	unrevealed	Unplugged	Teaching philosophy: students worked in groups or individually, and the instructor worked as facilitator. Teaching content: designed from multiple-choice Bebras tasks.	No significant gender differences were found in the improvement of CT skills [38].
20	Grade 6	6 weeks	Plugged/unplugged (Plugged: Scratch, Code.org; unplugged: algorithm for motions; Tospaa unplugged coding game)	Teaching content: loop structure. Experiment group 1: Code.org. Experiment group 2: unplugged activities. Control group: Scratch.	The unplugged activities are significantly more beneficial for males, while no significant gender differences were found in the improvement of CT skills in the other two groups [34].
21	Grade 3, 7	5 sessions (4 h each)	Plugged (Scratch, Code.org, Code & Go robot, mBot robot)	Teaching philosophy: guided learning and discovery learning. Teaching content: Primary guided learning group: Code & Go robot, with activities from Code.org. Primary discovery learning group: design a guitar on Scratch. Secondary guided learning group: activities from Code.org; students simulated a Pong game in Scratch. Secondary discovery learning group: mBot robot circuit.	No significant gender differences were found in the improvement of CT skills in primary students, while secondary male students experienced significantly greater improvement [35].
22	Grade 7	8 lessons (40 min, conducted once a week)	Plugged/unplugged/mix (Plugged: Code.org; unplugged: graph paper programming, cups stacking, loop game on map, paper exercises)	Teaching philosophy: 5E pedagogical model and 4P learning theory. Plugged group: 8 plugged activities. Unplugged group: 8 unplugged activities. Plugged-first group: 4 plugged activities + 4 unplugged activities. Unplugged-first group: 4 unplugged activities + 4 plugged activities.	No significant gender differences were found in the improvement in different programming approaches [39].
23	Grade 1	8 lessons (40 min, conducted once a week)	Plugged/unplugged/mix	Plugged group: 8 plugged activities. Unplugged group: 8 unplugged activities. Plugged-first group: 4 plugged activities + 4 unplugged activities. Unplugged-first group: 4 unplugged activities + 4 plugged activities.	No significant gender differences were found in the improvement in different programming approaches [40].

4. Discussion

This systematic review explores the existing literature on gender differences in computational thinking skills among primary and secondary school students in different aspects and identifies the effective factors in the interventions contributing to gender equality in computational thinking skills.

This review divides the studies on gender differences in computational thinking skills into four categories: (1) studies on inherent gender differences in computational thinking skills; (2) studies that investigated the current situation of gender differences in computational thinking skills; (3) studies focused on the gender differences exhibited during CT task process; and (4) studies that explored the effects of interventions on computational thinking skills on different genders.

First of all, this review finds that Turkey and mainland China are the regions that have the most research on gender differences in computational thinking skills. The research in China mainly focuses on the interventions of gender differences in computational thinking skills, while the research in Turkey mainly focuses on the investigations of existing gender differences in computational thinking skills, and there remains a large gap in the research of inherent gender differences in computational thinking.

This review also found an imbalance in the focus of current gender-related research on age groups, with a primary concentration on the middle grades, while there is a relative lack of research on both lower and higher grade levels. The lack of research on lower grade and higher grade levels may be attributable to practicality factors. For lower-grade students, their reading ability, comprehension skills, and attention may limit their performance, and the implementation of the assessments [15] and additional management may be required during the process; for higher-grade students, they are facing the stress associated with academic advancement and college admissions. However, the lack of research on lower grade and higher grade levels may hinder the identification of timely equality-based intervention requirements for students. Research indicates that gender differences in certain cognitive abilities, such as mental rotation, which have been shown to be positively correlated with computational thinking skills (and this relationship is not influenced by gender) [41], begin to emerge in the early age range [42]. Meanwhile, higher-grade students are about to face the decision of whether to pursue STEM majors/careers. Computational thinking skills, as a potential factor that may influence their choices, also need to be examined for gender differences for the purpose of promoting a more gender-equal STEM community.

In terms of instruments assessing computational thinking skills in these gender-related studies, this review found that, first of all, the CT domains covered by instruments developed under different conceptual frameworks vary, and instruments with limited coverage or less detail may not be able to detect subtle gender differences. Secondly, the correlation between different tools also needs to be examined. Currently, only TechCheck has been reported to have a moderate correlation of r = 0.53, p < 0.001, with another instrument, TACTIC-KIBO [15]. It is necessary to further validate whether students at the same level and students of the same gender perform consistently across different evaluations, especially with self-design instruments. Additionally, as mentioned above, students’ reading skills are considered to be potentially correlated with their scores in the assessment, especially for younger students, as their literacy level may affect their understanding of the questions. Research indicates that girls generally demonstrate higher language skills compared to boys, which is attributed to influences from brain structure and function, as well as social factors [43]. Currently, the majority of the items in the instruments used for research on lower-grade students still rely on text-based prompts. Whether female students’ advantage in language fluency, vocabulary, and reading comprehension abilities also translates into benefits for them in the assessment requires further examination. Another important point to consider is that instruments like TechCheck have reported racial/ethnic-based differences [15]. Therefore, caution should be exercised when generalizing their application to study gender differences in other regions. In order to address these issues, the development of a clearer and more specific definition of computational thinking and corresponding assessment instruments is required.

In the past, some scholars believed that there were conflicting research results regarding gender differences in students’ computational thinking skills [24]. However, after classifying and summarizing the research in this review, it was found that the so-called “conflicting results” may be due to neglect and confusion in distinguishing between the first and second types of studies. It is difficult to identify the essential pattern of gender differences in computational thinking skills in this situation. Without excluding interfering variables, the gender differences observed in such investigations are the result of the influence of multiple factors, especially prior education. Within the review scope, only Jiang and Wong noticed this issue and focused on the inherent gender differences in computational thinking skills prior to CT education [19]. But their research only focused on students in grades 4–6 in China; the inherent gender differences in CT skills in different age groups and regions still need further exploration. Although the second type of research fails to reveal essential gender differences, they reflect the computational thinking levels achieved by students of different genders in educational environments across different regions. Combining the results of the first type of studies, they could reflect the advantages and disadvantages of local computational thinking education in the context of gender equality to some extent. According to the findings of the second type of study, the gender differences in CT skills among primary and secondary school students vary across different regions, and unexpectedly, female students in culturally patriarchal East Asian regions exhibited higher overall levels of CT skills. Further research is needed to identify what specific factors are contributing to this phenomenon. On the other hand, these investigations also indicated that female students have the potential to achieve the same or even higher levels of computational thinking skills as male students, which is an encouraging result for gender equality issues in education and STEM-related areas.

From a neurological perspective, functional magnetic resonance imaging (fMRI) result shows that the application of computational thinking skills recruited distributed cortical regions (including the posterior parietal cortex, the medial frontal cortex, and the left lateral frontal cortex) [44]. Research also indicates that males possess smaller parietal lobe gray-matter-to-white-matter ratios (GM/WM) than females, resulting in better performance in mental rotation tasks (which is considered to be closely associated with computational thinking skills) [41,45]. This neurological characteristic does not align with the current situation of gender differences in computational thinking skills observed in most regions summarized in this review, implying the influence of other factors (especially environmental factors). Another neurological indicator, the parietal lobe surface area, which positively correlated with the performance of psychological rotation tasks, exhibits a pattern of no significant gender difference during adolescence, which then develops into a situation where females are significantly disadvantaged in adulthood, also indicating the influence of the environment [45]. The collective evidence consistently points to the crucial role of interventions.

Regarding interventions aimed at improving CT skills, first of all, the traditional teacher-centered pedagogical approach has been confirmed to have a negative effect on female students’ computational thinking skills [33], indicating that teaching reform related to CT education is imperative. And two patterns have been found and can guide the design of future interventions: (1) Female students need to spend more time to achieve the same level of computational thinking skills as male students. This pattern is also reflected in the results of the interventions, where all interventions that yielded greater benefits for females had a duration of no less than 13 weeks. The longer learning time may be attributed to the inherent physiological characteristics, such as higher GM/WM. (2) During the task process, female students are more concerned with problem-solving and collaboration, while male students are more concerned with operational aspects. Aligned with the pattern, problem-based learning approaches have been confirmed that it can effectively improve female students’ critical thinking, algorithmic thinking, and problem-solving skills [33]. Through multiple comparisons, this review also identified several intervention measures with gender effects, such as the algorithm for motions and the Tospaa unplugged coding game. However, some of the interventions are difficult to distinguish whether the gender effects are attributed to teaching tools, teaching philosophy, or the combination of the two factors, such as gamified Python programming with a 5E pedagogical model and mBot robot with discovery learning. The research of Ma et al. can be imitated by changing the teaching philosophy based on established teaching tools or by changing teaching tools based on established teaching philosophy to determine the specific contributing factors. In addition to cognitive factors, this review also considers learning motivation as a potential factor that may influence the effectiveness of interventions. The validated approach of problem-based learning empowers students with the autonomy to control their learning process and encourages their active participation in appropriately challenging problem-solving activities through collaborative group work. From the perspective of Self-Determination Theory (SDT), this teaching philosophy successfully caters to students’ basic psychological needs: the need for autonomy, competence, and relatedness—the fulfillment of which promotes an increase in their motivation [46]. To address the current situation of female students being less motivated [47,48], such interventions may have the potential to rectify this unfavorable situation. Similarly, the design and implementation of other teaching philosophies or teaching tools should also take into consideration the influence of such affective factors. For the interventions that did not show significant gender effects, it is appropriate to promote and implement them to a certain extent in regions where gender differences in students’ computational thinking skills have not been found (such as Turkey).

5. Future Research

In general, further research is still needed to explore and better understand the gender differences in computational thinking skills. As this review points out, there is still insufficient research on the inherent gender differences in computational thinking skills, and the ultimate goal of gender research is to promote gender equality through practical actions; this review proposes frameworks and guidance for future research on these two specific topics based on the findings of this review.

As stated, there is still a large gap in the study of inherent gender differences in computational thinking skills. Future research can draw on the research methods of Jiang and Wong to screen students in a certain region who have not received education related to CT for research [19]. However, screening out students who never received CT-related education in different regions and age groups can be a challenging task, particularly in areas where CT is widely valued and popularized. Given the diverse regional contexts, variables including education factors, cultural factors, parenting factors, race, etc., could serve as moderating variables, and multi-group comparative analysis across different age groups should be conducted so that an in-depth exploration into the inherent gender differences in computational thinking skills can be pursued (see Figure 6).

Currently, the gender differences in computational thinking skills exhibited by students in different regions are formed by the inherent differences and the differences during the CT task process under existing educational interventions. By utilizing the aforementioned framework, inherent differences could be identified. Combining the findings of the differences exhibited during the CT task process, gender-difference-oriented educational interventions could be designed. Also, based on the comparison between the current differences and the inherent differences in CT skills, factors that contribute to the gender effect in current interventions could be identified, thereby promoting the design of new interventions targeting gender equality in CT-related education. Meanwhile, the design of interventions also needs to incorporate relevant factors from other studies, such as those related to motivation and identity. The framework for future research is shown in Figure 7.

And as for the research objects, future studies can focus on higher grades (grades 10, 11, and 12) and lower grades (grades 1, 2, and 3) to fill the current research gap and expand the research area to the Americas, Europe, Africa, and other regions in a global perspective.

On a deeper level, there remains a need to investigate the physiological factors that underlie the potential gender differences in computational thinking skills, as well as the underlying mechanisms behind the differential effects observed among different genders under specific interventions. Physiological studies involving neuroscience, neurology, etc., would provide stronger evidence and reveal the essential causes of potential intrinsic gender differences in computational thinking skills.

6. Conclusions

To conclude, this systematic review synthesized current research findings and identified several critical issues and gaps in the existing research landscape, including the noticeable imbalance in research across different regions and age groups and the insufficiency in the studies on inherent gender differences in computational thinking skills; the requirement for the development of a clearer and more specific assessment instrument for a more detail identification of gender differences in computational thinking skills; the regular pattern of gender differences in computational thinking skills among students across different regions under the existing educational environments; and the effective interventional factors contributing to gender-equitable CT education. Future research is encouraged to address the identified imbalance and insufficiency, and corresponding frameworks are also proposed in this review for future reference.