Playful programming,Social Resilience,and Persistent Actions as Drivers of Preservice Early Childhood Teachers’ Engagement in Computer Science

Conceptual Framework

The core grounding of our conceptual framework in the present study is playful programing. We envisioned play during programing that can invite participants to consider CS to be approachable and fun. Play in our framework is two fold. One is dramatic play (with robots) whose role is essential within early childhood education. The other is play with (robot) programing in which tinkering is natural in debugging and lesson design for preschoolers. Our framework is grounded in literatures not only on play (e.g., Ashiabi, 2007; Hurtado-Mazeyra et al., 2022; Vygotsky, 1978) and collaborative programing (e.g., Dawson et al., 2018; Echeverría et al., 2017; Nagappan et al., 2003), but also on resilience (e.g., Heljakka, 2023; Luthar & Cicchetti, 2000; Southwick et al., 2014), As such, it justifies the potential role of play in collaborative programing that can invite early childhood teachers to exhibit resilience in the face of challenges. We explain the framework in the following two sections labeled “playful programing” and “resilience.”

Playful Programming

Many studies indicate that non-CS majors often learn programing most effectively in the context of collaborative learning projects (Dawson et al., 2018; Echeverría et al., 2017; Nagappan et al., 2003). Extensive research also indicates that collaborative programing boosts academic achievement and interest (e.g., Wu et al., 2019). Furthermore, the present study theorized that collaborative programing could counter stereotypical conceptions of CS as solitary, non-interpersonal work that is only for certain groups of people. Play, a unique element of the study framework, was expected to generate collective playfulness. Play with robots was essential to program robots and also to design lessons for young children’s play with robots. At the same time, this study’s conceptual framework helped to frame children’s dramatic play with robots in an imaginary situation and roles to fill following associated rules during field experience in preschools (Bodrova & Leong, 2003; Hostettler Scharer, 2017; Vygotsky, 1978).

Play-based activities have the potential to be highly motivating among higher education learners (Boysen et al., 2022; Jørgensen et al., 2023). One argument is that exposing preservice, early childhood teachers to play and playfulness in the design of their teacher education class activities can give them an enhanced appreciation of the benefits of play and playfulness in teaching and an understanding of how such approaches may be experienced by learners (Boysen et al., 2022; Galbraith, 2022; Kemple et al., 2015). Notably, play is critical to cognitive and socio-emotional development in early childhood (Ashiabi, 2007; Hurtado-Mazeyra et al., 2022; Vygotsky, 1978). Play contributes to such development by encouraging children to engage in semiotic mediation and by inviting children to adopt/use rules inherent to the culture to which the children belong (Duncan & Tarulli, 2003; Vygotsky, 1978). Within sociodramatic play, semiotic mediation is inherent in the object substitution by which children use a stuffed animal to refer to a person, or a building block to represent a phone (Guo & Mackenzie, 2015; Ma, 2014). That is, children use cultural knowledge to imagine a possible world, the entities within that world, and what can be used to represent the different entities. In this way, they begin to engage in the type of symbolic representation that is inherent to logic. They also need to negotiate their imagined world and the semiotic mediation used to create that world with their peers. Play-based activities also tend to be more intrinsically motivating than other activities for early learners (Duncan & Tarulli, 2003). As such, play is central to early childhood education. Preservice early childhood teachers are prepared to base curricula and lesson plans around play; indeed, early childhood is often called the play years (Galbraith, 2022). Still, not all early childhood education is entirely play-based; pressure from parents, policy and standards often leads many early childhood educators to choose a more didactic approach (Bubikova-Moan et al., 2019; Stipek & Byler, 1997). Some research indicates that with an enhanced appreciation and understanding of play, preservice early childhood teachers are more likely to commit to integrating play within their teaching (Jung & Jin, 2015).

Embedding collaborative play in collaborative programing could invite new conceptions of robots and robot programing. Considering that “how people and the things that they create help to shape the ways in which they and others view the world” (Smagorinsky, 2007, p. 62), it was expected that teachers’ thinking about robots for children’s collaborative play would shape robots and robot programing. Besides, play is essential to young children’s learning, but also integrating robots as co-players in imagined situations that include roles and rules is aligned with dramatic play (Bodrova & Leong, 2003; Hostettler Scharer, 2017; Vygotsky, 1978).

Playful programing can also help women engage more fully within programing (Regal et al., 2024; Tellhed et al., 2023). This is critical as the vast majority of early childhood teachers are women (Saluja et al., 2002; Van Laere et al., 2014). When engaged in collaborative programing tasks, women often assume roles that are stereotypically associated with females, such as documentation and project management (Fowler & Su, 2018). Engaging in such aspects does not positively predict persistence in CS and information technology majors (Weston et al., 2019). But when workload is evenly distributed, collaborative programing can be helpful for women programmers (Werner et al., 2004). One way this can help is by inviting women to see programing as a collaborative activity, thereby countering the stereotype of CS as a solitary profession (Werner et al., 2004). Collaborative programing can also lead to less programing errors (Bravo et al., 2013) and higher quality projects (McDowell et al., 2002; Nagappan et al., 2003; Preston, 2005) than solo programing. This is especially important among women students who often attribute struggles with CS tasks to lack of ability (Koch et al., 2008). Collaborative programing is often seen as more enjoyable than solo programing, especially among non-CS majors (Boyer et al., 2008; Nagappan et al., 2003).

Resilience

It is also important to consider the role of playfulness on cognitive and motivational outcomes within teacher education courses themselves. Resilience in the face of adversity is a critical attribute among university students (Brewer et al., 2019; Chua et al., 2023; Turner et al., 2017), and global incidence of low resilience among university students was estimated to be 36% in a systematic review (Chua et al., 2023). Resilience can be defined as positive adaptation in the face of adversity (Luthar & Cicchetti, 2000; Southwick et al., 2014), and can explain mental health and academic outcomes (Luthar & Cicchetti, 2000; Turner et al., 2017). Playfulness can contribute to both resilience and creativity (Heljakka, 2023). Self-efficacy also plays a key role in explaining/predicting resilience (Beltman et al., 2011; Yada et al., 2021). When non-CS majors learn to program, they often face fears regarding their abilities to understand and master the content and to learn to program competently (Hogan et al., 2023; Shell & Soh, 2013). This can result from struggles producing adequate code to solve a problem, and struggles in identifying and solving bugs (Gorson & O’Rourke, 2019). These struggles can lead such learners to have low programing self-efficacy, especially considering that mastery experience is the strongest contributor to self-efficacy (Bandura, 1997; Usher & Pajares, 2008).

The vast majority of preservice early childhood teachers are women (Saluja et al., 2002; Van Laere et al., 2014); stereotypes about computing (e.g., that computer scientists are almost always men who work long hours alone) can turn women away from computing (Cheryan & Markus, 2020; Koch et al., 2008; Main & Schimpf, 2017). Women are likely to have faced lacks in curriculum and encouragement to pursue CS throughout their education (Ryoo, 2019). Resilience in the face of fear and lack of confidence is especially critical among women of color who are learning to program in “White-male dominated spaces” (Williams et al., 2024, p. 21:3).

Incorporating play within university courses may be a promising method to foster playfulness and resilience among university students (Magnuson & Barnett, 2013). This is important because while didactic resilience training focused on psychological constructs (e.g., mindfulness, positive psychology) exists (Brewer et al., 2019; Kunzler et al., 2020), evidence for effectiveness of such interventions on resilience or mental health improvement as assessed by meta-analytic methods is mixed (Kunzler et al., 2020). Play may allow teacher education students to engage in imaginal coping; for example, to sing instead of feel distress, frustration or despair (Clark, 2016). Especially as they are learning computer science, many challenges can emerge, as content being learned is novel and bugs prevent successful execution of code (McCauley et al., 2008). Singing can lead to physiological changes that contribute to enhance resilience (Kang et al., 2018).

Research Design

This case study (Stake, 1978) was designed to elicit an in-depth understanding of participants’ programing and debugging experiences. The unit of analysis was a class in which early childhood education (ECE) preservice teachers learned to program robots over the course of a 3-week unit. We used a phenomenological lens to ask “[w]hat is the nature or essence of the experience of learning” of this particular group of programing learners (Van Manen, 1990, p. 10). We were specifically interested in the meanings they ascribed to their lived processes of collaborative programing and reacting to challenges. Data included a pre-survey about motivation and experience related to STEM, robotics and programing, classroom videos, reflections, and interviews. Analysis strategies included open and axial coding and conversation analysis.

Study Context and Participants

Participants were 14 preservice, ECE majors enrolled in an arts-based early childhood education course in a large university in the southeastern United States. They worked in pairs during a 3-week programing unit. All were female. The average age was 19.79 years. Two were African Americans, one was Asian American, one was multiracial, and the rest were Caucasians. Nine indicated no to low prior programing knowledge and five indicated intermediate. Six were video-recorded and five participated in individual interviews. Table 1 lists demographic information and data completion. All participant names have been changed to pseudonyms.

OPEN IN VIEWER

Table 1.

Participant Data Summary.

Team	Pseudonym	Video recording	Interview	Age	Race/ethnicity	Computer programing knowledge	Prior robot programing experience (context)
1	Zoey	Yes	Yes	19	Caucasian	No knowledge	Yes (Undergrad ed tech class: Ozobot)
1	Joy	Yes	Yes	20	Caucasian	Intermediate	Yes (Undergrad ed tech classes: Scratch, Makey Makey, Ozobot)
2	Mia	Yes	Yes	19	African American	Intermediate	Yes (Middle school: Lego Mindstorms, robotics class, engineering program)
2	Ellen	Yes	No	20	Hispanic and Caucasian	No knowledge	No
3	Luna	Yes	No	20	Caucasian	Low	No
3	Liz	Yes	No	20	African American	Low	No
4	Pam	No	Yes	19	Asian	No knowledge	No
4	Ava	No	No	19	Caucasian	No knowledge	No
5	Haley	No	No	20	Caucasian	Intermediate	Yes (Undergrad ed tech classes)
5	Belle	No	No	20	Caucasian	Low	Yes (Undergrad ed tech class)
6	Lucia	No	Yes	21	Caucasian	Intermediate	Yes (Undergrad ed tech classes: Scratch, Makey Makey, Ozobot)
6	Stella	No	No	21	Caucasian	Low	No
7	Cara	No	No	20	Caucasian	Low	No
8	Moira	No	No	19	Caucasian	Intermediate	No

Note. Teams with only one row mean that the participant’s partner in that team was a non-participant. Programing knowledge is self-assessed prior to the robot programing unit began in the present study. The undergraduate ed tech course title is blinded. Based on participants’ other data sources, the listed Ozobot experiences were from using color marker coding, not from block-based coding. There is no other data about Scratch and Makey Makey experiences (e.g., if each indeed included a robot element).

Robot Programming Learning Unit

The overall design of the unit was grounded in the conceptual framework of the study. Collaborative play and collaborative programing were embedded throughout the unit. Collaborative teams consisted of teaching partners who went to preschools together for field experience. Robots were portrayed as co-players in dramatic play not only in programing activities but also in lesson design activities for their field experience preschool classrooms.

The robot programing unit was comprised of three 3-hour classes and two sessions of field experience within a preschool. In Class 1, participants were presented with an overview of the history of educational robotics and learned basics of coding using Hour of Code and Ozoblockly. They also learned in pairs to teach with a sample lesson that involved dramatic play; in it, robots represented vegetables and needed to navigate through a supermarket to get to the section that is consistent with the type of vegetables (i.e., leafy green for the spinach bot or root vegetable for the carrot bot). The sample lesson included a song that they could sing based off of the muffin man song: “Do you know the carrot bot, the carrot bot, the carrot bot, do you know the carrot bot who lives in the root vegetable section? Yes, I know the carrot bot, the carrot bot, the carrot bot, yes I know the carrot bot who lives in the root vegetable section.” They were provided with the supermarket map and Ozoblockly code that programed the robot to get to its correct section (by skating past the frozen food section because it was freezing cold, speeding up past the snack section to avoid unhealthy food, and spinning and displaying fireworks in the root vegetable or leafy green vegetable sections; see Figure 1). They learned to code their robots and run them through the supermarket map. They practiced the lesson and then taught it in their field experience preschool classroom with their carrot and spinach bots during Week 1.

OPEN IN VIEWER

Figure 1.

Example code (left) that programed the robot to navigate through the supermarket map (right).

The lesson used in Week 2 was designed by the participants in pairs. Each pair created their own lesson, programed their robot’s movements needed to perform dramatic play with preschoolers within the lesson together, and taught the lesson with the programed robots in field experience. In Class 2, before designing their own lesson, they worked on a series of programing practice activities. For example, one activity was to program the robot to move at different speeds per its top light color. Programing practice activities were done in association with potential classroom activities for children to play with robots. This way, robot programing involved dramatic play not only between participants and their robot but also within their preschool classroom. Participants were invited to look up example code (e.g., Figure 2) when needed, and create alternative code if they looked up the code. The design of this series of programing practice activities was also to create natural opportunities for tinkering and playful exploration but also for their design for play among preschoolers in field experience. In Class 3, participants were introduced to other educational robots such as Lego Mindstorms, Lego WeDo, Ozobot Evo, Dash & Dot, and Roborobo for them to play with.

OPEN IN VIEWER

Figure 2.

Programing practice activity example code.

Data Collection

At the beginning of the unit, all participants took a presurvey covering motivation and experience related to STEM, robots, and programing. The survey was used in prior research (Belland et al., 2021), subscales of the survey received Cronbach’s alphas ranging from 0.7 to 0.95. Three paired groups were video-recorded in three classes, and five participants were interviewed after the unit ended (see Table 1). All participants responded to reflection prompts. Reflection prompts were designed to guide participants’ reflection-in-action and reflection-on-action in each class (Umutlu & Kim, 2020; Schön, 1987). Prompts for reflection-in-action were given in the middle of programing and those for reflection-on-action about programing were given at the end of each class. Three sets of prompts were given in Class 1 and Class 2. Two sets of prompts were given in Class 3. Three to five prompts were given per set. Responses were collected individually so even when the pair was attempting to debug the same robot together during collaborative programing, their responses were not the same. Reflection prompts included “What do you like about working with ozobots? Why?”“What do you find frustrating about working with ozobots? Why?”“What did you find the most challenging about working with ozobots today? Why?” and “Explain what you did to address the challenges.” Prompts for reflection-on-action about their field experience teaching were given in the beginning of Class 2 and Class 3.

The length of interviews ranged from 16 to 24 minutes. The semi-structured interview questions asked participants to talk about the processes of collaborative programing, play, and teaching, and the challenges that they encountered and resolved. Depending on specificity of responses, further questions were asked. Example questions are “What problems did you encounter when programing your robot?”“How did you handle the problems that you ran into?”“How did you know that was the problem?” and “How did you find the cause of the problem?”

Data Analysis

Interviews and classroom videos were transcribed, and imported into NVivo 12 along with reflections for analysis. An initial coding scheme was constructed by the first author based on her open coding of transcripts and literature related to collaborative programing, debugging, and tinkering (e.g., Kim et al., 2018; McCauley et al., 2008), attributional theories including foci on resilience and emotions such as playfulness and joy (e.g., Kim & Pekrun, 2014; Weiner, 1985), and stereotypical conceptions about CS and other disciplines (e.g., Cheryan, Meltzoff, & Kim, 2011; Cheryan, Siy et al., 2011). The coding scheme was revised based on discussions between the first, second, and third authors. Then, the third author used the revised coding scheme to code two interview transcripts, three reflections, and one classroom video. The first and second authors reviewed the third author’s coding independently, and then discussed the coding and revised the coding scheme with the third author. The third author applied the re-revised coding scheme to another set of an interview transcript, reflections, and a classroom video, and the three authors discussed the process once more and finalized the coding scheme. Sample coding scheme nodes are listed in Table 2. Then, the third author coded all classroom videos, interviews, and reflections, and the first author reviewed all coded data to reconcile disagreements with the third author. The second author randomly selected 5% of the coded data and reviewed it independently to check for agreement. The three authors met again to reach consensus, and the third author updated his coding using nodes based on the last consensus discussion. After completion of coding, the first, second, and third authors went through multiple rounds of meta-coding of coded data. That is, a list of salient observations was created upon the reviews of NVivo documents produced per coding scheme node and the list was discussed. The list was revised until a set of themes emerged. Themes were continuously refined while writing the present paper and revising the coded data. This process involved cognizant, constant efforts in describing the participants’ lived experience with collaborative programing through a phenomenological lens (Van Manen, 1990).

OPEN IN VIEWER

Table 2.

Sample Coding Scheme Nodes and Sample Excerpts.

Framework components	Sample nodes	Sample excerpts
Play and playfulness	Collaborative playful exploration	Mia: We’re getting closer. This is good. This is happening. This is real. This is me. Ellen: Exactly. … Ellen: Boom. Mia: We’re killing it. Ellen: We’re geniuses. Mia: Obviously. Oh, that was good!
	Playful fun	Zoey: It [coding] was fun for me … it’s fun to be able to see the robot do what you program it to do. And I also like it because it’s easy for kids and fun for kids.
Collaborative programing, debugging, and tinkering	Looked at the robot immediately after noticing the problem	Lucia: “When it didn’t do exactly, I think we first looked at the robot just because we were, oh, something’s wrong with this. But then we would go on to realize, okay, well maybe it’s in the coding, maybe there’s something in the coding that we can fix.”
	Reviewed only the suspected area(s) in the code	Zoey: “We would look at its [the robot’s] movements first and then just based on where it was going or how far it was turning. That’s what would make us decide to change it to a different movement or different turn [in the code].”
Resilience	Persistent behaviors or utterance	Mia: “Okay. Yes. Then you turn. Yes. Yes, yes. Skate around a little bit. Turn left. No! Ah, I see what the issue is… See this whole skating thing is so inconsistent. I don’t know what to do about this stuff. I’m gonna try this again.”
	Non-persistent behaviors or utterance	Ellen: “Maybe she’s not picking up that it’s red. And I don’t know how. I’m giving up. We did the code right and it’s just not working.”
Stereotypical conceptions	Stereotypical comments	Belle: “Robot makes me too anxious, I’m not good with tech”Lucia: “… I’d never really thought of myself as somebody that could do programing.”
Attribution to the success in programing	Collaborative programing	Pam: “The program itself was very simple and I had my friends around me if I was stuck.”
	Adaptive attribution	Mia: “I think if you’re more interested, you’re probably going to be more talented. … I don’t think that there’s anything that makes people innately better at doing programing. I just think that if you’re patient, and willing to take the time to try and figure it out, then you’ll be fine.”
	Self-efficacy	Lucia: “It [my programing] was definitely better than I expected.”

Theme 1: Through Playful Coding, Preservice Early Childhood Teachers Learned That Computer Science is Approachable and Fun

Zoey began the programing unit believing that computer science would be too difficult for her. Such a conception seems to have resulted from no early exposure to computer science. As she explained during the interview, “I grew up not doing any of that [coding] and I just thought coding was really difficult and that I would never be able to do it.” Later her stereotypical conception about computer science seemed to have disappeared, as hinted in her other interview comments below:

It [coding] was fun for me because, well, first it’s easy. When you hear the word coding, I had never had much experience with it before. And you see the long game, like codes that don’t even look like English. But with Ozoblockly it’s just block coding, you’re just dragging blocks and it’s a lot easier than what people think. That’s why I liked it and it’s fun to be able to see the robot do what you program it to do. And I also like it because it’s easy for kids and fun for kids.

This elimination of her stereotypical conceptions about computer science was observed also with Zoey’s improved self-efficacy. Her reflections depict progressively more positive sentiment as she succeeded in collaborative debugging in class. While challenges were reported in all her reflections (note: reflection prompts asked about the challenge that they were experiencing at the moment and those they experienced in each class), her excitement about observed success in her team robot was recorded at the end of each class. For example, as listed in Table 3, Zoey noted the challenge of inconsistency in robot behaviors in her reflection-in-action (i.e., reflection prompts were given in the middle of debugging) in both Class 1 and Class 2 but her notes contained less frustration in Class 2 than Class 1. She also engaged in reflections on mastery experience. For example, she noted, “The most exciting moment was finally getting the ozobot to follow the right path,” which suggests growth in her self-efficacy considering that mastery experience is a critical source in building self-efficacy (Bandura, 1997; Beatson et al., 2018).

OPEN IN VIEWER

Table 3.

Excerpts from Zoey’s Reflections.

Class 1Reflection-in-action	Class 1Reflection-on-action	Class 2Reflection-in-action	Class 2Reflection-on-action
“Another challenge was keeping it from veering off the page and staying on the path.”	“It was most challenging to figure out which movements to change and keep.”	“We first struggled making the ozobot respond to colored lines.”	“The biggest challenge was keeping the robot from moving and pushing the puzzle pieces.”
“It was frustrating having to keep re-load[ing] the ozobot, and it wasn’t always consistent.”	“The most exciting moment was finally getting the ozobot to follow the right path.”	“The ozobots can sometimes be inconsistent.”	“It was exciting to see the robot complete the shapes that we made.”

Similarly, Lucia’s stereotypical conceptions about computer science appeared to dissipate as her self-efficacy improved. In her interview, she said:

I thought I did better than I expected…. Just because I’d never really thought of myself as somebody that could do programming. I had done a little bit in the past, so I felt confident and okay, I kind of knew what ozoblockly was, but I hadn’t had much experience. (…) But it was definitely better than I expected just because I had low expectations. (…) I think everybody can do a little bit of it. Some people caught on how to exactly program it faster than others, but I think that it’s not a skill that some definitely have and some don’t. Everybody can somehow use ozoblockly.

Joy explained changes in her conceptions about computer science during the interview:

I figured coding and programming sound really hard and really challenging and I had never done anything like it before other than the [name blinded, another course in which ozobots were introduced] class. So before that, I didn’t think that that was something that I could do. So then I liked to use these [coding] programs that were so easy and user friendly. It was really nice.

In the pre-survey, Joy listed Scratch, Makey Makey, and Ozobot as what she learned from the general educational technology course that she had taken prior to the present study. Based on other data sources from the participants who took the prior course (see Table 1), it is inferred that the prior course taught students to use color marker coding without computer use to control Ozobots. For example, Moira noted in her Class 1 reflection, “It’s something I’ve never done before.” Lucia said, “I had done a little bit of programing in the past (…) I kind of know what ozoblockly is, but I hadn’t had much experience.” And Zoey said, “I had never had much experience with it.” However, it should be noted that data about these participants’ experiences from the prior course is limited because only three of these five participants did the interview and the semi-structured interview questions were not designed to ask about the prior course specifically. Consequently, relevant data were insufficient to make full sense of the participants’ prior programing self-assessment data which ranged no knowledge to low and intermediate knowledge. Joy self-assessed her prior programing knowledge as intermediate prior to her engagement with the programing unit in the present study. All her reflections from Class 1 to Class 3 included a notion about ease of robot programing. For example, in Class 1, she noted “They are fairly easy to program and fun to use.”

Changes in conceptions about computer science did not seem to result simply from exposure to programing (also considering that six participants were exposed to block-based coding prior to the present study); rather, the method of integrating programing into the course seems to have played a critical role. That is, the robot was introduced as a co-player in dramatic play in which young children can engage. For example, the sample lesson that the participants learned to teach in their preschool field experience with their robots aimed to create a dramatic play context in which children helped the robots navigate a supermarket as the robots looked for their home (i.e., the carrot bot was trying to find the root vegetable section and the spinach bot was trying to find the leafy green vegetable section). Robot programing in the present study embodied objects and activities such as dramatic play that were not stereotypically associated with computer scientists. These design elements in the unit may have impacted our participants’ stereotypical conceptions about programing, considering that learning environments without objects that convey CS stereotypes can positively influence CS interests of female undergraduates (Cheryan, Meltzoff, & Kim, 2011). When Zoey reported that she did better with programing than she had expected prior to the unit, she pointed out that her unexpectedly positive experience with robot programing resulted from the connection of programing to preschoolers’ play and learning:

I think that was the coolest part for me and that kind of surpassed my expectations to see how you can incorporate technology like that [robot programming] into a lesson even for four-year-olds and then bring it into the classroom.

In contrast, Belle appeared to retain her stereotypical conception. In her Class 2 reflection, Belle said, “Technology stresses me out” and in Class 3, she noted “Robot makes me too anxious, I’m not good with tech… Also, I don’t find it interesting so I don’t think I would teach it well.” Despite such comments, Belle highlighted the robots being user-friendly and cute in her Classes 1 and 2 reflections and “how excited the kids get when they see them [robots]” in her Class 3 reflection. Belle’s partner, Haley, seems to have become self-efficacious. While she self-assessed her programing knowledge prior to the unit as intermediate, she reported her struggles with robot programing in her Class 1 reflection and expressed her desire to become proficient. In her Class 2 reflection, she reported no challenge by saying “not really any problems, just a little bit of trial and error to get it perfect.” In her Class 3 reflection, she indicated that there was nothing else she needed to learn and said, “I feel like the block programing is pretty easy to use.” Neither Belle nor Haley was interviewed or video-recorded so there is no other data to further discuss their stereotypical conception. Given Belle’s reflection about Haley’s help during debugging (“I had my partner double check my code after I fixed it” in Class 1), it is probable that their collaborative programing and play facilitated Belle’s positive views on robot use in ECE classrooms despite her dislike for technology.

This does not mean that programing went well all the time. Joy encountered multiple bugs and had struggles with them (see the episode in Theme 2, e.g.). In her Class 1 reflection, Joy noted, “The ozobot wouldn’t follow the same path even with the same programing.” She also noted in her Class 2 reflection, “I tried re-coding but I still had the same problem with it not following the same path every time.” During the interview, she noted tinkering during debugging that had to be done as part of programing:

Sometimes it [the robot] doesn’t do exactly what you code it to do. So when it turns and you have it do a slight turn, it’ll do a slight turn, but then when you have it do a left turn, it’ll still not do a full left turn. So I had to figure out how to code it, maybe do a turn and then a slight turn to try to get it to do what I wanted it to do. But other than that, I mean it worked pretty well.

As observed often in her collaborative debugging with Zoey, much of the inconsistency in robot performance they experienced resulted from inconsistency in where (e.g., where exactly within the supermarket store map in Class 1) the ozobot began its movement. This seems natural considering the inherent complexity in the robot programing unit that required the participants not only to learn to program robots based on the given or designed choreographies of robots but also comprehend the choreography within their field experience context teaching preschoolers. In fact, Joy expressed her desire to learn to fully understand Ozoblockly in her reflection-on-action during the last day of the unit.

Despite the difficulties she encountered, Joy reported in the interview that she programed better than she had thought and attributed her success to her ability with technology:

I’m pretty good with technology. That was always something that my mom said, yeah, you need to go into technology because you’re so good at it. Just figuring out problems and stuff. So I think that could have helped.

Joy’s attributional notion is interesting in that she attributed her successful programing to her ability with technology only after the unit. She also cited verbal persuasion received from her mom as a factor in her assessment of technology ability (Bandura, 1997). She said in her interview anybody can do it because it is easy. Nonetheless, she appeared to believe that there are people who learn more quickly than others:

It’s so straightforward. I think maybe some people would grasp it quickly, more quickly. But I don’t think… I mean it’s so easy to get and so straightforward that I think pretty much anyone could do it.

Moira wrote in her Class 1 reflection that “The thought of coding seems so difficult, but it really wasn’t.” Later in the unit, the class was offered the opportunity to write their own code. Regarding this part of the unit, Moira wrote in her Class 2 reflection that the most exciting thing was “creating my own [code], even though it was hard.” Moira was aware of the increasing difficulty in coding tasks but despite the challenge, writing code from scratch was still exciting for her. She was not concerned about difficulty as something that would stop her from doing the activity any longer. Her changed stereotypical conception about computer science was related to her new self-efficacy. Changes in stereotypical conceptions about computer science through exposure to block-based programing have been previously observed elsewhere (Kim et al., 2015). Likewise, most participants in the present study became comfortable and self-efficacious as they learned block-based programing; thus, their stereotypical conceptions about computer science became too obsolete to associate computer science with special groups of people. As Pam noted in the interview, even those who cannot read yet can program:

This program specifically, I think anyone can do it, especially younger kids. It’s not very hard. Even if they can’t even read what the words are saying, they can use the mouse or the tablet and move the pieces together.

The pre-reader level of the platform was not used in the present study, but “the thought of coding” that Moira had as well prior to the unit (see her Class 1 reflection above), seemed to have changed as she became self-efficacious. Pam’s increased self-efficacy may have led to her desire to incorporate more robot moves into her teaching as expressed in her Class 3 reflection from the last day of the robotics unit: “I want to learn to program more [robot] dances or if there is a jumping feature I could use.”

Theme 2: Preservice, Early Childhood Teachers Engaged in Adaptive Attribution

Attributional factors mentioned during interviews as success enablers were observable in the participants’ actions during debugging. In other words, what they did was aligned with what they thought would help them do well with programing. During the interview, when asked what would take to be good at programing, Zoey responded:

I think it’s more of an interest thing…. I think if people don’t have that interest then they might not build up those [programming] skills. … And people who are really interested in it, I mean they’ll be spending more time doing it and I think in the end they might have more knowledge and more skills to do it.

Zoey thought that time investment based off interest would lead to success in programing. Such attribution is adaptive in that the locus of control for both time and interest are internal in contrast to the external locus of control for luck, for example (Weiner, 1985). Considering that the interview was conducted after the robot programing unit, her adaptive attribution may have resulted from her removed stereotypical conception about computer science being only for special groups of people because of its difficulty. After all, she believed that to be good at programing, one does not have to have special talent; rather, time investment matters. Joy, Lucia, Mia, and Pam also highlighted the ease of block-based programing in their interview (see Theme 1).

This finding related to adaptive attribution is meaningful not only for teacher learning but also for computer science education because such a growth mindset could be cultivated in teachers possibly through block-based programing experience. And growth mindsets could become a part of the school cultures to which these teachers belong. Growth mindset as a culture is critical (Dweck, 2014). Fundamental beliefs of teachers are impactful in their teaching practice (Kim et al., 2013). Thus, if future teachers have a growth mindset about computer science, they are likely to invite all students to computer science, including those from underrepresented populations in computer science. Note that women are underrepresented in computer science (Beyer, 2014). A recent study showed that teaching with a growth mindset led to students’ development of interest in computer science (Burnette et al., 2020).

Aligned with her adaptive attribution that successful programing comes with invested time, Zoey did not give up in the middle of programing, as illustrated in the following dialog with her partner Joy during debugging. Zoey’s utterances were often negative as shown in the debugging episode below (emphasis added to her negative utterances). When attribution to success in programing was adaptive, even when participants wanted to give up and move on, they still continued their work.

Zoey had good reason to be frustrated also as explained in her interview: “My least favorite part was probably just how it could be a little frustrating, like with the inconsistencies.” The inconsistencies were pointed out by other participants. For example, Cara’s reflections listed the robot “being so sensitive” as a key challenge of debugging. Despite challenges from inconsistencies, Stella, Cara’s partner, reported persistent actions like Zoey’s in her reflections:

The most difficult part is calibrating the bot because sometimes it said it worked when it actually didn’t. It was also hard to make it go exactly where you wanted it to go. I would just keep trying to calibrate it until it worked and I would position it differently to try to let it end up in different spots.

Still, less negative emotions are likely to be experienced when one has adaptive attribution (Kim and Pekrun, 2014) but Zoey expressed frustration throughout this episode of debugging and still pushed herself to keep going despite constant frustration. A study (Garcia & Rimé, 2019), albeit outside of educational research, found that venting negative emotions in a community was associated with social resilience within the community. It should be also noted that collaborative programing partners also taught using their robot lessons together during field experience in the same preschool classroom for the entire semester as described in the method section. Also in Mia’s case, she had adaptive attribution about successful programing (i.e., time investment) and her persistent actions were aligned with it but not with her utterances, like Zoey and Joy as shown in the following debugging episode (emphasis added).

(When Mia noticed the robot was not following the intended pathway, she steered it with her hand to correct its movement)

The multiple debugging episodes shown above suggest that, for Mia, singing was a way of fostering resilience (Clark, 2016; Gunnestad, 2006; Kang et al., 2018). It may be possible to integrate singing into scaffolding for problem-solving with uncertainty. Especially in early childhood education contexts, such a scaffolding strategy can be integrated seamlessly considering private singing is often observed among children and used by teachers (Thibodeaux et al., 2019). Managing uncertainty is important in collaborative problem solving (Jordan, 2015; Jordan & McDaniel, 2014). In her interview, Mia reported that robot programing was not as easy as it may seem when compared to other robots like Lego Mindstorms. When asked how well she had done during the robotics unit, Mia said she did only average even though she had done some programing prior to the present study. When a task is easy, it is not too difficult to keep working on it; when difficult, especially more difficult than you thought, it is easy to quit (Weiner, 2010; Wigfield & Eccles, 2000). Mia’s perseverance in action in the midst of difficulties is aligned with her adaptive attribution that highlighted time for effort.

Another participant, Ava, mentioned in her Class 2 reflection that persistence helped her overcome challenges during debugging, but she reported that challenges were not too difficult. No other data is available to corroborate Ava’s notion. Further research is needed to understand learners like Mia who still persist despite difficulties beyond their comprehension. Learning to work through difficulties is essential not only to CS education but also to any other educational discipline. It seems logical to think that teachers with such a quality would facilitate perseverance in children. Research on action control and volition, that explains effort made where things seem unattainable (Kim & Bennekin, 2016; Corno, 1993; Heckhausen, 2007; Keller, 2017; Kuhl, 1987), could be applied to CS education.

Theme 3: Preservice, Early Childhood Teachers Engaged in Joint Celebration When They Observed Each Other’s Successes During Collaborative Tinkering

All participants reported successful trial and error. For example, Zoey’s Class 1 reflection corroborates her use of trial and error. When prompted to explain how she addressed challenges in making the robot do what she wanted it to do, she wrote “I made small adjustments to see which movements worked best at keeping it on the path. I mostly just used trial and error to see what worked best.” It is not unusual to see trial-and-error (e.g., Fitzgerald et al., 2010; Grigoreanu et al., 2006; Murphy et al., 2008) especially among novice programing learners. This is often the case in the block-based programing contexts (Kim et al., 2018, 2022). Ozobot Bits were less precise than ideal, but inconsistencies in robot performance also resulted partly from where robots were positioned on the supermarket map as briefly discussed earlier (see Theme 1). In her Class 1 reflection, Mia reported, “I had to try and change the algorithm so I could get my intended target.” At times, physical manipulation (e.g., maintaining the first position of the robot in all trials) was sufficient to debug, but missed. Interestingly, Belle’s Class 1 reflection pointed out the imperfect process of “adjusting the starting point.”

Enjoyment from collaborative explorations and discoveries and expectation for such enjoyment seemed to drive repetitive cycles of tinkering as depicted in discourse between Joy and Zoey during debugging (see Theme 2), and in Mia and Ellen’s discourse below. Here, Mia and Ellen debugged the robot to visit different animals and light up based on the animal color such as brown for kangaroos, green for alligators, and pink for flamingos. Emphasis is added to text relevant to the finding discussed here.

Vicarious enjoyment from seeing others’ discoveries helped participants get going. Joint celebrations about successful debugging played a role of verbal persuasion that positively impacted self-efficacy (Bandura, 1997; Usher & Pajares, 2008). A scene from Luna and Liz’s collaborative debugging in Class 1 exemplifies this (emphasis added):

Other debugging episodes, such as Zoey’s dialog with her partner Joy in Theme 2, exhibited bidirectional verbal persuasion that helped each other be resilient in the face of their own struggles. Pam’s comment, “I had my friends around me if I was stuck,” also hints at resilience enabled through collaborative and social programing.

While it is often argued that trial-and-error should be eliminated (Murphy et al., 2008), the same argument is not applicable to the present study. Even when participants were frustrated with a prolonged period of debugging, their trial-and-error appeared to enable social resilience or vice versa. Also with this experience, teachers will be able to notice such practice among children in their future classrooms. With first-hand knowledge of what it is, why it has to happen, and how it is done, these future teachers’ capacity is expanded. Experiencing enjoyment from collaborative explorations through perseverance is not trivial as well. As reported earlier, when facing success after struggles through trial and error, Mia’s utterance depicts the joy of achievement “I did it. Oh my God. Oh my goodness. Wow. That was the best moment of my life. Wow. The best moment of my life.” The present study adds to the evidence that trial-and-error is not something that should be regarded as unproductive (e.g., Kim et al., 2021; Berland et al., 2013), but it should be studied as part of a unique process of social resilience during collaborative programing. Educators still would want to minimize aimless or endless trial and error; however, scaffolding is needed to allow sufficient time to explore and make sense of collaborative experiences (Kim et al., 2021) during perseverance, discovery, and enjoyment.

Overall Discussion

The study findings provide a sense of how preservice early childhood teachers exhibit resilience during collaborative programing and collaborative play. Their resilience was informed by playfulness (Boysen et al., 2022; Jørgensen et al., 2023), and resilience (Luthar & Cicchetti, 2000; Southwick et al., 2014). Taking a playful approach allowed the preservice teachers to acknowledge their struggles but also engage in playful coping (e.g., singing) to regulate negative emotions (Clark, 2016; Kang et al., 2018). This resilience appeared to be associated with gains in self-efficacy and productive attribution of success, and budding interest in and views countering stereotypical conceptions of CS as they saw how it can engage and facilitate the learning of young children. The productive attribution and lack of unproductive attribution appeared to result from participants’ experience of the course as a safe play space in which it was okay and expected to make mistakes (Lyon & Clayton, 2021; Pinkard et al., 2017). The budding interest appeared to result from participants’ successful use of robots and coding within field experience in preschools. This is important because women are drastically underrepresented in CS (Beyer, 2014; Weston et al., 2019), and simply including computer science content in P-12 settings would not change this. Furthermore, many participants cited the joy of the preschool learners in field experience as influential to their persistence in learning CS. In this way, playfulness both within the preservice classroom and the preschool field experience classroom served to promote resilience among the preservice teachers.

These findings are critical in that challenges and opportunities reported here provide potential guidance for improving teacher education for CS teaching. For example, the finding that preservice teachers can come to find CS to be approachable and fun through playful programing provides direction for CS teacher education (Theme 1). Specifically, robots as co-players in dramatic play in this study embodied CS in a way that disconnected programing from stereotypical activities of computer scientists (Cheryan, Meltzoff, & Kim, 2011). It is also important to note that programing in this study was also connected to the intrinsic interests of our participants, non-CS majors, in teaching young children through lesson design and field experience using their programed robots. Further, negatively valanced utterances during collaborative programing were not necessarily a roadblock to successful debugging (Theme 2). Rather, it was a way of coping with the encountered difficulties alongside peers in collaborative programing. Singing also seems to have played a role in fostering resilience (Clark, 2016; Kang et al., 2018). These findings give insights for scaffolding that could be used with novice programing learners, especially in teacher learning contexts. Perseverance is valued in the present study and thus, trial-and-error is viewed as having positive benefits, especially where social resilience emerged through vicarious success and joint celebration during collaborative debugging (Theme 3).