Treffer: Trends and Patterns in K-12 Computer Science Education: Data Analysis from Twitter

K-12 computer science (CS) education has emerged as a vital component of modern education, nurturing computational thinking, problem-solving, and digital literacy. This study examines the K-12 CS education dynamics, emphasizing its impact and implications, particularly in the context of equity. Twitter data from 2017 to 2021 were collected, focusing on English-language tweets within the United States. This collection was completed before Elon Musk's acquisition of the company and its subsequent rebranding to X. Three keyword sets span CS education, computational thinking -- a core competency of CS learners and CS education organizations and conferences. The findings indicate: (1) a significant decrease in tweet volumes for each set of keywords after 2019, (2) the critical role of coding within a broader STEM education framework, and (3) the centrality of students in semantic networks formed by the tweets, highlighting the pertinence of a student-centered learning strategy in K-12 CS education. To ensure equitable access and opportunities, K-12 CS education in a broader STEM ecosystem should adopt student-centered learning, with teachers facilitating coding, programming, and technology education. These insights inform educators, policymakers, and researchers about K-12 CS education's significance in preparing students for the future, with a strong emphasis on equity and inclusion.

As Provided

AN0183372527;5b101mar.25;2025Mar04.04:13;v2.2.500

Trends and patterns in K-12 computer science education: data analysis from twitter 

K-12 computer science (CS) education has emerged as a vital component of modern education, nurturing computational thinking, problem-solving, and digital literacy. This study examines the K-12 CS education dynamics, emphasizing its impact and implications, particularly in the context of equity. Twitter data from 2017 to 2021 were collected, focusing on English-language tweets within the United States. This collection was completed before Elon Musk's acquisition of the company and its subsequent rebranding to X. Three keyword sets span CS education, computational thinking – a core competency of CS learners and CS education organizations and conferences. The findings indicate: (1) a significant decrease in tweet volumes for each set of keywords after 2019, (2) the critical role of coding within a broader STEM education framework, and (3) the centrality of students in semantic networks formed by the tweets, highlighting the pertinence of a student-centered learning strategy in K-12 CS education. To ensure equitable access and opportunities, K-12 CS education in a broader STEM ecosystem should adopt student-centered learning, with teachers facilitating coding, programming, and technology education. These insights inform educators, policymakers, and researchers about K-12 CS education's significance in preparing students for the future, with a strong emphasis on equity and inclusion.

Keywords: K-12 computer science education; Twitter/tweet; descriptive content analysis; CSTIM project; semantic network

Introduction

In recent years, the integration of computer science (CS) education in K-12 schools has gained importance, focusing on computational thinking, problem-solving, and digital literacy. These skills are essential for students to succeed in today's digital world. Equitable access to quality CS education has become a key objective, as evidenced by initiatives such as the CS for All movement, which advocates for widespread, inclusive access to CS education, ensuring that all students, regardless of background, are equipped with the computational skills necessary for success in the 21st century (Lynch et al., [22]).

Social media platforms like Twitter play a crucial role in facilitating discussions about CS education.[1] CS education organizations and conferences, such as International Computing Education Research (@ICER_C), Association for Computing Machinery's (ACM) Special Interest Group on Computer Science Education (@SIGCSE_TS), National Science Foundation (NSF)-funded CS for All Teachers (@CSforAllTchrs), Computer Science for All (@CSforALL), Computer Science Teacher Association (@csteachersorg), and Grace Hopper Celebration (@ghc) actively engage on Twitter, fostering conversations on key trends, challenges, and innovations. These discussions, captured through hashtags and account activity, provide valuable insights into the evolving landscape of CS education (see Figure 1).

Graph: Figure 1. Some CS education related Twitter accounts and conference hashtags.

This study analyzes Twitter data from 2017 to 2021 to identify trends in K-12 CS education, with a focus on computational thinking, coding, and STEM (Science, Technology, Engineering, and Mathematics) integration. By examining discussions from key CS education organizations and educators, it offers insights for educators, scholars, and policymakers on implementing effective K-12 CS programs. The findings also inform efforts to promote equitable access to CS education, preparing students for future careers in technology.

Literature review

The field of K-12 CS education has gained prominence due to the growing recognition of the significance of computational thinking and technological literacy in preparing students for the digital era. This review presents key concepts, research trends, and theoretical frameworks in K-12 CS education.

Importance of K-12 CS education

Integrating CS education in K-12 schools fosters computational thinking skills, including problem-solving and logical reasoning, crucially promoting critical thinking and creativity (Code.org, CSTA, & ECEP Alliance, [7]; CSforALL, [10]). It also cultivates digital citizenship, equipping students with ethical knowledge about privacy and responsible online behavior (Organisation for Economic Co-operation and Development [OECD, [26]]; World Economic Forum [WEF], [33]). Integrating CS education empowers students with computational thinking skills, contributing to their digital literacy and citizenship in the 21st century (Kafai & Proctor, [20]).

Student-centered learning approach

A student-centered approach, prioritizing students' active involvement, is endorsed by the American Psychological Association (American Psychological Association [APA], [1]). Teachers act as facilitators, allowing students to co-create educational standards, identify areas of struggle, and adapt teaching to meet their needs (Brown, [4]; Wright, [34]). In the context of K-12 CS education, this approach strives to nurture intrinsic motivation and empower students as co-creators in the learning process, fostering engagement, computational thinking skills, deeper understanding, and retention of CS concepts (Anwar et al., [2]; Zhang et al., [35]).

Integration of CS education into STEM

K-12 CS education is an integral part of STEM education, promoting collaboration, problem-solving, and innovation, all of which are essential components of computational thinking (Boulden et al., [3]; Lee et al., [21]; Shaby et al., [29]). Various organizations like the National Science Foundation (NSF, see https://beta.nsf.gov/funding/opportunities/stem-computing-k-12-education-stemc) and the U.S. Department of Education (see https://www.ed.gov/stem) support the integration of computing in STEM education. Researchers, practitioners, policymakers, and organizations like the Computer Science Teachers Association (CSTA, see https://www.csforall.org/members/computer%5fscience%5fteachers%5fassociation/) drive the integration of CS in K-12 STEM ecosystems, working to develop strategies and curricula that provide a comprehensive educational experience for students (Boulden et al., [3]; Shaby et al., [29]).

Teacher professional development

Teachers play a critical role in K-12 CS education (Code.org, CSTA, & ECEP Alliance, [7]; World Economic Forum [WEF] , [33]). Teacher professional development programs are vital to equip educators with content knowledge, pedagogical skills, and confidence, which are needed to facilitate computational thinking (Anwar et al., [2]; Israel et al., [19]; Kafai & Proctor, [20]; Veletsianos et al., [31]). Effective teacher professional development promotes student engagement and learning outcomes by nurturing computational thinking (Goode et al., [17]; McGee et al., [23]).

Equity considerations

Equitable access to K-12 CS education is crucial (Fletcher & Warner, [14]; Vakil, [30]). Efforts are being made to address the underrepresentation of certain groups, such as females and minorities (Denner & Campe, [11]; Goode, [16]). Inclusive pedagogical strategies and reducing gender biases contribute to a diverse and inclusive learning environment (Denner & Campe, [11]). Culturally relevant examples, connecting CS to students' experiences, and fostering an inclusive classroom culture promote equitable opportunity to develop computational thinking skills (Denner & Campe, [11]; Fletcher & Warner, [14]; Goode, [16]).

Role of social media in K-12 CS education discourse

When compared to other media platforms, Twitter stands out as a primary choice for academic research due to its robust API (application programming interface) support, enabling researchers to efficiently access and collect vast amounts of data. In the context of K-12 CS education, Twitter serves as the conduit through which discussions on K-12 CS education, computational thinking, and related topics are shared and disseminated. It offers insights into field trends and discussions (Chi et al., [6]) and supports educator collaboration and connection through hashtags (Carpenter et al., [5]; Du et al., [12]). Additionally, it increasingly aligns with educational objectives as school policies evolve to embrace its educational use (Crockett et al., [9]).

However, prior research on the use of Twitter in K-12 education institutions is relatively limited compared to higher education (Wang, [32]). This research aims to fill this gap by examining current trends and patterns in K-12 CS education in the United States. It focuses on the roles of teachers and students and the integration of CS in STEM. Through Twitter data analysis, it provides insights into information dissemination, educator collaboration, and discussions in the K-12 CS education community.

Materials and methods

Data collection

This study received approval from the Institutional Review Boards (IRB) at (university name, blind for review). We collected tweets related to K-12 CS education from January 2017 to December 2021 using the Twitter API. In strict adherence to Twitter's terms of service and privacy policy, it's important to note that no personal information was gathered during this data collection process. Due to the monthly quota limitations imposed on querying Twitter data, the data collection process extended over seven months, spanning from 2021 to 2022. For data retrieval, we employed three separate sets of keywords.

The first set includes six keywords related to CS education in K-12 schools: "k-12 computer science education," "elementary school computer science education," "primary school computer science education," "secondary school computer science education," "middle school computer science education," and "high school computer science education."

The second set stemmed from interviews conducted in 2020 with five K-12 CS teachers and eight CS professors from the Computer Science Teachers in Michigan (CSTIM) project, which explored key aspects of K-12 CS education, including its necessity, core competencies for CS learners, prevailing trends and issues, effective teaching strategies, and teacher competencies in K-12 CS settings (Zhu & Wang, [36], [37]). Our findings underscored that the central goal in K-12 CS education extends beyond programming languages; it revolves around fostering the core competency of problem-solving through computational thinking. Therefore, the second set of keywords includes six terms: "k-12 computational thinking," "elementary school computational thinking," "primary school computational thinking," "secondary school computational thinking," "middle school computational thinking," and "high school computational thinking."

The third set includes keywords related to CS education organizations and conferences, selected for their active role in promoting key conversations on K-12 CS education, professional development, and equity. These influential voices within the CS education community include "@ICER_C," "@SIGCSE_TS," "@CSforAllTchrs," "@CSforALL," "@csteachersorg," and "@ghc." These keywords were combined with "k-12," "elementary school," "primary school," "secondary school," "middle school," and "high school," resulting in a total of 36 keywords.

The flowchart outlining the process of data collection, cleaning, and analysis is depicted in Figure 2.

Graph: Figure 2. Flowchart for data collection and analysis.

Data processing

After executing the queries, we obtained three distinct sets of Twitter data, each containing user IDs, tweet creation timestamps, tweet texts, and geolocations where available. We converted all tweets to lowercase, excluded retweets, and specifically focused on tweets in English from the United States. This approach ensured the study's relevance to the US K-12 CS education context, allowing us to explore nuances, trends, and discussions related to curriculum standards, policy developments, and US-specific educational initiatives. Furthermore, as all six CS education organizations and conferences selected are US-based, this geographical confinement enabled a meaningful comparison of results generated by the three sets of keywords.

The first set of keywords resulted in 6,727 tweets from 4,512 unique users, the second set had 2,769 tweets from 1,411 unique users, and the third set had 1,467 tweets from 613 unique users. There were 85 overlapping cases between the first and second sets, as shown in Figure 3. Stop words, including articles, auxiliary verbs, conjunctions, prepositions, pronouns, question words, less meaningful words, and special HTML character codes, are excluded from the analysis.

Graph: Figure 3. Number of tweets queried by three sets of keywords.

Data analysis

This study uses descriptive content analysis on tweets collected using three sets of keywords. First, word frequencies and their changes are examined over time. Second, word co-occurrence in tweets is investigated. Third, semantic networks are generated from the previous step, creating graphical representations of co-occurring words and their associations (Galoyan et al., [15]; Rafiq et al., [27]). These networks are constructed to reveal and visualize the underlying semantic structure, providing insights into centrality (the prominence of focal words in the network), core and peripheral positions (whether focal words are central or marginal within the network), and community (clusters of words and connections between clusters).

Results

Findings in tweets from the first set of keywords

The first set of keywords focuses on CS education in K-12 schools. As shown in Figure 4, the number of tweets peaked in 2017 and decreased thereafter. The number of tweets in 2021 is only about 70% of that in 2017.

Graph: Figure 4. Numbers of tweets queried by each set of keywords.

Figure 5 displays the top 30 most frequent words in the tweets, with "student(s)" appearing over 2,000 times, followed by "tech," "teacher(s)," "code," "learn," "state(s)," "stem," "program(s)," and "new," all appearing over 500 times. When standardized by the number of tweets (6,727), "student(s)" appears about 31 times, "tech" 19 times, "teacher(s)" and "code" 16 times each, "learn" 14 times, "state(s)" 13 times, "stem" and "program(s)" 11 times each, and "new" 9 times per 100 tweets.

Graph: Figure 5. Top 30 frequent words in tweets queried by the first set of keywords.

Table 1 illustrates rank changes for the top 30 words in tweets from the first query between 2017 and 2021. "Student(s)" maintains the highest frequency throughout the years, and the top eight words consistently remain in the top 10. All top 10 words remain in the top 30, indicating a high level of reliability. Ranks for words 11 to 30 show more variability.

Table 1. The yearly ranks of the top 30 frequent words in tweets queried by the first set of keywords between 2017 and 2021.

Rank	Word	2017	2018	2019	2020	2021
1	student/students	3	1	1	1	1
2	tech	1	2	5	5	4
3	teacher/teachers	12	5	2	3	3
4	code	2	3	4	4	6
5	learn	28	8	8	2	2
6	state/states	9	4	3	7	7
7	stem	10	6	7	8	8
8	program/programs	27	7	6	6	5
9	new	13	9	9	9	11
10	teach	61	11	11	10	10
11	future	5	29	23	17	62
12	class/classes/classroom/classrooms	33	15	10	26	19
13	work/workforce	54	12	13	12	17
14	expand	19	17	32	16	25
15	access	73	10	12	31	20
16	support	20	14	21	34	51
17	csedweek	49	18	50	11	14
18	make	102	13	15	13	26
19	csforall	16	19	20	53	43
20	help	83	16	17	21	33
21	partner/partnership	41	47	47	29	16
22	course/courses	117	27	29	41	12
23	math	129	31	14	32	28
24	fund/funding	11	50	33	46	147
25	take	47	25	19	82	44
26	nation/national	85	26	30	40	23
27	job/jobs	81	52	16	57	21
28	inspire/inspiration	74	53	91	25	15
29	standard/standards	35	22	28	116	80
30	lead/leader/leaders	112	40	27	18	34

The co-appearance analysis of the top 30 words in the 6,727 tweets forms a semantic network with 432 undirected ties. Notable word pairs appearing in over 200 tweets include "student(s)" and "learn" in 419 tweets, "student(s)" and "code" in 394 tweets, "student(s)" and "program(s)" in 287 tweets, "student(s)" and "teacher(s)" in 268 tweets, "student(s)" and "tech" in 265 tweets, "student(s)" and "inspire/inspiration" in 238 tweets, "student(s)" and "csedweek" (computer science education week is "an annual call to action to inspire k-12 students to learn computer science, advocate for equity, and celebrate the contributions of students, teachers, and partners to the field"; see https://www.csedweek.org/) in 221 tweets, and "future" and "tech" in 221 tweets.

To examine the barebone structure of the semantic network, we experimented with different threshold values for word co-appearance frequencies in tweets. Setting the threshold too high would exclude meaningful ties, while setting it too low would result in a noisy graph. The optimal threshold of 110 was selected for Figure 6.

Graph: Figure 6. Barebone structure of semantic network between the top 30 frequent words in tweets queried by the first set of keywords.

The network graph in Figure 6, generated using the Harel-Koren Fast Multiscale layout in NodeXL software, represents the top 30 words (the network graph "zGraph1.xlsx" is available from https://github.com/socnetfan/cstimp3). The lines indicate co-appearance of words in tweets over 110 times, with line width indicating the frequency of co-appearance.

The semantic network centers around the word "student(s)" and reveals a compelling pattern. On the left, a community forms with words like "code," "program(s)," "future," "stem," "tech," "teacher(s)," "teach," "learn," "csedweek," and "inspire/inspiration": "student(s)" are actively engaged in learning "code," "program(s)," and "tech" under the guidance of their "teacher(s)"; "tech" is closely linked to "stem" education, "program(s)," and the "future" considerations for "student(s)"; and the mention of "csedweek" signifies a unique opportunity to "inspire" "student(s)" to embrace CS in K-12 schools. On the right, a star-like structure emerges, with words like "work/workforce" and "partner/partnership" exclusively connecting to "student(s)."

Table 2 presents key network statistics for each word in Figure 6. Among the three centrality measures, degree centrality represents the number of co-appearances of a word with other words in the tweets, closeness centrality measures the proximity of a word to all other words in the network, and eigenvector centrality is determined by a word's connections to well-connected words. The k-core, indicating core or marginal network position, is a subgraph where each word is adjacent to a minimum of k other words (Seidman, [28]). The top five words, along with "program(s)," "csedweek," and "inspire/inspiration," have higher values on all centrality measures, indicating their central status in the semantic network. These words are located within 3-core subgraphs, highlighting their core positions.

Table 2. Key network statistics of the top 30 frequent words in tweets queried by the first set of keywords.

<table><thead><tr><td>Rank</td><td>Word</td><td>Degree centrality</td><td>Closeness centrality</td><td>Eigenvector centrality</td><td><italic>k</italic>-core</td></tr></thead><tbody><tr><td>1</td><td>Student/students</td><td>18</td><td>0.950</td><td>0.571</td><td>3</td></tr><tr><td>2</td><td>Tech</td><td>6</td><td>0.594</td><td>0.327</td><td>3</td></tr><tr><td>3</td><td>Teacher/teachers</td><td>4</td><td>0.559</td><td>0.241</td><td>3</td></tr><tr><td>4</td><td>Code</td><td>3</td><td>0.528</td><td>0.223</td><td>3</td></tr><tr><td>5</td><td>Learn</td><td>7</td><td>0.613</td><td>0.376</td><td>3</td></tr><tr><td>6</td><td>State/states</td><td>1</td><td>0.500</td><td>0.104</td><td>1</td></tr><tr><td>7</td><td>Stem</td><td>2</td><td>0.514</td><td>0.164</td><td>2</td></tr><tr><td>8</td><td>Program/programs</td><td>4</td><td>0.543</td><td>0.273</td><td>3</td></tr><tr><td>9</td><td>New</td><td>1</td><td>0.500</td><td>0.104</td><td>1</td></tr><tr><td>10</td><td>Teach</td><td>1</td><td>0.365</td><td>0.044</td><td>1</td></tr><tr><td>11</td><td>Future</td><td>2</td><td>0.514</td><td>0.164</td><td>2</td></tr><tr><td>12</td><td>Class/classes/classroom/classrooms</td><td>1</td><td>0.500</td><td>0.104</td><td>1</td></tr><tr><td>13</td><td>Work/workforce</td><td>1</td><td>0.500</td><td>0.104</td><td>1</td></tr><tr><td>15</td><td>Access</td><td>1</td><td>0.500</td><td>0.104</td><td>1</td></tr><tr><td>17</td><td>Csedweek</td><td>3</td><td>0.528</td><td>0.212</td><td>3</td></tr><tr><td>20</td><td>Help</td><td>1</td><td>0.500</td><td>0.104</td><td>1</td></tr><tr><td>21</td><td>Partner/partnership</td><td>1</td><td>0.500</td><td>0.104</td><td>1</td></tr><tr><td>22</td><td>Course/courses</td><td>1</td><td>0.500</td><td>0.104</td><td>1</td></tr><tr><td>25</td><td>Take</td><td>1</td><td>0.500</td><td>0.104</td><td>1</td></tr><tr><td>28</td><td>Inspire/inspiration</td><td>3</td><td>0.528</td><td>0.212</td><td>3</td></tr></tbody></table>

Findings in tweets from the second set of keywords

The second set of keywords focuses on computational thinking in K-12 schools. As shown in Figure 4, the number of tweets increases from 2017 to 2018 and then decreases continuously. The number of original tweets in 2021 is only about 29% of that in 2018.

Figure 7 displays the top 30 frequent words in the tweets (excluding the keywords). The overlapping words within the top 10 frequent words from the first and second sets of keywords include "code," "learn," "student(s)," "teacher(s)," "tech," "teach," and "stem." Standardizing these words by the number of tweets (2,769), we would expect to see "code" and "learn" 19 times each, "student(s)" 18 times, "teacher(s)" 17 times, "educator(s)" 15 times, "class(s)/classroom(s)" 12 times, "tech" 11 times, and "teach" and "stem" 10 times each per 100 tweets.

Graph: Figure 7. Top 30 frequent words in tweets queried by the second set of keywords.

Table 3 displays the top 30 words from the second query, showing larger rank variations compared to the first set of keywords. However, the top four words have consistently remained in the top 10 over the five-year study period, and all top 10 words have consistently been in the top 30 since 2018, indicating a relatively high level of reliability.

Table 3. The yearly ranks of the top 30 frequent words in tweets queried by the second set of keywords between 2017 and 2021.

Rank	Word	2017	2018	2019	2020	2021
1	code	1	1	6	2	7
2	learn	2	3	3	4	2
3	student/students	6	4	4	3	1
4	teacher/teachers	7	5	2	8	3
5	educator/educators	51	18	1	1	20
6	class/classes/classroom/classrooms	22	2	15	9	11
7	tech	8	7	8	17	12
8	teach	20	8	11	13	8
9	stem	23	10	7	22	6
10	skill/skills	37	9	14	12	5
11	program/programs	24	19	12	7	4
12	new	4	14	9	38	23
13	use	15	13	20	23	15
14	integrate/integration	18	20	27	11	14
15	way/ways	58	58	5	54	64
16	develop/development	25	24	29	14	10
17	curriculum/curriculums	30	15	19	45	16
18	university/universities	264	215	13	6	24
19	math	153	12	32	27	19
20	problem/problems	68	22	16	60	32
21	free	78	32	30	16	9
22	future	11	17	48	157	41
23	chat	36	25	42	21	29
24	weave	-	6	250	200	-
25	solve	120	30	17	69	43
26	work/workforce	38	35	23	36	26
27	resource/resources	54	11	94	43	55
28	digital	14	26	55	42	111
29	make	28	36	41	28	60
30	help	72	34	33	15	97

The co-appearance of the top 30 words in tweets from the second query results in 416 ties. Notable word pairs in more than 100 tweets include "student(s)" and "learn" in 128 tweets, "solve" and "problem(s)" in 124 tweets, "student(s)" and "skill(s)" in 121 tweets, "university/universities" and "educator(s)" in 119 tweets, "teacher(s)" and "stem" in 113 tweets, and "teacher(s)" and "learn" in 106 tweets.

In Figure 8, the optimal threshold of 60 is set to depict the basic structure of the semantic network between the top 30 words in tweets from the second query (the network graph "zGraph2.xlsx" is available from https://github.com/socnetfan/cstimp3). The semantic network centers on "student(s)," with two communities highlighting the importance of "learn," closely linked with words like "code," "teacher(s)," and "stem" education. On the right side, connections exist between "teach," "code," and "solve" "problem(s)," contributing to students' computational thinking. Additionally, "develop(ment)" and "program(s)" are linked exclusively to "student(s)." Notably, "university/universities" and "educator(s)" are closely connected.

Graph: Figure 8. Barebone structure of semantic network between the top 30 frequent words in tweets queried by the second set of keywords.

Table 4 provides key network statistics, highlighting the centrality and core positions of the top four words: "code," "learn," "student(s)," and "teacher(s)." These words, along with "stem" and "skill(s)," demonstrate their core positions within the semantic network.

Table 4. Key network statistics of the top 30 frequent words in tweets queried by the second set of keywords.

<table><thead><tr><td>Rank</td><td>Word</td><td>Degree centrality</td><td>Closeness centrality</td><td>Eigenvector centrality</td><td><italic>k</italic>-core</td></tr></thead><tbody><tr><td>1</td><td>code</td><td>4</td><td>0.633</td><td>0.273</td><td>3</td></tr><tr><td>2</td><td>learn</td><td>9</td><td>0.800</td><td>0.476</td><td>3</td></tr><tr><td>3</td><td>student/students</td><td>10</td><td>0.867</td><td>0.463</td><td>3</td></tr><tr><td>4</td><td>teacher/teachers</td><td>7</td><td>0.733</td><td>0.386</td><td>3</td></tr><tr><td>5</td><td>educator/educators</td><td>1</td><td>-</td><td>0.000</td><td>1</td></tr><tr><td>6</td><td>class/classes/classroom/classrooms</td><td>2</td><td>0.500</td><td>0.175</td><td>2</td></tr><tr><td>7</td><td>tech</td><td>2</td><td>0.500</td><td>0.175</td><td>2</td></tr><tr><td>8</td><td>teach</td><td>2</td><td>0.522</td><td>0.149</td><td>2</td></tr><tr><td>9</td><td>stem</td><td>3</td><td>0.600</td><td>0.269</td><td>3</td></tr><tr><td>10</td><td>skill/skills</td><td>3</td><td>0.633</td><td>0.258</td><td>3</td></tr><tr><td>11</td><td>program/programs</td><td>1</td><td>0.489</td><td>0.094</td><td>1</td></tr><tr><td>12</td><td>new</td><td>2</td><td>0.500</td><td>0.175</td><td>2</td></tr><tr><td>15</td><td>way/ways</td><td>2</td><td>0.500</td><td>0.175</td><td>2</td></tr><tr><td>16</td><td>develop/development</td><td>1</td><td>0.522</td><td>0.120</td><td>1</td></tr><tr><td>18</td><td>university/universities</td><td>1</td><td>-</td><td>0.000</td><td>1</td></tr><tr><td>20</td><td>problem/problems</td><td>2</td><td>0.522</td><td>0.118</td><td>2</td></tr><tr><td>25</td><td>solve</td><td>2</td><td>0.522</td><td>0.118</td><td>2</td></tr></tbody></table>

Findings in tweets from the third set of keywords

The third set of keywords is related to tweets mentioning CS organizations and conferences in K-12 schools. Figure 4 shows a decrease in tweets from 2017 to 2021, with 2021 having about 57% of the tweets in 2019 and 58% of the tweets in 2020.

In Figure 9, familiar words like "teacher(s)," "student(s)," "code," "learn," "tech," and "stem" appear in tweets from previous queries. Standardizing the most frequent words by the number of tweets (1,467), we would expect to see "education" 35 times, "teacher(s)" 27 times, "student(s)" 24 times, "code" 21 times, "learn" 18 times, "equity" 15 times, "csta" 14 times, "tech" 13 times, and "stem" 12 times for every 100 tweets.

Graph: Figure 9. Top 30 frequent words in tweets queried by the third set of keywords.

Table 5 shows that most top 10 words remained in the top 30 consistently over the study period, indicating a relatively high level of reliability. The word "equity" has gained popularity since 2017, ranking third in 2019 and remaining in the top 10 thereafter.

Table 5. The yearly ranks of the top 30 frequent words in tweets queried by the third set of keywords between 2017 and 2021.

Rank	Word	2017	2018	2019	2020	2021
1	education	3	1	2	2	1
2	teacher/teachers	2	4	10	1	2
3	student/students	6	2	1	4	5
4	code	11	5	4	3	4
5	learn	8	11	6	5	3
6	equity	609	52	3	6	9
7	csta	5	7	5	15	10
8	tech	25	8	13	7	7
9	stem	22	9	12	17	6
10	chat	4	42	11	10	27
11	teach	12	19	17	9	12
12	support	19	10	30	8	26
13	join	18	61	21	11	8
14	educator/educators	-	23	7	14	16
15	make	30	17	9	24	20
16	standard/standards	1	3	50	124	23
17	ai	563	44	8	26	13
18	state/states	14	6	19	177	11
19	share	38	25	20	19	18
20	program/programs	13	15	31	28	15
21	work/workforce	15	36	14	33	24
22	great	21	18	15	41	33
23	thank	23	14	26	29	32
24	use	27	73	40	12	39
25	class/classes/classroom/classrooms	65	12	28	40	31
26	summit	44	22	18	50	22
27	need	36	28	77	13	72
28	new	9	30	49	36	19
29	interest	20	81	27	32	28
30	think	29	26	32	35	30

The co-appearance of the top 30 words in tweets results in 435 ties. For word pairs that appear in more than 60 tweets, "student(s)" and "education" appear in 88 tweets, "student(s)" and "learn" in 76 tweets, "teacher(s)" and "education" in 71 tweets, "code" and "education" in 70 tweets, "learn" and "education" in 63 tweets, "tech" and "code" in 62 tweets, "stem" and "education" in 62 tweets, and "learn" and "teacher(s)" in 62 tweets.

In Figure 10, with an optimal threshold of 40, the semantic network's basic structure is revealed (the network graph "zGraph3.xlsx" is available from https://github.com/socnetfan/cstimp3). In the semantic network, a central community features words such as "education," "student(s)," "code," "learn," "tech," and "stem," aligning with the top 30 frequent words from the first keyword set. A separate community in the lower right section includes words like "learn," "education," "student(s)," "csta," and "teacher(s)." Furthermore, two-star subgraphs are evident. One involves words like "equity," "chat," "state(s)," and "join," associated with "education." The other subgraph features words like "teach," "use," and "support," linked with "teacher(s)."

Graph: Figure 10. Barebone structure of semantic network between the top 30 frequent words in tweets queried by the third set of keywords.

Table 6 presents key network statistics, with the top five words ("education," "teacher(s)," "student(s)," "code," and "learn") exhibiting high centrality values, indicating their central role in the semantic network. The words "education," "student(s)," "code," "learn," "tech," and "stem" form 4-core subgraphs within the dense community, while "teacher(s)" forms a 3-core subgraph, highlighting their core positions in the network.

Table 6. Key network statistics of the top 30 frequent words in tweets queried by the third set of keywords.

Rank	Word	Degree centrality	Closeness centrality	Eigenvector centrality	k-core
1	education	12	0.833	0.509	4
2	teacher/teachers	7	0.652	0.301	3
3	student/students	6	0.625	0.403	4
4	code	5	0.536	0.354	4
5	learn	4	0.577	0.304	4
6	equity	1	0.469	0.099	1
7	csta	2	0.536	0.157	2
8	tech	4	0.517	0.305	4
9	stem	4	0.517	0.305	4
10	chat	1	0.469	0.099	1
11	teach	1	0.405	0.058	1
12	support	1	0.405	0.058	1
13	join	1	0.469	0.099	1
15	make	1	0.469	0.099	1
18	state/states	1	0.469	0.099	1
24	use	1	0.405	0.058	1

Table 7 summarizes the rankings, degree centrality, and k-core values for common top frequent words across three distinct keyword sets. Notably, words such as "student/students," "tech," "teacher/teachers," "code," "learn," and "stem" consistently occupy top positions in all sets, emphasizing their significance in K-12 CS education discussions. These results illuminate both common and unique focal points within Twitter conversations related to K-12 CS education, offering valuable insights into the prevailing themes of these discussions.

Table 7. Comparing rank, degree centrality, and k-core among common top frequent words in tweets queried by each set of keywords.

Word	Rank in keyword set	Degree centrality in keyword set	k-core in keyword set
1	2	3	1	2	3	1	2	3
student/students	1	3	3	18	10	6	3	3	4
tech	2	7	8	6	2	4	3	2	4
teacher/teachers	3	4	2	4	7	7	3	3	3
code	4	1	4	3	4	5	3	3	4
learn	5	2	5	7	9	4	3	3	4
stem	7	9	9	2	3	4	2	3	4
program/programs	8	11	20	4	1	0	3	1	0
new	9	12	28	1	2	0	1	2	0
teach	10	8	11	1	2	1	1	2	1
class/classes/classroom/classrooms	12	6	25	1	2	0	1	2	0
work/workforce	13	26	21	1	0	0	1	0	0
make	18	29	15	0	0	1	0	0	1

Discussion

This study analyzed tweets using three keyword queries: K-12 CS education, K-12 computational thinking, and Twitter accounts of six CS education organizations and conferences discussing CS education in K-12 schools. Despite their independence, the descriptive content analysis found significant similarities among the tweets.

Firstly, the analysis of tweet frequency from 2017 to 2021 suggests a decline in discussions related to K-12 CS education since 2019. This decline may be attributed to various factors, including the effects of the COVID-19 pandemic. Remote learning policies during the pandemic posed challenges for students and teachers in K-12 schools regarding CS education (Zhu & Wang, [36], [37]; Crick et al., [8]). The shift of CS education conferences to online formats limited participation and interactions, consequently impacting the growth of CS education in K-12 schools (Falk & Hagsten, [13]; Mubin et al., [24]).

Secondly, the frequency of certain words in the tweets consistently indicated their reliability across the three sets of keywords. Overlapping words like "student(s)," "tech," "teacher(s)," "code," "learn," and "stem" underscored the importance of "tech," "code," and "stem" in K-12 CS education. This finding aligns with our previous study (Zhu & Wang, [36], [37]) and highlights the strong link between CS education and general STEM learning in K-12 schools. Efforts have been made to incorporate computational thinking in STEM education, engaging learners in coding to solve real-world technical problems (e.g., Hershkovitz et al., [18]; Mumcu et al., [25]). Effective CS education necessitates active participation from both students and teachers, moving beyond passive acceptance of assigned tasks, as revealed through interviews conducted with K-12 CS teachers and CS professors in the CSTIM project (Zhu & Wang, [36], [37]).

Thirdly, the constructed semantic networks from the top 30 words in the tweets showed similar patterns. "Student(s)" held a central position in the first two networks and was important in the third. "Learn," "tech," "code," and "stem" were core words in all three networks. This highlights the importance of adopting a student-centered learning strategy in K-12 CS education, where students become co-creators guided by their curiosity, creativity, and personal interests within the broader STEM education ecosystem (Anwar et al., [2]; American Psychological Association [APA], [1]; Boulden et al., [3]; Brown, [4], Shaby et al., [29]; Wright, [34]; Zhang et al., [35]). Additionally, "teacher(s)" played a central role in each network, emphasizing their role as facilitators in supporting student-centered learning (Brown, [4]).

Moreover, the frequency of the term "equity" in our Twitter data highlights the growing recognition of the need for equitable access and opportunities in K-12 CS education. This indicates a shared commitment among educators, researchers, and CS organizations to create a diverse and inclusive learning environment. These discussions on Twitter contribute to broader efforts in fostering equal opportunities for all students, promoting a more inclusive educational landscape.

In conclusion, this study offers valuable insights into K-12 CS education by analyzing tweets from different queries. The findings indicate a decline in public awareness, emphasize student-centered learning and teacher facilitation, highlight the integration of CS education with STEM, and underscore the importance of equity. Addressing these aspects can lead to effective strategies and policies that foster inclusive K-12 CS education, preparing students for the future in a rapidly evolving digital landscape.

Limitations and future research

This study has limitations that should be acknowledged. Firstly, data collection was limited to specific K-12 CS education-related Twitter accounts and hashtags, potentially excluding other conversations on the topic. Future research could expand the search scope to capture a broader view. Secondly, the data focused on the United States and English-language tweets, limiting generalizability to other countries and languages. Including data from various countries and languages would offer a more global perspective. Thirdly, the study relied on descriptive analysis of large-scale Twitter data, lacking in-depth understanding of individual contexts. Future studies could analyze the detailed context of tweets to explore factors influencing discussions. By recognizing these limitations, the study opens up avenues for further research that can overcome these constraints and provide a more comprehensive understanding of the topic at hand.

Conclusion

This study provides rich, critical insights into the evolving discourse of K-12 CS education by analyzing Twitter data from 2017 to 2021, with a focus on computational thinking, coding, and key education organizations. Our findings reveal notable trends, including a decline in tweet volumes post-2019 and the central role of coding within the broader STEM education framework. The prominence of student-centered learning also emerged as a key theme, emphasizing the need for more active student engagement in K-12 CS education. One of the most critical insights from this research is the persistent concern regarding equity in access to CS education. Addressing disparities in access is not only a pressing challenge but also essential for ensuring that all students, regardless of background, are equipped with the computational skills necessary for the future. Furthermore, this study demonstrates the immense potential of social media platforms, such as Twitter, to offer valuable insights into educational trends and shifts in public discourse. As digital platforms continue to evolve, educators, policymakers, and scholars must leverage these tools to remain responsive to emerging trends while ensuring that equitable and effective educational practices remain at the forefront of CS education.

Disclosure statement

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

By Cheng Wang and Meina Zhu

Reported by Author; Author