Effects of games on students’ emotions of learning science and achievement in chemistry

ABSTRACT

This study investigated the time effect of cooperative games on students’ emotions of learning science and the treatment effect on their chemistry achievement. This quasi-experimental study compared the use of cards, board games, and riddles, and the use of conventional paper-and-pencil exercises to learn the basics of chemical elements and compounds, for a duration of 4 weeks. One hundred and fourteen ninth graders at an urban public high school in Taipei were involved. The results revealed that the experimental group had significantly higher positive emotion and lower negative emotion throughout the intervention period. While no time effect was observed for the experimental group, a significant time effect on positive and negative emotions in the comparison group using exercises was found: high achievers decreased their positive emotion, and middle to high achievers increased their negative emotion. Furthermore, low and middle achievers performed better when using games. However, no significant difference for the high achievers of the two groups was discovered. This study showed that conventional exercises were detrimental to middle and high achievers’ learning emotions, although their concepts improved. Science teachers may try innovative activities such as collaborative games to maintain students’ positive emotions and facilitate low achievers’ conceptual learning.

KEYWORDS:

• Academic emotion
• chemistry
• game-based learning
• learning emotion

Introduction

This study focused on both chemistry achievement and emotions related to learning science (hereafter academic or learning emotions). Emotion is at the heart of the attitudes students develop towards science, and is considered as a core aspect of the affective dimension of learning (Maria et al., 2003). The integration of both cognitive and affective learning experiences is essential for meaningful learning (Novak, 2010). Chemistry involves basic and important concepts such as elements and compounds, upon which complicated concepts are built. They are composed of a heavy load of facts for students to memorise. In this early stage of chemistry learning, an innovative teaching approach is crucial to engage students and arouse interest. Among various teaching approaches, game-based learning (GBL) is both educational and entertaining (Jabbar & Felicia, 2015), which attracts students to participate continuously in activities (Garris et al., 2002). Educators make use of this motivation to design materials to promote students’ intellectual development (Petsche, 2011). Accordingly, the current study investigated the chemistry learning emotions and conceptual learning of students by applying GBL in chemistry classes.

Emotions are feelings of arousal, pleasure, or displeasure arising from a complex interaction among subjective and objective factors, mediated by neural and hormonal systems (Kleinginna & Kleinginna, 1981). Much discussion of emotion research has been based on the pleasure-activation model, including pleasure-displeasure or valence dimension and activation or intensity of a valenced affective response dimension (Feldman Burrett & Russell, 1999). Feldman Burrett and Russell also pointed out that more information can be captured if emotions are consciously directed to a specific emotional episode, such as the science learning in the current study. Pekrun et al. (2011) identified various emotions related to learning spread around the four phases by the two dimensions. Typical de- and activated pleasant learning emotions are relief and enjoyment, respectively, whereas de- and activated unpleasant learning emotions are such as boredom and anxiety, respectively. Watson and Tellegen (1985) redefined the affective space by two independent dimensions – positive and negative affective states. The positive and negative emotions are combinations of different degrees of pleasantness and activation.

In the classroom setting, students experience diverse emotions during the learning process which affect their subsequent learning participation and academic performance. Positive emotions could promote effort and perseverance, whereas negative emotions such as sadness, anxiety, frustration, and boredom would diminish involvement (Pekrun et al., 2007). Domain-specific learning emotion and achievement are correlated (Giannakos, 2013). Pekrun et al.’s (2017) longitudinal study revealed that they reciprocally predicted each other over time. Positive emotions (i.e. enjoyment and pride) positively predicted subsequent achievement, and such achievement also in turn positively predicted the learners’ emotions after controlling other factors, including learners’ gender, intelligence, and family socioeconomic status. On the other hand, negative emotions (i.e. anger, anxiety, shame, boredom, and hopelessness) were negative predictors for achievement, and vice versa. For other educational variables, emotions are crucial for students’ personality development, health (Linnenbrink, 2006; Linnenbrink-Garcia & Pekrun, 2011), class climate, self-regulation (Pekrun et al., 2009), classroom engagement (Gong & Bergey, 2020), academic self-efficacy (Oriol-Granado et al., 2017), and life-career decisions (Hartung, 2011). Moreover, emotions shape the relationship between chemistry teachers and their students (Maria et al., 2003). In science learning, emotions un-/stimulate motivation (Bell et al., 2009) and drive autonomous learning of difficult subjects such as physics (Laukenmann et al., 2003). When instructors engage students through enjoyable/boring learning, they encourage/discourage students’ building of longer-term interest, which will in turn develop future engagement and a solid understanding of concepts (Hidi & Renninger, 2006).

Previous studies have shown that GBL is an innovative strategy to engage students actively along with improving their emotions (Jabbar & Felicia, 2015) because of the reward and feedback mechanism (Sadler et al., 2013; Tan et al., 2013). However, evidence is scanty regarding whether games really maintain their impacts and outcomes over time (Jabbar & Felicia, 2015) and are effective for students with different levels of ability. First of all, emotions are momentary occurrences within a given situation and at a specified point in time (Pekrun et al., 2011). Students’ emotions may change over multiple occurrences of gameplay as the novelty wears off. Emotions over time or time effect should be monitored. Second, O’Neil et al. (2005) asserted that individual abilities (e.g. low, middle, and high achievers) should be taken into account if educational games are to be used as a method to enhance learning. The difficulties of gameplay and task challenge are perceived differently by different achievers. Therefore, this study examined the time effect of games on positive and negative emotions of learning chemistry and the treatment effect of card and board games and riddles for the experimental group, and paper-and-pencil exercises for the comparison group on achievement. This study sought to answer the following research questions:

1. What are the effects of games on learners’ positive and negative emotions of learning chemistry over time?

2. What are the effects of games on learners’ chemistry achievement?

3. How do the effects differ among low, middle, and high achievers?

Effects of educational games on emotion

An inherent nature of games is competition (McGonigal, 2011). Volet et al. (2019) found that the competitive nature of collaborative science learning governed students’ positive emotions, such as joy and interest-related emotions. Competitiveness exists in science game-based learning and affects students’ engagement, motivation, and achievement (Jayakanthan, 2002). It motivates students to participate in uninteresting or routine educational activities, thus stimulating their interest. Both review research (Randel et al., 1992) and recent experimental studies on educational games (Hung et al., 2014; Wichadee & Pattanapichet, 2018) have concluded that GBL is consistently perceived as being more interesting than conventional instruction. GBL could evoke interest, enjoyment, involvement, and confidence (Garris et al., 2002). Enjoyment is at the core of game characteristics and is an important positive emotion in educational games to increase learning achievement (Giannakos, 2013). Moreover, joyful learning could attract the students’ engagement, making them more absorbed in the activities and evoking intrinsic motivation to make them like learning and keep studying even outside of class. Such intellectual engagement and intrinsic motivation base performance more on competence than on rewards (McMillan & Hearn, 2008).

Board and card game studies have reported that GBL is enjoyable for learning science and mathematics (Mariscal et al., 2012; Morris, 2011; Mosher et al., 2012; Ng et al., 2007, May; Rastegarpour & Marashi, 2012). A survey by Ng et al. (2007, May) revealed that 87.7% of students enjoy mathematical card games. Mosher et al. (2012) pointed out that students enjoy playing board games and show an exothermic reaction as an alternative to other review activities during class time. Games make learning easier, happier, more fun, more student-centered, and even more effective (Prensky, 2003). For example, Morris (2011) administered the Go Chemistry card game in which students earn points by using cards to correctly form the formulas of covalent and ionic compounds. He reported that most students enjoyed the game and understood compound formation and nomenclature better as a result of playing. Mariscal et al. (2012) stated that the Families of Chemical Elements Game provided an enjoyable way of working together to learn the names and symbols of the chemical elements, the periodic table, and the objects of daily life in which the different elements are found. On the other hand, in terms of negative emotions, a limited number of previous studies have focused on anxiety. Hung et al. (2014) used a digital game on e-books and assessed elementary school students’ mathematical anxiety. No significant differences were found among three groups (i.e. the GBL approach using e-books, a learning system on e-books, and traditional teacher-directed instruction). Nevertheless, mathematical anxiety ratings from the pre- to post-questionnaire of both the GBL and e-learning groups decreased after intervention, while that of the traditional instruction group increased. How to reduce anxiety is an important issue (Li et al., 2011), but it is not easy to achieve. To the best of our knowledge, no previous studies have measured the effect of multiple interventions on anxiety. The existing literature focusing on emotions using GBL mainly applies one-shot interventions. As emotions are crucial to learning, how they change over time deserves further evaluation.

Effects of educational games on achievement

GBL is an effective form of instruction to improve the learning achievement of elementary school students (Hobbs & Yan, 2008), middle school students (Giannakos, 2013), high school students (Papastergiou, 2009), and even undergraduate students (Boeker et al., 2013). In terms of board and card games, a previous study found that students who participated in burn game activities achieved the main learning aims, namely increasing their knowledge of burn care and stimulating discussion (Whittam & Chow, 2017). Likewise, a pediatric board game was found to motivate students to advance their knowledge of pediatric medicine (Ogershok & Cottrell, 2004). Similar experimental studies have revealed that board and card game instruction is excellent for improving achievement (Dieser & Bogner, 2016; Ng et al., 2007, May; Rastegarpour & Marashi, 2012). For chemistry learning, board and card games have been successfully used to teach various topics, such as chemical compounds (Rastegarpour & Marashi, 2012), the periodic table of the chemical elements (Alexander et al., 2008), organic chemistry in terms of recognition and identification of organic functional groups (Welsh, 2003), stereochemistry of carbohydrates (Costa, 2007), and elemental symbols (Alexander et al., 2008). In the current study, we focused on atomic symbols, atomic number and weights, chemical properties, the atomic positions in the periodic table, relative activity, and simple compounds.

Finally, O’Neil et al. (2005) articulated that educational games are not sufficient for learning, asserting that individual differences should be taken into account if educational games are to be used as a method to enhance learning. For example, Wrzesien and Raya (2010) found that sixth graders reported higher motivation and engagement levels as a result of playing a science-based game; however, there was no evidence to show that the game led to significant learning advancements over the traditional class. Individual differences such as academic ability and emotion should be taken into account to fully understand the impact of a game on students’ learning.

The study design

This study investigated secondary students’ emotions and achievement as a consequence of playing board and card games and riddles in chemistry. The unit of pure substance was chosen due to its dull and boring content. Each week, the students were informed of the schedule and were encouraged to study the textbook before the class. In the class, games for the experimental group and paper-and-pencil exercises for the comparison group were prepared for reviewing concepts including atomic symbols, atomic number and weights, chemical properties, the atomic positions in the periodic table, relative chemical activity, and simple compounds. To wrap up the class, the instructor guided the students to discuss the concepts learned from the games or exercises and to clarify problems raised during the activities.

Game design

Hays (2005) defines a game as a competitively constructed activity having a specific goal and a set of rules in a particular context. In other words, games are systems with clear rules in which participants compete for a goal (Deterding et al., 2011, September). Several traits are shared by games: a goal, rules, and feedback. First of all, the goal is the specific result that players strive to achieve. It provides players with a sense of mission for them to concentrate and continuously participate in (McGonigal, 2011). Second, players are restricted by rules to achieve the goal which then pushes them to foster strategic thinking. Some games allow players to create their own rules, which facilitates creativity. Third, a feedback system tells players their distance from the target, usually in the form of points, levels or progress columns, or provides the players with objective results at the end of the game. Timely feedback guides the players toward the goal, letting them know that the goal could be achieved, and giving them continued motivation. In the current study, the goals of the game were similar to the learning objectives in each unit: (a) The periodic table, (b) The metal and non-metallic elements, (c) Forming compounds, and (d) Chemical formulas. The rules combined pedagogies such as collaborative learning and active learning, and concepts such as the chemical properties. To win a game, students had to understand and apply chemical concepts.

Four games played by the students in the experimental class during the four weeks were:

Chemistry Uno

This activity was inspired by the classical card game Uno. Each card was printed with an element symbol and its atomic number and weight. Through the game process, students could distinguish metallic from non-metallic elements, understand the atomic and mass numbers, understand the number of neutrons, protons, and electrons, classify the elements based on the same family (i.e. having similar chemical and physical properties) and period (i.e. having the same number of electron shells), and compare the activity of elements. The rules and implementation process of this game are described in Table 1.

Table 1. Chemistry Uno.

Number of Players	4–5 students
Time	25 min
Resources	80 symbolic cards
Game Play	Each student draws one card. Everyone puts their card on the table at the same time. The student who has the highest atomic number is the first player. The other students take turns to play. Every player starts with 10 cards. The rest of the cards are placed in a draw pile facing down. The first player puts a card on the table and names the rule, that is, what to match/compare. For example, the card should be a larger/ smaller number of electron/ proton/ neutron/ mass, the same metal or non-metal element, the same family/ period elements, and more or less active, et cetera. Gameplay follows a clockwise direction. If a player has no matches, he/she has to draw a card from the draw pile. If that card matches the rule, play it. Otherwise, the player has to keep the card and the game moves on to the next person. In addition, a universal card could be used to Pass, Swivel, or Change the rules. Pass means skip the next player. Swivel means that the order of the play is switched to the previous player and the opposite direction. Change the rules allows initiation of a different rule.
Game Score	The first-out player scores 10 points, the second player scores 8 points, and the third, fourth, and fifth score 6, 4, and 2 points, respectively.

The Ultimate Decryptor

This game used riddles (Figure 1) and collaborative learning (Gokhale, 1995). Students discussed in groups of 4–5 to solve the riddles. Each riddle had four hints and was presented by PowerPoint. Each group wrote their answers on the worksheet (Figure 2). The first row was their guess after the first hint was revealed. They could change it and put down different answers in the second to fourth rows as more hints were revealed. They were encouraged to discuss, share their thoughts, listen carefully to the comments of each member of the group, and reconsider their own judgements/opinions. They then collectively decided on the answer. The group received 5, 3, 2, and 1 points if they solved the riddle using 1, 2, 3, and 4 hints, respectively. Through this game, students could review the characteristics and application of metal and non-metallic elements.

Figure 1. PPT slide with four hints of sulfur. (1) 16; 32, (2) Blue flame, (3) Black powder, (4) Yellow solid.

Figure 2. Student’s worksheet for the Ultimate Decryptor.

Chemical spelling

This game was about forming compounds using two or more chemical element counters and calculating the molecular weight. Table 2 displays the details of the game.

Table 2. The chemical spelling setting.

Number of Players	4–5 students
Time	25 min
Resources	Each student draws one atomic counter. Everyone puts their counter on the table at the same time. The student who has the highest atomic number is the first player. Every player starts with nine atomic counters. The rest of the counters are placed in the bag. The first player puts one or more counters on the table (Fig. 3). Each counter must have at least one side connecting to the others to form correct compounds (s). The number of counters for each turn is unlimited. It is not permitted to flip/change the face of the atomic counter once it is on the table. Gameplay follows a clockwise direction. Players can move the atomic counters to combine with others to create a desired compound, but they cannot move a counter that is surrounded by five or more atomic counters. If the player has no matches or cannot move any counters to form different compounds, he/she has to take a counter from the bag. When any team member has played all their counters, or the game can no longer continue, the game ends.
Game Score	The first-out player scores 10 points, the second player scores 8 points, and the third, fourth, and fifth score 6, 4, and 2 points respectively. If all players still have atomic counters, the player with the least is the winner.
Extension Activity	Students write the compounds spelled on the table on the worksheet (the more the better, not repeatable), and calculate the molecular weight. The group that has the highest number of correct compounds wins the prize.

Chemical formulas

This game was similar to the Ultimate Decryptor, but requires more reasoning and deep thinking. Through this game, students could identify the composition and characteristics of acids, bases, salts, and organic compounds. The rules and scores are the same as those of the Ultimate Decryptor.

Pedagogy underpinning the games

Experiential learning theory

GBL is supported by the experiential learning theory (Kolb, 1984). Experiential learning relies much on the meaning-making process of the individual’s direct experience. Learning is a result of the interaction among experience, cognition, and behaviour (Kolb, 1984). Concepts are constructed via concrete and active experiences as well as reflective observation (Wrzesien & Raya, 2010). Through game rules and game play, students could engage in direct experience, reflect on learning, and receive immediate feedback from peers. For example, in the game Chemistry Uno, the first player or whoever plays a universal card names a rule, whether the next card should be a larger/ smaller number of electron/ proton/ neutron/ mass, the same metal or non-metal element, the same family/ period elements, and more or less active, et cetera. The following players engage in direct experience of applying the concepts. A player then receives feedback from others if she/he plays the wrong card. Such feedback and active participation sustain effective learning.

Collaborative learning

This study used cooperative games, which are rooted in social constructivism. Collaborative learning is a form of instruction whereby students with various performance levels work in groups to achieve certain goals. It builds a relationship among learners that fosters positive interdependence, individual accountability, and interpersonal skills (Gokhale, 2012). In the current study, several rules were applied to promote peer evaluation and discussion during the games. For example, in Weeks 2 and 4, students discussed in groups the answers to riddles, for which they had to apply chemistry concepts and use their imagination. They had to teach and support each other because any of them could be randomly assigned to answer. In Week 3, the extension activity of Chemical Spelling asked each group to collectively identify different chemical compounds formed by atomic counters. Through collaboration, students’ knowledge and abilities developed. The success of collaborative learning could be explained by two main perspectives, namely cognitive and social. Collaborative learning promotes conceptual understanding through mutually shared cognition (Webb & Palincsar, 1996), and social interaction is supported by social constructivism (Wu et al., 2012).

Finally, we used collaborative learning to moderate the challenge level of the games. Challenges are needed for motivating players to play and for learners to learn. Too much challenge could be detrimental for engagement in a game, especially for low achievers. By applying appropriate game rules, students are socially situated to construct knowledge through peer interaction. As a result, low achievers would not be discouraged by difficult chemistry concepts, but would be able to participate in and learn from the games.

Paper-and-pencil design

The comparison group worked individually on the paper-and-pencil exercises that the teacher had compiled. The classroom was in a highly structured setting, where students sat row by row with limited free space, and chemistry was taught by lecture-dominated instruction and paper-and-pencil exercises. Both the setting and instruction style are typical in Chinese culture (Gong & Bergey, 2020). In the current study, the exercises were multiple-choice questions covering the same topics and concepts as those in the games. The instructor checked the answers to the exercises and explained them. An example question is: Regarding the description of sulfur, which following option is wrong? (Figure 3)

1. A purple-black solid.

2. An important chemical used to make black powder.

3. When burning, a blue flame appears.

4. Its atomic number is 16.

Figure 3. (a) Player 1 uses 3 counters to form H₂O. (b) Player 2 adds 3 more to form CH₃OH. (c) Player 3 adds Na and Cl. (d) Player 4 moves Na to form a new compound.

Method

Participants

One hundred and fourteen students from four chemistry classes participated in this study. The four classes were randomly assigned to be the experimental (n = 56) and comparison (n = 58) groups. These students were ninth graders at an urban public high school in Taipei and were aged 14–15 years (57.9% males, 42.1% females). The pure substance unit was taught in the eighth grade. In the ninth grade, the participants reviewed the concepts to prepare for the national examination.

Procedure

In this quasi-experimental study, the card and board games and the riddles were implemented with the experimental group, and the paper-and-pencil exercise was administered to the comparison group. Both groups were taught by the same instructor. This instructor had been teaching chemistry for 9 years. The intervention lasted for 4 weeks. Students reviewed the content of pure substance for 45 min/week with four topics covering: (a) The periodic table, (b) The metal and non-metallic elements, (c) Forming compounds, and (d) Chemical formulas. The content comprised atomic symbols, atomic numbers, atomic and molecular weights, the atomic positions in the periodic table, the number of neutrons, protons, and electrons, the groups of metal and non-metallic elements, classification of the elements based on the same family (i.e. similar chemical and physical properties) and period (i.e. the same number of electron shells), relative activity of elements, chemical properties, the characteristics and application of metal and non-metallic elements, and formulas and properties of common acid, base, salt, and organic compounds.

The teaching time, content, and instructor were controlled to be the same for both groups. The following data were collected. First, both groups completed a conceptual test before and after the intervention to measure the learning achievement. Second, their emotions were measured weekly at the end of the class using a shortened version of the Achievement Emotion Questionnaire.

Conceptual test and survey

A conceptual test was administered to assess the students’ achievement of pure substance concepts. The test time before and after the experiment was 45 min. In the current study, four topics were covered in the 40 multiple-choice items. The content validity of the multiple-choice questions was established based on experts’ reviews, including those of two teachers and a science professor. A two-way specification matrix was used to align the test items with the chemical concepts and cognitive levels to make sure that the number of items was distributed evenly across the four topics and covered different cognitive levels. There were 11 items (3, 5, and 3 at the knowledge, comprehension, and application and analysis levels, respectively) regarding common elements and the periodic table, 13 items (3, 6, and 4 at the knowledge, comprehension, and application and analysis levels, respectively) testing chemical properties, 8 items (3, 3, and 2 at the knowledge, comprehension, and application and analysis levels, respectively) related to common acid, base, and salt, and 8 items (2, 4, and 2 at the knowledge, comprehension, and application and analysis levels, respectively) about chemical formula. Based on the pre-test, the reliability was high, KR 20 = .89. The test questions are the same before and after the intervention, but the number and the options of the multiple-choice questions were rearranged.

The Achievement Emotion Questionnaire which was coded using a 5-point Likert scale ranging from strongly disagree ( = 1) to strongly agree ( = 5) was developed by Pekrun et al. (2011) for assessing college students’ emotions. The original questionnaire includes seven emotions: enjoyment, pride, anger, anxiety, shame, hopelessness, and boredom. Its 60 items are framed in the contexts of attending class (n = 19), studying and doing homework outside class (n = 18), and taking tests and examinations (n = 23) in a specific subject such as science or mathematics. To reduce the response time, we selected only items from the dimensions that were more directly relevant to the current study. We first considered those in the context of attending science class. Then out of the seven emotions, we took enjoyment, pride, anxiety, and boredom and excluded anger, shame, and hopelessness based on the relatedness to GBL. As a result, 11 items, including four items from enjoyment, two each from pride and anxiety, and three from boredom, comprised the positive and negative emotion sub-scales. They were translated into Chinese and then back into English to ensure the same meaning. Both sub-scales had high internal consistency (α = .91 for both).

The confirmatory factor analysis (CFA) was performed to assess the construct validity of the positive and negative emotions in the current data. The recommended values for a good model fit were: χ²/df < 3 (Kline, 2016; Zhang, 2016), the comparative fit index (CFI) and the Tucker-Lewis index (TLI) ≥ .90 (Zhang, 2016), and Standardized Root Mean Square Residual (SRMR) < .05 (Byrne, 2016). SRMR was used due to the small sample size (Iacobucci, 2010). The CFA result showed that both the positive and negative emotions had a well-fitting model (Table 3; Figure 4). Based on the internal reliability and CFA analyses, the positive and negative emotion scales were reliable and valid.

Figure 4. The measurement model of the achievement emotions.

Table 3. Model Fit of Positive and Negative Emotions.

Emotion	N	χ2/df	CFI	TLI	SRMR
Positive	114	2.97	.95	.91	.04
Negative	114	1.73	.99	.97	.02
Pooled	114	1.96	.96	.95	.04
Criteria		<3	>.90	>.90	<.05

Data analysis

The students were divided into low, middle, and high achievers based on the pre-test scores. Those with scores above 69 are high achievers (top 33%), those who had scores between 47.5∼69 are middle achievers, and those whose scores were below 47.5 are low achievers (bottom 33%). To test any significant difference between the comparison and experimental groups in learning emotions across the 4 weeks, Mixed ANOVA, also called repeated-measures analysis of variance (RM-ANOVA) with treatment as a between-subject factor, was performed on the low, middle, and high achievers. In the Mixed ANOVA, Mauchly’s test could detect if the data violated the assumption of sphericity. The correction was selected to resolve the violation based on epsilon. The results and degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity (ε < .75) or Huynh-Feldt estimates of sphericity (ε >.75). If there was a significant interaction between time and group, an independent t test on groups and a repeated measures of ANOVA on the emotions were performed to test both main effects.

Concerning conceptual learning, an independent t test was carried out to detect any difference between the experimental and comparison groups in the beginning. A paired t test was run to measure whether the high, middle, and low achievers in the two groups were different before and after the intervention. Moreover, ANCOVA was planned to measure the extent to which the treatment influenced the high, middle, and low achievers’ learning achievement. In the cases where the data violated the assumption of homogeneity of regression slopes, that is, the correlation between the pre- and post-test differed significantly across the comparison and experimental groups, an alternative correction by Johnson-Neyman (Fraas & Newman, 2005) was applied to find the point of intersection (crossover point) of the regression lines. It revealed for which region there was a significant difference between groups.

Results

Emotion

This section presents a comparison of the students’ learning emotions while playing the games and completing the paper-and-pencil intervention using mixed ANOVA. The means and standard deviations are summarised in Tables 4 and 5, and are visualised in Figures 5 and 6.

Figure 5. Average scores of positive emotion from Weeks 1–4. Note: LA = Low achievers; MA = Middle achievers; HA = High achievers.

Figure 6. Average scores of negative emotion from Week 1 to Week 4. Note: LA = Low achievers; MA = Middle achievers; HA = High achievers.

Table 4. Means and standard deviations of positive emotion between the comparison and experimental groups over 4 weeks.

Group	n	Week 1 M (SD)	Week 2 M (SD)	Week 3 M (SD)	Week 4 M (SD)
High achievers
Comparison	17	3.60 (.47)	3.20 (.67)	3.22 (.71)	3.35 (.63)
Experimental	16	3.70 (.73)	3.86 (.60)	4.00 (.56)	3.69 (.76)
Middle achievers
Comparison	21	2.87 (.49)	2.74 (.62)	2.58 (.55)	2.57 (.68)
Experimental	21	3.63 (.66)	3.92 (.70)	3.79 (.58)	3.60 (.68)
Low achievers
Comparison	20	1.73 (.40)	1.98 (.50)	2.09 (.45)	1.89 (.52)
Experimental	19	3.45 (.74)	3.13 (.46)	3.07 (.59)	3.35 (.88)

Table 5. Means and standard deviation of academic negative emotion between the comparison and experimental groups over the 4 weeks.

Group	n	Week 1 M (SD)	Week 2 M (SD)	Week 3 M (SD)	Week 4 M (SD)
High achievers
Comparison	17	2.09 (.42)	2.18 (.53)	2.44 (.49)	2.33 (.55)
Experimental	16	1.83 (.66)	1.65 (.64)	1.44 (.43)	1.87 (.69)
Middle achievers
Comparison	21	2.93 (.46)	3.11 (.63)	3.14 (.53)	3.33 (.31)
Experimental	21	1.95 (.60)	1.81 (.66)	1.92 (.62)	2.05 (.70)
Low achievers
Comparison	20	4.08 (.57)	3.96 (.58)	3.98 (.55)	4.00 (.52)
Experimental	19	2.69 (.59)	2.37 (.60)	2.73 (.71)	2.74 (.71)

Positive emotion

There was a significant interaction between time and group for the low [F_{Time x Group} (1.76, 65.17) = 5.58, p = .008] and the high achievers [F_{Time x Group} (1.78, 55.32) = 4.16, p = .025], but not for the middle achievers [F_{Time x Group} (2.73, 109.08) = 2.24, p = .094]. The significant interaction indicated that the groups changed over time in different ways (Figure 5).

For the particular case of the low achievers, it could be seen that the positive emotion of the experimental group decreased from Weeks 1–3, whereas that of the comparison group increased. Nevertheless, further analysis of the low achievers revealed that there was a statistically significant difference between the comparison and experimental groups for positive emotion during the 4 weeks, t(37) = 9.04, 7.44, 5.83, and 6.36, p < .001, respectively. Those in the experimental group always had higher positive emotions. In fact, their emotions were as high as those of the high achievers in the comparison group, as shown in Figure 5. Moreover, the RM-ANOVA pointed out a significant effect of time on positive emotion in the comparison group, F_Time (3, 57) = 4.36, p = .008, η² = .186, but not in the experimental group, F_Time (1.33, 24.57) = 2.22, p = .143, η² = .110. This indicated that the positive emotion of the students in the experimental group did not change during the 4 weeks. A Bonferroni post hoc analysis was performed on the comparison group and found a significant improvement in positive emotion from Week 1 to Week 3 (1.73 ± .40 vs. 2.09 ± .45). However, positive emotion was still far below that of the experimental group, suggesting an absence of enjoyment and pride in the conventional paper-and-pencil instruction.

The analysis of high achievers using an independent t test showed that there was a significant difference between the comparison and experimental groups for Weeks 2 [t(31) = 3.00, p = .005] and 3 [t(31) = 3.50, p = .001]. The RM-ANOVA revealed that there was a significant difference over time on positive emotion in the comparison group, F_Time (3, 48) = 4.87, p = .005, η² = .233. According to the Bonferroni post hoc tests, the conventional paper-and-pencil intervention decreased students’ positive emotion from Week 1 to Week 2 (3.60 ± .47 vs. 3.20 ± .67) and from Week 1 to the last week (3.60 ± .47 vs. 3.35 ± .63), which was statistically significantly different (p = .006, p = .031, respectively). This indicated that the 4 weeks of conventional instruction reduced the high achievers’ sense of enjoyment and pride. However, for the experimental group, the RM-ANOVA with a Green-house Geisser correction determined that the mean of positive emotion did not differ statistically significantly between time points, F_Time (1.31, 19.69) = 1.30, p = .281, η² = .080.

For the middle achievers, for which no significant interaction between time and group was detected, the results of RM-ANOVA informed no significant main effect of time on positive emotions, F_Time (2.73, 109.08) = 2.29, p = .089, η² = .054. However, there was a significant main effect of group, F_Group (1, 40) = 47.95, p < .001. This suggested that students who learned via the board game activities had significantly higher enjoyment and pride than the students in the traditional paper-and-pencil class throughout the 4 weeks.

Negative emotion

There was significant interaction between time and group for the high achievers [F_{Time x Group} (2.74, 84.90) = 4.95, p = .004, η² = .138], while no interaction was detected for the low [F_{Time x Group} (3, 111) = .97, p = .409, η² = .026] or middle achievers [F_{Time x Group} (3, 120) = 1.32, p = .272, η² = .032]. The experiment exhibited a significant effect on negative emotion for both low [F_group (1, 37) = 101.56, p < .001, η² = .733] and middle achievers [F_group (1, 40) = 76.36, p < .001, η² = .656]. Those in the experimental group had consistently lower negative emotions than those in the comparison group. Regarding the time effect, there was significant difference for the middle achievers [F_Time (3, 120) = 3.00, p = .033, η² = .070], but not for the low achievers [F_Time (3, 111) = 1.69, p = .173, η² = .044]. A RM-ANOVA showed that middle achievers’ negative emotion changed significantly over time for the comparison group [F_Time (3, 60) = 4.05, p = .011, η² = .168], but not for the experimental group [F_Time (2.68, 53.65) = .94, p = .421, η² = .045]. According to the Bonferroni post hoc analysis, the middle achievers’ anxiety and boredom increased significantly from Weeks 1–4 (2.93 ± .46 vs. 3.33 ± .31) when conventional paper-and-pencil practices were used, whereas the negative emotion remained low when games were used.

For further analysis of high achievers, an independent t test revealed that there was a statistically significant difference between the comparison and experimental groups for Week 2 [t(31) = 2. 57, p = .015], Week 3 [t(31) = 6.21, p < .001], and Week 4 [t(31) = 2.09, p = .045]. Also, there was a significant effect of time on negative emotion in the comparison group only [F_Time (3, 48) = 3.96, p = .013, η² = .198]. High achievers in the comparison group significantly increased their anxiety and boredom from Weeks 1–3 (2.09 ± .42 vs. 2.44 ± .49).

Simply put, for low and middle achievers, when they learned via game activities, they had significantly less and stable feelings of anxiety and boredom. Likewise, high achievers of the experimental group tended to sense less negative emotion than the comparison group, except in Week 1. Moreover, in the comparison group, low achievers always possessed high negative emotion, and the middle and high achievers increased their negative emotion over time.

Achievement

The pre-test scores showed that the comparison [M (SD) = 57.05 (18.98)] and experimental [M (SD) = 57.90 (20.13)] groups had no significant difference, t(112) = .24, p = .811, in the beginning. It was confirmed that the two groups randomly assigned by the researchers had no difference in their initial level of knowledge. Based on the results of paired-sample t tests, there was a statistically significant improvement from the pre- to post-tests in both the game and paper-and-pencil groups for all achievement levels (Table 6). According to Cohen’s d value, the experimental group ranged from 1.69–2.42, and the comparison group ranged from 0.64–1.26, indicating that the two teaching methods both had large effects on the improvement of low, middle, and high achievers.

Table 6. Means, standard deviations, and comparisons of pre- and post-tests conceptual understanding for low, middle, and high achievers.

Group	n	Pre- test^a M (SD)	Post- test^a M (SD)	Paired t-test	Cohen’s d^b
Experiment
High Achievers	16	82.81 (8.00)	89.69 (7.00)	7.42***	1.85
Middle achievers	21	58.69 (7.57)	69.17 (10.79)	7.74***	1.69
Low achievers	19	36.05 (8.05)	46.32 (8.75)	10.55***	2.42
Comparison
High achievers	17	80.29 (6.72)	86.03 (7.50)	5.17***	1.25
Middle achievers	21	58.33 (5.49)	63.45 (6.82)	5.77***	1.26
Low achievers	20	35.88 (7.36)	38.38 (6.55)	2.87*	0.64

*p < .05, ***p < .001.

^aTotal score was 100.

^bEffect size d = < .3 small effect, .5∼.8 medium effect, and >.8 large effect.

When comparing the experimental and comparison groups, the data violated the assumption of homogeneity of regression slopes, F (1, 110) = 6.34, p < .05. This indicated that the correlation between the pre- and post-tests differed significantly across the comparison and experimental groups. Therefore, the Johnson-Neyman procedure was performed to calculate the simultaneous regions of significance, that is, the range of the pre-test scores for which the experiment had a significant influence on the post-test. The result revealed that for students with a pretest score between 81.12 and 100, that is, the highest 18.4%, there was no significant difference in the post-tests of the experimental (n = 10) and comparison groups (n = 11). However, for students with pre-test scores lower than 81.12, the post-test scores of the game group were significantly higher than those of the comparison group after the pre-test was controlled (Figure 7).

Figure 7. The regression line of the comparison and experimental groups. No significant differences in the post-test score for pretest scores above 81.12 between the game and paper-and-pencil classes.

Discussion and conclusion

This study contributes to an understanding of GBL in science education by revealing the effects on emotions over time, and the different effects for students of different abilities. The results revealed that the students, especially the low and middle achievers, learned better chemistry concepts with considerably more enjoyment and pride and less anxiety and boredom than their peers in the comparison class. Similar positive results of implementing board and card games instruction on achievement (Dieser & Bogner, 2016; Ng et al., 2007, May; Rastegarpour & Marashi, 2012), enjoyment (Mariscal et al., 2012; Morris, 2011; Mosher et al., 2012; Ng et al., 2007, May; Rastegarpour & Marashi, 2012), and anxiety (Hung et al., 2014) were found to those in the literature. Rastegarpour and Marashi (2012) investigated the effects of a card game, a computer game, and the traditional teaching method on learners’ chemistry concepts and revealed that teacher-made instructional card games are effective for learning chemistry. The current study further showed that the impact on middle and low achievers is profound.

Unlike low and middle achievers who always prefer games, for certain weeks there was no significant difference between the effects of learning via games and conventional instruction on the positive and negative emotions of the high achievers. Moreover, the top 18.4% of students performed similarly in the post-test regardless of the treatment. This indicates that high achievers’ emotion and conceptual learning are less influenced by the instruction. Some previous studies have found that high achievers tend to have better learning strategies (Law et al., 2008; Yip & Chung, 2005) and self-efficacy. These enable them to actively process information and sustain motivation, thereby influencing their mastery of material and subsequent academic achievement (Pintrich et al., 1993).

Furthermore, there was no time effect on positive and negative emotions in the experimental class. Low, middle, and high achievers all enjoyed, felt proud, and expressed low levels of anxiety and boredom during the 4 weeks of game intervention. On the other hand, the time effect on positive and negative emotions was shown to vary for the conventional paper-and-pencil class. Low achievers increased their positive emotion, while high achievers reduced it. This indicates that low achievers might be able to accept traditional learning strategies, although game intervention could render significantly higher positive emotion. In addition, middle and high achievers increased their anxiety and boredom, feeling bored if instructors kept using traditional instruction.

Finally, the inherent challenge in a game should be carefully considered (Jabbar & Felicia, 2015). Usually, too much challenge has a preventive effect rather than inspiring enthusiasm to engage (Ke & Abras, 2013). A balance between challenge and cognitive skills creates an optimal flow, which may encourage the participants to pursue greater learning interest (Yang et al., 2020). In the current study, students played games in groups. Collaborative learning might moderate the challenge of the current games to be suitable for even low achievers. Collaboration makes learning more effective and efficient than individual learning because it decreases task complexity (Paas & Sweller, 2012). It may then influence the students’ emotion during activities. Hence, the low and middle students who learned via games had significantly more positive and less negative feelings. Our results contradict those of Bottino et al. (2007) who found that low achievers remained anxious about succeeding in a game even after the level was set to be suitable for their skills, and high achievers were quickly frustrated due to making mistakes. Another study by Hung et al. (2014) mentioned no significant difference in the effects of the GBL approach using e-books, a learning system with e-books, and traditional teacher-directed instruction on anxiety, although a pre- to post-anxiety questionnaire showed that anxiety tended to decrease after game exposure. In the current study, which incorporated collaborative learning, we found that anxiety and boredom significantly decreased after using game instruction. This is very important due to many low achievers becoming paralysed by anxiety, and hence failing to use their learning strategies to be successful (VanZile-Tamsen & Livingston, 1999).

Limitation of the study

This study measured only emotions after innovative and conventional strategies were applied each week. It did not measure pre-positive and negative emotions, and hence students’ initial emotions were unidentified. Moreover, this study provided evidence of the effect of games on subject matters that are linked to low cognitive levels such as recitation and comprehension. More class time and different teaching designs may be needed for the learning of high cognitive objectives. In addition, future research may implement qualitative methods to understand learning strategies that high achievers use, so as to explain how they manage learning emotions in different learning conditions.

Implication

This study has several theoretical and practical implications. First, we showed that GBL could increase positive emotion and decrease negative emotion in multiple-shot interventions. The experimental group felt enjoyment and pride and were not anxious or bored when learning the rote materials in chemistry. This intervention was also excellent in terms of improving students’ achievement. Secondly, instructors can use games to motivate students’ autonomous learning. In the present study, the teacher did not provide instruction. Students autonomously prepared themselves for the games. Third, GBL positively synergized with collaborative learning. Students could understand better based on collaborative activity and experience. Collaboration, problem solving, communication, and critical thinking are referred to as higher order thinking skills/ twenty-first-century skills/ future ready skills. Game-based collaborative learning can help students develop the skills that are essential for their future jobs. Future research may investigate these skills in more depth. Lastly, board and card games allow teachers to design and deliver creative instruction using minimum resources, which may benefit education systems that have limited resources (Husnaini & Chen, 2019).

Disclosure statement

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

Additional information

Funding

This work was supported by The Academy of Finland: [grant number 318380]; Ministry of Science and Technology, Republic of China: [grant number MOST 105-2511-S-011-009-MY3,MOST 108-2511-H-011-002-MY4].