Individual or collaborative projects? Considerations influencing the preferences of students with high reasoning ability and others their age

ABSTRACT

Conditions influencing 328 students’ (Grades 6-8) preferences for collaborating or working alone on challenging projects were investigated, as well as their potential interactions with ability, grade and sex. Each student completed the Cognitive Abilities Test (Form 7) and Project Context Survey. No overall preference for individual or collaborative projects was found. Students’ preferences were sensitive to features of the context (subject, nature of the task and social dynamics). Individual projects were preferred in art and shared projects in science and social studies. Students with high ability and boys preferred individual projects in Math. Principal components analyses revealed three contextual considerations influenced students’ desire to work on projects alone (enjoyment, optimizing the outcome, and risk management) and five influenced the appeal of collaborating (inclusiveness and trust, access to the strengths of others, their perceived need for support, familiarity, and fair assessment). High ability students were more concerned with the efficiency and quality of their work, and their grades while others their age were more influenced by the potential for fun. Grade 8 students were more concerned with risk management and the assessment process than younger students. If the safe, supportive, fair conditions they sought for collaborating were not available, students’ default preference was to work alone on a challenging project.

KEYWORDS:

• High ability
• Collaborative
• Individual
• Projects
• Preference

Theorists’ and practitioners’ enthusiasm for collaborative learning and its potential benefits are at an all-time high, however students’ perspectives on those experiences have always been mixed. While Jang, Reeve, and Halusic (2016) indicated that teaching students in the ways they prefer can play a significant role in students’ learning, the considerations influencing their preferences for working with others and individually have received less attention. One’s preferences for working alone and with others have been found to be sensitive to the nature of the task and conditions surrounding the task (Cantwell & Andrews, 2002; French, Walker, & Shore, 2011; Kanevsky, 2011; Koutrouba, Kariotaki, & Christopoulos, 2012; Walker & Shore, 2015; Wismath & Orr, 2015), as well as one’s cognitive ability (French et al., 2011; Pyryt, Sandals, & Begoray, 1998; Ramsay & Richards, 1997; Ricca, 1984; Samardzija & Peterson, 2015; Sears & Reagin, 2013; Walker & Shore, 2015).

In the past, a student’s preference for learning alone or with others was conceptualized as a global, dichotomous, stable trait, a “learning style” (Pyryt, 1991), however an accumulating body of research and critique has challenged the concept of fixed learning styles (e.g., Burns, Johnson, & Gable, 1998; Cuevas, 2015; Curry, 1990; Pashler, McDaniel, Rohrer, & Bjork, 2008; Reynolds, 1997; Sadler-Smith, 1997). It argues that one’s preferences are context-specific and has broadened the discussion from viewing them narrowly as a trait (a characteristic of the individual) to seeing them as a state that results from complex, dynamic interactions between a learner and the learning context (Curry, 1990; Riding, 1997; Walker, Shore, & French, 2011).

Researchers sharing a context-sensitive orientation to learning preferences have found that students’ desire to learn with or without others varies depending on what is to be learned, how, with whom, and other considerations (e.g., French et al., 2011; Kanevsky, 2011, 2015; Koutrouba et al., 2012; Samardzija & Peterson, 2015; Walker & Shore, 2015). For example, more students prefer to work with others when the conditions for working in a group include working with others they have chosen. Alternatively, most students prefer working alone when they feel collaborating may result in potential conflict, unfair workload, or a poor grade (Kanevsky, 2011, 2015; Samardzija & Peterson, 2015; Walker & Shore, 2015). This study further investigates the ways in which interactions among learner characteristics (reasoning ability, age and sex) and features of the context influence the appeal of working alone or with others.

“Contexts can consist of subject-matter domains (e.g., science), specific tasks/problems (e.g., a textbook problem to solve), social interactions (e.g., caretaking routines between a parent and child), and situational/physical settings (e.g., home, classrooms, museums, labs)” (APA Coalition for Psychology in Schools and Education, 2015, p. 10). With this definition in mind, we held the second element of context, the task, constant rather than studying preferences broadly, across tasks. We sought a better understanding of the ways in which known or new combinations of these features of the context influence learners’ task-specific preferences for individual and group work when they are engaged in a frequently assigned task, a challenging project. Projects were identified as the target task because: (a) they can be undertaken alone or as a shared endeavor, (b) they are an essential element of many forms of inquiry-based learning that were strongly recommended and frequently implemented in schools participants attended, (c) they often offer students opportunities for deeper learning than smaller academic tasks, (d) all students had experienced individual and group projects by the time they reached middle school (Grades 6, 7, and 8), and (e) students in our focus groups considered them complex, “high stakes” assignments that played a significant role in their grades. The popularity of complex, shared assignments makes it increasingly important to provide educators with the evidence-based guidance they need to create contexts that are not only effective but attractive to learners.

Academic and social outcomes of learning alone versus with others

Numerous meta-analyses have examined the academic and social outcomes of group work as well as differences in the outcomes of collaborative and individual learning activities (e.g., Abrami, Lou, Chambers, Poulsen, & Spence, 2000; Kulik & Kulik, 1987; Lou, Abrami, & d’Apollonia, 2001; Lou, Abrami, & Spence, 2000; Lou et al., 1996; Pai, Sears, & Maeda, 2015; Roseth, Johnson, & Johnson, 2008; Slavin, 1987, 1990). For example, Pai et al. (2015) reported an estimated overall effect size of 0.30 in favor of small group learning over individual after analyzing the results of 24 studies that had compared students’ performance on transfer tasks. Roseth et al. (2008) also found “higher achievement and more positive peer relationships were associated with cooperative rather than competitive or individualistic goal structures” (p. 223) in their meta-analysis of 148 studies involving middle school students. These authors repeatedly reminded readers to consider the variability evident in the findings of individual studies included in their meta-analyses. Many attributed it to differences in the nature of the individual, cooperative, and collaborative learning contexts involved (e.g., Lou et al., 1996).

Contextual considerations related to working alone and with others

The influences of the four aspects of the context on individual and group work have been extensively investigated. This brief review of the literature was limited to one setting: school. In classrooms, students can share tasks with one or more peers. These activities are often described as either cooperative or collaborative. In practice, distinctions between the two blur. This became evident in the first author’s debriefing sessions with participants while field-testing the data collection instrument. All had worked on projects alone and with one or more classmates. Although their informal accounts included many features of collaborative learning, none were aware of the distinction made between it and cooperative learning in the literature. They mentioned that they had chosen workmates and had co-constructed projects and assessment criteria with their teacher. Members of their group worked without differentiated roles or responsibilities (Gillespie & Richardson, 2011) and occasionally collaborated with other groups. Power and responsibility for learning were redistributed from the teacher to students (Abrami et al., 1995). These experiences are more often associated with collaborative than cooperative learning, therefore the word “collaborative” will be used to refer to activities in which students work with another or others as it more accurately captures the meaning it had for students.

Subject

Surprisingly few studies have compared students’ preferences in different school subjects. This is likely because students’ preferences had been believed to be stable “styles” that were consistent across content areas. Pyryt’s (1991) survey data revealed no differences in gifted students’ preferences across math, English, social studies, and science. However, Li and Adamson (1992) found differences related to subject and sex. In math, their gifted secondary students preferred to learn alone. And in science, boys preferred individual work more than girls while girls preferred it more in English. More recently, Cowan (2014) reported 75% of his high school students preferred group work in social studies so they could share the workload, be more creative and social, and have more fun. Lou et al. (1996) found the effects of within-class grouping (vs. no grouping) on academic performance were positive in all school subjects; however, they were greater in math and science than in reading, language arts, and other courses. As mentioned above, they suggested this finding may have reflected differences in the types and difficulty of the tasks students associated with each subject rather than the subject. In other words, the tasks in math and science were considered more challenging or better suited to groups than those offered in reading and language arts.

Nature of the task

The structure and features of a task also affect what is learned and how learning occurs (Cohen, 1994; Walker & Shore, 2015). Relatively simple, factual, well-defined tasks involving specific procedures have been found to be more appropriate for individuals than groups while complex, ill-defined, difficult activities requiring a variety of roles, skills, and knowledge are better for dyads or groups (Fuchs, Fuchs, Hamlett, & Karns, 1998; Gillies, 2014; Lou et al., 2001). “A group task is a task that requires resources (information, knowledge, heuristic problem-solving strategies, materials, and skills) that no single individual possesses so that no single individual is likely to solve the problem or accomplish the task objective without at least some input from others” (Cohen, 1994, p. 8). For example, Diezmann and Watters’ (2001) found increasing task difficulty played an essential role in gifted students’ (11–12 years old) collaborations with peers when solving a series of challenging math problems. “If a task is sufficiently difficult, students tend to seek interaction with someone who can provide either cognitive or affective support” (Diezmann & Watters, p. 26). Additional evidence can be found in Sears and Reagin’s (2013) comparison of individuals and dyads solving problems in “traditional” (mixed ability) and “accelerated” (high ability) math classes. Pairs performed better than individuals in the traditional classes while the opposite was true in accelerated classes. “In other words, for students who were able to solve the problem successfully alone, collaboration was more of a hindrance than a benefit to performance” (p. 1167). This may reflect differences their expectations regarding their effort in group work. Cera Guy, Williams, and Shore (2019) found that while high-achieving students expected to work as hard alone as they did in a group while others expected to work less in a group.

Social interactions

Interpersonal dynamics generate most of the challenges and benefits students associate with group work. Their concerns regarding group composition, fair distribution of workload (Orbell & Dawes, 1981; Robinson, 1990; Salomon & Globerson, 1989), and status differentials (Cohen, 1994) often underlie their desire to work alone. When these concerns are not resolved, they can result in a sense of injustice, hurt feelings, and conflicts that erode motivation, learning, and relationships. Orbell and Dawes (1981) felt these “social dilemmas” (p. 39) were often the result of a lack of individual accountability within groups, such as all members of a group receiving the same grade regardless of the nature or extent of their contribution (Neber, Finsterwald, & Urban, 2001). Students have reported frustration associated with social loafing (when a peer exerts less effort in a group than they would if they worked alone), the free-rider effect (when a peer relies on others to do her or his share of the work [Salomon & Globerson, 1989]), the sucker effect (when the group expects the most capable student to do most of the work [Orbell & Dawes, 1981]), status differentials (“when higher status members dominate group activity” [Salomon & Globerson, 1989, p. 95]), and ganging-up effects (when “the whole group adopts a work-avoidance tendency, spending as little effort as possible” [Neber et al., 2001, p. 201]). Although some enjoyed peer tutoring (Ristow, Edeburn, & Ristow, 1985), others resented being treated as a “junior teacher” (Coleman, Gallagher, & Nelson, 1993; Robinson, 1990, 2003). Gifted students’ concerns about group work were greatest in mixed-ability (heterogeneous) groups (e.g., Clinkenbeard, 1991; French et al., 2011; Neber et al., 2001; Robinson, 1990, 1991; Schmitt & Goebel, 2015; Willard-Holt, Weber, Morrison, & Horgan, 2013), however, “When members contributed equally, students preferred working in groups to working alone” (Samardzija & Peterson, 2015, p. 245). Differences mentioned earlier in expectations students have for the effort involved in individual and group work are likely to contribute to high-achievers’ frustrations when others expect to work less in groups than they do (Cera Guy et al., 2019)

Same- or mixed-ability groups

Grouping students with others of similar ability, in homogeneous groups, or heterogeneously, in mixed-ability groups, may be the most contentious aspect of group work. Fuchs et al.’s (1998) investigation of same- versus mixed-ability dyads working on complex, challenging tasks found “homogeneous pairs worked more collaboratively and with greater cognitive conflict and resolution, focus on interacting, and helpfulness and cooperation” (p. 247); however “heterogeneous groups may be used appropriately with less complex tasks” (p. 251). Other researchers were concerned that in mixed-ability classes, homogeneously grouping students may enhance the efficiency of high ability students’ learning at the expense of lower ability peers’ because high ability students were not there to support their growth (e.g., Hooper & Hannafin, 1991).

Recent studies have found high ability students, and often their peers, preferred group to individual work when they were able to work with others of similar ability (Schmitt & Goebel, 2015; Walker & Shore, 2015) or those who learned at the same pace (Kanevsky, 2011). Neber et al.’s (2001) meta-analysis of research on ability grouping gifted students during cooperative learning supports these findings. They concluded “high achievers’ performances improve if they learn in homogeneous groups with other high-achieving students. Lower performances result if these students either learn individually or together with lower achieving students in heterogeneous groups” (p. 210, emphasis in original). “Often when gifted students are faced with working in a mixed ability group of agemates, they tend to express a desire to work alone” (Rayneri, Gerber, & Wiley, 2006, p. 115).

Walker and Shore (2015) explored interactions among ability, task characteristics, and preferences via a questionnaire and interviews. When asked if the impact a task had on their grade would influence their desire to collaborate, students in high performing classes, more than those in mixed-ability classes, wanted to work alone on “high stakes” assignments, such as projects. This result, as well as others from their study, provide evidence of the complex relationships among preferences, characteristics of learners and the contexts in which they learn.

Working with friends and choosing workmates

The opportunity to work with friends has been one of the reasons students prefer and enjoy group work more than working alone (Cowan, 2014; Fisher & Frey, 2012). Myers (2012) found “students who self-select their classroom work group members do report higher levels of commitment, trust, and relational satisfaction, as well as more affective learning and more cognitive learning, than students who are randomly assigned to classroom work groups” (p. 50). Collaborating was more attractive when students were able to work with a friend or choose their partner or members of their group (French & Shore, 2009; Kanevsky, 2011, 2015; Walker & Shore, 2015). When choosing their workmates, Kanevsky (2011) found more than 83% of her sample (Grades 3 to 8, gifted and nongifted) wanted to work with a partner or in a group. More than half disliked being assigned to dyads or groups by their teacher. Similarly, most of the twice-exceptional learners in Willard-Holt et al.’s (2013) study also wanted to choose their workmates.

It is intriguing that students became less enamored with choosing their groupmates over time in high school science classes (Mitchell, Reilly, Bramwell, Solnosky, & Lilly, 2004). Students felt obligated to choose friends in group tasks. They reported spending more time socializing than working in student-selected groups and felt the social consequences could be detrimental beyond the science class. Many did not want to take responsibility for their choices if the group did not work well and came to appreciate their teacher’s group-building expertise. The short- and long-term academic, social, and emotional consequences of choosing group members and working with friends on learning and behavior are not clear at this point and in need of further studies (Hanham & McCormick, 2018).

Group size

Studies of students’ preferred group size have also yielded inconsistent results however they favor small groups. Riding and Read (1996) found students preferred dyads and groups to working alone, as did Kanevsky (2011), however Kanevsky’s participants’ preferences were dependent upon having a choice of workmates. Working with a partner or in a small group were also favored by the nine to 16 year-old high- and otherachieving students (Cera Guy et al., 2019).

Most studies comparing individual students’ academic, affective and social outcomes when learning in groups of different sizes have concluded groups have better outcomes than individuals, and smaller groups (e.g., two to four students) are better than large groups (e.g., five or more students) most of the time (Wilkinson & Fung, 2002). Students working in pairs learned more than groups or individuals (e.g., Lou et al., 2001; Wilkinson & Fung, 2002) and developed higher levels of social self-esteem (e.g., Bertucci, Conte, Johnson, & Johnson, 2010). Both were superior to individual learning. Similarly, Fuchs et al. (2000) found pairs of 3rd and 4th graders solving complex math problems “earned higher scores than small groups on participation, helpfulness, cooperation, quality of talk, and PA [performance assessment] work” (p. 183). It appeared that students not only liked but benefited most from working with up to three peers.

Learner characteristics

Inconsistent results have also emerged from efforts to associate preferences with learner characteristics such as ability, sex, and age or grade.

Academic or cognitive ability

For many years, textbooks on the education of gifted individuals portrayed them as “loners” who preferred to learn independently and indicated this distinguished them from their nongifted peers. Research has not provided consistent support for this claim. Although some found they preferred independent study (e.g., Boultinghouse, 1984; Chan, 2001; Li & Bourque, 1987; Ricca, 1984; Ristow et al., 1985; Stewart, 1981) and learning alone (Ewing & Yong, 1993; Griggs & Dunn, 1984; Griggs & Price, 1980a, 1980b; Li & Adamson, 1992; Pyryt et al., 1998), others did not (e.g., Burns et al., 1998; Dunn & Price, 1980; Ewing & Yong, 1992; Kanevsky, 2011, 2015; Rayneri et al., 2006; Walker & Shore, 2015). Based on the findings of their survey data from 247 gifted, high achieving and nonidentified students in Grades 4 through 12, French et al. (2011) concluded “Some gifted students prefer to work alone some of the time” (p. 155) and this desire grows stronger when the social conditions work against them. “Gifted students who felt that their work was appreciated by teachers and fellow students reported the strongest preference to work with others” (p. 145).

These conditional preferences for individual and group work were also evident in Kanevsky’s (2011) findings. The 416 gifted and nongifted students’ (Grades 3–8) responses to a survey of their preferences for different types of differentiation shed further light on ability-related differences. With respect to working alone or with others, she found only two conditions of 18 related to collaboration distinguished the preferences of individuals who had been identified as gifted from those who had not. A larger majority of gifted students wanted to work with others some of the time and alone at other times and a smaller proportion disliked sitting alone. The reasons for keeping their options open became apparent in related results. A majority of all participants wanted to work with others when they were able to work with their choice of workmates, when their workmates worked at the same pace as they did, and when they sat in clusters. They did not want to be taught by peers or assigned to groups by their teacher. Similarly, students in Walker and Shore’s (2015) study also preferred group work to working alone when it was with “the right” workmates. More than their peers, high performing students wanted to work with others when it would have little impact on their grades but alone on those that did.

Age or grade

The findings of studies investigating age-related trends in preferences have also varied. Some found the desire to work alone increased with age or grade (e.g., French et al., 2011; Kanevsky, 2015) while others did not (e.g., Pyryt et al., 1998). Methodological differences (e.g., instrumentation, ranges of ages and ability of participants) make it difficult to distill any reliable patterns or insights from their findings.

Sex

Sex may play a role in learners’ preferences however, again, the evidence is not clear. Some researchers found boys preferred learning with peers more than girls (French et al., 2011; Pyryt et al., 1998; Ramsay & Richards, 1997), others the opposite (Johnson & Engelhard, 1992; Li & Adamson, 1992), or they found no difference (Kanevsky, 2015). French et al. suggested sex was likely to interact with other learner characteristics, such as ability, and may also be influenced by social roles. They found gifted girls preferred to learn alone more than nongifted boys and thought this might be due to being gifted rather than being female. These girls also enjoyed teaching peers more than gifted boys. Relationships between sex and other personal characteristics with learning preferences also deserve further study.

Theoretical value of learning preferences

Most studies of students’ preferences for working alone or with others have either had no theoretical orientation or claim social-constructivism frames the work (e.g., French et al., 2011). Self-determination theory offers a complementary perspective on the relationship between learning and preferences that focuses on student motivation. It accounts for the effect of accommodating students’ preferences on their engagement and motivation to learn (Jang et al., 2016). In order for students’ motivation to flourish, their basic needs for autonomy, competence, and sense of belonging must be met (Reeve, Deci, & Ryan, 2004; Ryan & Deci, 2000). Engagement can be increased by facilitating autonomy support, students’ feelings of competence, and relationships among class members. Learning in ways they prefer has been found to have these benefits as well as enhancing conceptual outcomes (Jang et al., 2016). Providing students with opportunities to choose or control organizational, procedural, or cognitive aspects of their learning can also achieve these goals (Patall, Cooper, & Robinson, 2008; Reeve, Jang, Carrell, Jeon, & Barch, 2004; Stefanou, Perencevich, DiCintio, & Turner, 2004). Allowing them to choose their workmates is an example of organizational autonomy support (Stefanou et al., 2004). Of course, enhancing engagement is not that simple. Jang, Reeve, and Deci (2010) found that autonomy support and structure provided by the teacher (e.g., clear directions and guidance, keeping students on task, managing behavior) both contributed to student’s observed engagement, however “only autonomy support was a unique predictor of students’ self-reported engagement” (p. 588).

When grouping students and creating activities for those groups, educators expect students to collaborate and share their resources so their group’s work will reflect the participation and contributions of all group members. They also expect students’ relationships and interpersonal competence will be enhanced by opportunities to explore each other’s “reasoning and viewpoints in order to construct a shared understanding of the task, … propose and defend their own ideas, and to ask their peers to clarify and justify any ideas they do not understand” (Goos, Galbraith, & Renshaw, 2002, pp. 196–197). By the time North American students reach middle school (10–14 years old) most have had many experiences with group work, some in which those outcomes were actualized and others in which they were not. For example, Chichekian and Shore (2017) found academically high performing adolescents appreciated others who, like themselves, “held firm” to their perspectives when they differed from peers’. As a result, they developed evidence-based preferences for a variety of conditions associated with individual and group work.

In sum, studies investigating conditions influencing students’ eagerness to work alone and with others indicate their preferences likely involve interactions among characteristics of the learner and context. We sought to improve understandings of the context-specific interactions that influence students’ desire to work on projects alone and with others on a challenging project, a complex, popular task that can be assigned to individuals, partners or groups. Although studies using academic performance or program placement have often been used to distinguish high achieving or high ability learners from their peers, we used scores on a standardized measure of cognitive reasoning ability to provide greater precision than other means of operationally defining ability.

Research questions

1. Do middle school students generally prefer to work alone or with others on projects, or are their preferences context-specific?

2. Do students’ desires to work alone or with others on projects vary across school subjects? And do these subject-specific preferences vary with students’ cognitive ability, grade, and sex?

3. What contextual considerations underlie middle school students’ desire to work alone or with others on projects? In addition, do cognitive ability, grade, and sex contribute to students’ preferences?

Method

Participants

A total of 328 students enrolled in three middle schools (Grades 6 to 8) in one suburban Canadian school district participated in the study (191 girls, 137 boys; 130 in Grade 6, 105 in Grade 7, and 93 in Grade 8). The sample consisted of 74% individuals who self-identified as Anglophone Canadian, 15% as Asian, 7% as European, 2% as Middle Eastern, Pilipino, or Jewish, and 2% did not respond. Two cognitive ability groups were formed using students’ composite scores on the Cognitive Abilities Test, Form 7 (CogAT7; Lohman, 2011) and the 90th percentile threshold for giftedness proposed by Gagné (2017). The 62 students who scored at or above the 90th percentile were included in the cognitively highly able (CHA) group. The 263 who scored at the 89th percentile or below were included in the general comparison (GC) group. Table 1 provides a summary of the number and percentage of participants by cognitive ability, grade, and sex.

Table 1. Summary of participants by sex, grade, and cognitive ability

Research participants (N = 328)		n	Percent
Cognitive ability	CHA	62	19.1
	GC	263	80.9
	Total	325^a
Grade	Grade 6	130	39.6
	Grade 7	105	32.0
	Grade 8	93	28.4
	Total	328
Sex	Boys	137	41.8
	Girls	191	58.2
	Total	328

^a3 missing values.

In this school district, most middle school students had one teacher for their core academic subjects (mathematics, language arts, science, social studies and often for French and physical education as well). Students also rotate through an additional block of elective studies with different teachers three or four times in a year. These electives include drama, digital literacy, music, art, and other options.

Pedagogies involving collaborative learning were strongly encouraged by the school district however the nature and extent of their implementation varied. Many students in the focus and field test groups reported they had worked with classmates on many of their assignments in recent years while others in the same school told us this was only occasionally the case in their classes. All had extensive experience with projects, challenging or not, from Kindergarten until the data were collected.

Instruments

The Cognitive Abilities Test, Form 7 (CogAT7; Lohman, 2011) was used to assess academic reasoning ability and a Project Context Survey (PCS) which was developed to assess the influence of variety of contextual considerations (features of the context) on a student’s desire to work alone or with others on a challenging project.

CogAT7

The CogAT7 (Lohman, 2011) was used to provide an individualized, standardized assessment of ability. It is a group-administered assessment of learned “reasoning abilities in the three symbol systems most closely related to success in school: verbal reasoning, quantitative reasoning, and nonverbal reasoning” (Lohman, 2012a, p. 1). Its nine scales are clustered in three batteries: Verbal (Verbal Analogies, Sentence Completion, and Verbal Classification), Quantitative (Number Analogies, Number Puzzles, and Number Series), and Nonverbal (Figure Matrices, Paper Folding, and Figure Classification). Composite scores were used to form the two ability groups. Reliability coefficients for composite scores computed using the part-test method were .96 for Grades 5 to 8, indicating this measure has high internal consistency (Lohman, 2012b). The .76 correlation between CogAT7 composite scores and Full Scale WISC IV IQ scores provides evidence of its concurrent validity (Lohman, 2012b).

Project context survey (PCS)

The PCS consisted of two subscales: Working Alone (PCS-A, eight unique items specifically related to working alone) and Working with Others (PCS-O, 19 unique items specifically related to working with others). In additional to the unique items, each subscale included five parallel, subject-specific items. They assessed students’ desire to work alone (PCS-A) or with others (PCS-O) on projects in five school subjects: art, language arts, mathematics, social studies, and science. Because students’ preferences were not presumed to be dichotomous, the matched (but independent) items on the two subscales enabled students to express preferences for working with others or alone on projects in each subject that were not dichotomous. Following the subscales, students responded to a final multiple-choice item asking if they had a general preference for working alone or with others on projects.

The survey began with demographic questions (sex, grade, home language, etc.), followed by the PCS-A subscale, the PCS-O subscale, and ended with the general preference item. Within the PCS-A and PCS-O subscales, the five subject-specific items were answered first, then the eight and 19 unique items respectively. Other than the final item, all items were formatted on a Likert-type 5-point scale: “Strongly Disagree,” “Disagree,” “Neutral,” “Agree,” and “Strongly Agree.” Items on the PCS-A began with the stem “I prefer to work on a challenging project on my own,” followed by the item content, such as “when I feel my grade will be higher than if I work with others.” Similarly, items on the PCS-O began with the stem “I prefer to work with others on a challenging project when,” followed by the item content, such as “when all group members’ ideas are included to create the final product.” The five subject-focused items followed the same response format, e.g., “I prefer to work on a challenging project on my own when … it is in Art” on the PCS-A, and “I prefer to work with others on a challenging project when … it is in Art” on the PCS-O. Again, students indicated the strength of their agreement with each item.

Development of the PCS

An initial draft of the PCS was based on a review of research investigating conditions influencing students’ preferences for individual and group work. Two focus groups of students in Grades 6 to 8 were recruited to complete the draft survey items, provide feedback on their relevance and clarity, and suggest additional items. Revisions were based on input from the focus group, a research assistant, and a subject matter expert (SME) in gifted education. This process was repeated with the revised instrument and the same students. Another round of revisions was made before field-testing the survey in two mixed-ability classes.

The data collection instrument included the demographic items, the two subscales (PCS-A had 15 unique items and PCS-O had 40), and the general preference item. The number of items on the survey was approximately three times the number retained. This is consistent with Standards set by the American Educational Research Association, the American Psychological Association, and the National Council on Measurement in Education (2014). Two other SMEs, one in gifted education and the other in measurement, conducted item analyses after the data was collected. Ambiguous items and those with intercorrelations greater than .90 were excluded. As a result, eight and 19 unique items from the PCS-A and PCS-O subscales respectively were retained and included in the principal component analyses.

Procedure

Participants completed the PCS and the grade-appropriate level of the CogAT7 in three or more sessions led by the first and third authors. Most sessions took place in participants’ classrooms. Some were held in the school’s cafeteria or library when students were drawn from two or more classrooms at the same time. Sessions varied in duration from 25–60 minutes depending on students’ availability.

Results

General or context-specific project preferences for working alone or with others

Participants indicated their general preference for working alone or collaborating on challenging projects in one item by choosing one of five options: “always alone,” “always with a partner,” “always in a group of three or more,” “sometimes alone and sometimes with others,” and “don’t care.” A significant majority, 69% (n = 210) indicated their preferences varied by choosing the flexible option, “sometimes alone and sometimes with others.” In contrast, 20% (n = 61) always wanted to work with others on projects and only 7% (n = 21) always wanted to work on them alone. The remaining 5% (n = 15) reported no preference. The large proportion selecting the conditional “sometimes” option indicated most students’ preference for working alone or with others on a challenging project was sensitive to the context.

Subject-specific project preferences and their interactions with cognitive ability, grade, and sex

As mentioned earlier, items on each subscale asked students to rate the extent to which they wanted to learn alone (on the PCS-A) or with others (on the PCS-O) in five school subjects (art, language arts, mathematics, science and social studies). These items were rated separately on each subscale so participants were able to like or dislike both working alone and working with others on a project in each subject. Bivariate correlations were computed between all participants’ ratings for each school subject on the two subscales to see if their desire to work alone or collaboratively on projects was stronger in some subjects than others. They were −.65. −.55, −.58, −.50, and −.55, respectively. All were statistically significant (p < .001) and were considered large effect sizes. The negative direction of these correlations indicated these students preferred either working alone or collaborating on projects in all five subjects but not which one. The next set of analyses were undertaken to find out.

Subject preference scores (SPS; the difference between a student’s rating for each subject on the PCS-A and PCS-O) indicated whether they preferred working alone or with others on projects in a particular school subject. The range for each response was −2 to +2. For example, if a student strongly agreed (+2) with working alone on projects in math and strongly disagreed with collaborating on them (−2), their SPS would be +4 (2 – [−2] = 4). Another student might agree with both working alone and with others on math projects which would result in a SPS of zero (1–1 = 0), indicating no subject-specific project preference in math. Using this formula, the SPS could range from +4 (strong desire to work alone in a specific school subject) to −4 (strong desire to work with others in that subject). SPS were used in this initial analysis of subject-specific preferences, and later in the ability-, grade-, and sex-related comparisons.

Single-sample t-tests were conducted to see if the sample’s mean SPS for each school subject was statistically different from zero. Table 2 summarizes those results as well as the results of pairwise comparisons undertaken to investigate interactions between cognitive ability, grade, or sex and students’ preference in each school subject. An overall alpha level of .05 was considered for these analyses however it would have been reduced to .003 for each comparison after the Bonferroni adjustment so we accepted Ranstam’s (2016) recommendation and used an alpha of .01, acknowledging the risk of a false positive result. Exact alphas appear in Table 2 for readers’ consideration.

Table 2. Summary of group comparisons of subject-related project preferences

		Art	Language Arts	Math	Science	Social Studies
Overall	Mean (SD)	1.33 (2.15)	0.05 (1.89)	−.011 (2.16)	−0.82 (1.88)	−0.79 (1.87)
	Significance^a	p < .001^e	p = .55	p = .45	p < .001^e	p < .001^e
	(Cohen’s d)	0.62			0.44	0.42
Cognitive ability Mdn (IQR)	CHA	1.02 (1.90)	0.29 (1.86)	0.58 (2.06)	−0.66 (1.89)	−0.79 (2.04)
	GC	1.41 (2.21)	−0.01 (1.90)	−0.27 (2.17)	−0.85 (1.87)	−0.79 (1.88)
	Significance^b	p = .09	p = .39	p = .006^e	p = .73	p = .81
	Hedges’ g			0.17
Grade Mean (SD)	Grade 6	1.26 (2.16)	−0.12 (2.00)	−.040 (2.33)	−0.91 (1.96)	−0.76 (2.03)
	Grade 7	1.44 (2.02)	0.03 (1.73)	0.11 (2.07)	−1.10 (1.80)	−1.01 (1.80)
	Grade 8	1.37 (2.29)	0.35 (1.87)	0.10 (1.97)	−0.35 (1.76)	−0.53 (1.71)
	Significance^c	p = .80	p = .19	p = .13	p = .009^e	p = .22
	Partial η^2 (η^2p)				0.03
Sex Mean (SD)	Boys	0.98 (2.02)	0.26 (1.85)	0.33 (2.11)	−0.70 (1.96)	−0.91 (1.89)
	Girls	1.61 (2.28)	−0.07 (1.93)	−0.39 (2.18)	−0.89 (1.75)	−0.68 (1.86)
	Significance^d	p = .009^e	p = .12	p = .004^e	p = .35	p = .37
	Hedges’ g	0.29		0.33

The numbers in parentheses are standard deviations (SD), except for “Cognitive Ability,” where interquartile ranges (IQR) are provided instead.

^aSingle sample t-tests

^bMann-Whitney test

^cANOVAs

^dindependent sample t-tests

^ep < .01

Overall project preferences related to each school subject

The single-sample t-tests of SPS revealed that students preferred to work alone on a challenging project in art (M = 1.33, t[327] = −11.34, p < .001, Cohen’s d = 0.62), and with others in science (M = −0.82, t[327] = 7.80, p < .001, Cohen’s d = 0.44) and social studies (M = −.079, t[327] = 7.47, p < .001, Cohen’s d = 0.42). A moderate effect size was observed in art while small effect sizes were found in science and social studies (Cohen, 1992). No statistically significant differences were found in language arts and math which indicated the students’ preferences were diverse in those subjects.

Subject-specific project preferences and cognitive ability

To address the unequal group sizes and skew introduced by the 90^th percentile boundary between the ability groups, Mann-Whitney tests (nonparametric t-tests) were used to compare the median SPS of the CHA and GC groups. This revealed a statistically significant difference with a small effect size in math (CHA Mdn = 0.5, GC Mdn = 0, U = 9951, p = .006, Hedges’ g = 0.17) that showed students with high cognitive ability had a slight preference for working alone while students in the comparison group had no preference. No ability-related differences in project preferences were found in the other subjects.

Subject-specific project preferences and grade

One-way ANOVAs comparing the SPS of students in Grades 6, 7, and 8 resulted in a statistically significant age difference with a small effect size in science (F[2,325] = 4.73, p = .009, η²_p = .03). Only one of the post hoc comparisons of the means using the Tukey HSD test achieved significance. It suggested that in science, Grade 8 students (M = −0.35, SD = 1.97) were less inclined to collaborate on projects than the Grade 7 students (M = −1.10, SD = 2.07).

Subject-specific project preferences and sex

Independent sample t-tests of SPS revealed statistically significant differences between boys and girls in two subjects: art and math. In art, while both boys and girls preferred to work on projects alone, girls indicated a stronger inclination to do so (Boys’ Mean = .98, Girls’ Mean = 1.61, t[326] = −2.62, p = .009). In math, boys indicated a slight preference for individual projects whereas girls indicated a slight preference for working with others (Boys’ Mean = .33, Girls’ Mean = −0.39, t[326] = 2.92, p = .004). Girls’ and boys’ preferences did not differ the other three academic subjects.

Contextual considerations underlying students’ project preferences

Separate principal component analyses with Varimax rotation were conducted on 328 students’ ratings for the conditions described in the eight and 19 unique items on the PCS-A and PCS-O subscales, respectively. They reduced the number of items to a few distinct dimensions within each subscale. These extracted dimensions represent themes that were interpreted as contextual considerations that influenced students’ preference for working individually or collaboratively on a challenging project within the parameters of this survey. Bartlett’s test of sphericity indicated satisfactory interdependence among the items in the principal component analysis of the PCS-A (χ2 [28] = 426.61, p < .001) and PCS-O (χ2 [171] = 1264.65, p < .001) subscales. The Kaiser-Meyer-Olkin measure of sampling adequacy was .73 for PCS-A and .79 for PCS-O, which were sufficient for principal component analyses.

The number of principal components to be included in each subscale was determined in two ways. The content of the items was first qualitatively reviewed and overlapping items removed to optimize construct representation. Second, the R package paran (Dinno, 2012) was used to perform Horn (1965) parallel analysis for each subscale. It recommended three principal components be identified for the PCS-A and five principal components for the PCS-O. After ensuring that all eigenvalues were greater than 1 and loadings were greater than .40, three principal components were identified for the PCS-A subscale (see Table 3) and five for the PCS-O (see Table 4).

Table 3. Principal component analysis summary and themes for projects alone (PCS-A)

Item	Component loadings			Communality
	1	2	3
Theme 1: Enjoyment (3 items, α = .59)
1. When I know the teacher will help me if I get stuck	.79	−.04	−.04	.63
2. When it is about a subject I really enjoy	.71	.11	0.31	.51
3. When the final product is something I really enjoy developing	.66	.19	.14	.77
Theme 2: Optimizing the Outcome (2 items, α = .73)
4. When I feel I can do the project better or faster than if I work with others	.11	.87	.08	.77
5. When I feel my grade will be higher than if I work with others	.07	.85	.19	.52
Theme 3: Risk Management (3 items, α = .56)
6. When it involves an essay or a writing project	.07	−.01	.79	.49
7. When I think there will be conflict with my partner or among the members of the group	.07	.14	.70	.63
8. So I can be sure my grade will not be hurt by group members who do not do their share of the work	.28	.31	.60	.61
Eigenvalue	2.66	1.22	1.02
% Variance explained	20.82	20.44	20.33
% Total variance explained	61.48

Table 4. Principal component analysis summary and themes for projects with others (PCS-O)

Item	Component loadings					Comm-unality
	1	2	3	4	5
Theme 1: Inclusiveness (6 items, α = .75)
1. All group members’ ideas are included to create the final product	.69	.06	−.05	−.01	−.03	.53
2. Everyone in the group trusts me	.67	.10	−.05	.17	.15	.57
3. I trust the others I am working with	.64	.02	.38	.08	−.09	.59
4. I feel comfortable sharing my ideas	.63	.05	.03	.13	.09	.57
5. I feel my ideas are understood by my partner or group members	.57	−.05	.32	.16	−.08	.70
6. There is an atmosphere of trust among all members of the class	.55	−.13	.33	.18	.02	.49
Theme 2: Strengths of Others (3 items, α = .61)
7. I can rely on the others in my group to do the work better than I could	−.07	.73	.16	.06	−.08	.60
8. My partner or the members of my group are smarter than me	−.02	.73	.15	.07	.12	.57
9. I need others to keep me focused	.24	.50	.30	−.20	−.34	.51
Theme 3: Support (3 items, α = .58)
10. I know I will need help in order to complete the project. I know I can’t do it alone	.10	.19	.73	.11	.05	.57
11. I feel the project is going to be very challenging for me	.09	.19	.67	.01	.05	.66
12. Others in my group know something that will help me get a better grade than I would if I work on my own	.36	.34	.40	.00	.23	.34
Theme 4: Familiarity (4 items, α = .61)
13. I can choose my partner or the members of my group	.20	−.01	.05	.73	.01	.49
14. My partner or the members of my groups are my friends	.25	.47	−.11	.64	.02	.46
15. I work with a partner, not in a group	−.05	−.23	.35	.59	.03	.46
16. I know my partner or the members of my group well	.37	.35	.00	.55	−.04	.57
Theme 5: Assessment (3 items, α = .38)
17. My teacher gives me a separate grade for my contribution to the project	.04	.03	.21	.05	.79	.55
18. All members of the group receive the same grade no matter how much work they do^a	−.17	0.3	−.05	−.02	.65	.42
19. I can give myself a grade for my contribution to the project	−.27	.21	−.03	−.04	.48	.46
Eigenvalue	4.20	1.90	1.48	1.37	1.13
% Variance explained	15.29	11.20	9.54	9.29	7.80
% Total variance explained	53.11

^aItem 18 was reverse scored

Contextual considerations influencing students’ desire to work alone

As shown in Table 3, the three themes related to working alone on a difficult project were: (a) Enjoyment (the pleasure they associated with engaging in the topic and creating a product they enjoy while having the support of their teacher when it becomes difficult), (b) Optimizing the Outcome (working better or faster, and receiving a higher grade), and (c) and Risk Management (the desire to reduce the perceived risks to their writing, relationships and grade by completing a difficult project alone). The three themes explained 21%, 20%, and 20% of the total variance, respectively, thus 61% of the total variance was explained by the model. Reliability estimates (Cronbach’s alphas) for the three themes were .59, .73, and .56, respectively, with an overall alpha of .71 for the PCS-A subscale. None of the items in any of these themes, when deleted, resulted in a higher alpha, suggesting a preliminary fit in the component in which they were included.

Contextual considerations influencing students’ desire to work with others

The five components influencing the desire to work with others are presented in Table 4: (a) Inclusiveness (wanting an atmosphere involving mutual trust in which ideas can be shared safely and understood), (b) access to the Strengths of Others (being able to rely on the contributions of workmates who may be smarter and will maintain their focus on the project), (c) the Support provided by peers when they need help completing a difficult project and want a better grade than they could earn alone, (d) Familiarity (working on a project with a partner they have chosen who is a friend or someone they know well), and (e) fair Assessment (receiving an individual grade for their contributions to the project and having the opportunity to self-assess it as well). The variances explained by each of the five themes were 15% by Inclusiveness, 11% by Strengths of Others, 10% by Support, 9% by Familiarity, and 8% by Assessment. They accounted for 53% of the total variance. Their reliability estimates (Cronbach’s alphas) were .75, .61, .58, .61 and .38, respectively, with an alpha of .73 for the PCS-O overall. Again, deleting items did not result in a higher alpha for any component, suggesting a preliminary fit between each item and the component in which it was included.

The alphas for both subscales were acceptable (i.e., greater than .70; Kline, 2013), however the alphas for four of the five components were not. They indicate a lack of internal consistency among the items within them. This is likely due to the small number of items (three or four) included in the four themes with alphas less than .70. Although they were not as strong as desired, moderate and low alphas were tolerated to avoid construct under-representation and because this project is the first to examine the instrument’s psychometric properties. Although only two items loaded on it, the “Optimizing the Outcome” theme found in the PCS-A analysis was retained for the same reasons.

Differences in contextual considerations related to cognitive ability, grade, and sex

A theme score (a standardized weighted average by component loading) was computed for each participant for each component. Interactions were investigated between these theme scores and cognitive ability (using Mann-Whitney U tests), grade (using one-way ANOVAs), and sex (using independent t-tests with equal variance assumed).

Cognitive ability

Statistically significant differences between theme scores of students in the CHA and GC ability groups were found in two of the three themes related to working alone on projects (see Table 5) and one of the five related to working with others (see Table 6). Individual projects were more appealing to students in the comparison group than those with high ability when they felt working on a project alone would be more enjoyable (CHA Mdn = −0.28, GC Mdn = 0.11, U = 9975, p = .006, Hedges’ g = 0.28). In contrast, working alone was more attractive to high ability students than students in the GC group (CHA Mdn = 0.47, GC Mdn = 0.016, U = 6639, p < .005, Hedges’ g = 0.31) when Optimizing the Outcome was important to them. This included the opportunity to produce a higher quality project that could be completed more quickly and result in a higher grade.

Table 5. Group comparisons of PCS-A themes scores related to working alone on a challenging project

PCS-A		Enjoyment	Optimizing the Outcome	Risk Management
Cognitive ability Mdn (IQR)	CHA	−0.28 (1.24)	0.47 (1.35)	0.39 (1.16)
	GC	0.11 (1.40)	0.016 (1.48)	0.083 (1.22)
	Significance^a	p = .006^e	p = .005^e	p = .073
	Hedges’ g	0.28	0.31
Grade Mean (SD)	Grade 6	−0.038 (1.04)	0.055 (1.03)	−0.063 (1.00)
	Grade 7	0.065 (0.97)	−0.043 (0.90)	−0.15 (1.03)
	Grade 8	−0.020 (0.96)	−0.029 (1.06)	0.25 (0.91)
	Significance^b	p = .72	p = .72	p = .013
	Partial η^2 (η^2p)			0.03
Sex Mean (SD)	Boys mean	0.090 (1.01)	−0.033 (1.03)	−0.049 (1.01)
	Girls mean	−0.064 (0.98)	0.024 (0.97)	0.035 (0.98)
	Significance^c	p = .17	p = .61	p = .45

The numbers in parentheses are standard deviations (SD), except for “Cognitive Ability,” where interquartile ranges (IQR) are shown instead.

^aSingle sample t-tests

^bMann-Whitney tests

^cANOVAs

^dindependent sample t-tests

^ep < .01

Table 6. Group comparisons of PCS-O themes scores related to working with others on a challenging project

PCS-O		Inclusiveness	Strengths of Others	Support	Familiarity	Assessment
Cognitive ability Mdn (IQR)	CHA	−0.025 (1.06)	−0.044 (1.25)	0.55 (1.06)	0.26 (1.35)	0.10 (1.17)
	GC	0.071 (1.20)	−0.026 (1.15)	−0.006 (1.18)	−0.024 (1.39)	0.031 (1.37)
	Significance^b	p = .35	p = .76	p < .001^e	p = .22	p = .86
	Hedges’ g			0.48
Grade Mean (SD)	Grade 6	0.11 (1.11)	−0.027 (1.07)	0.082 (1.01)	0.14 (0.98)	−0.13 (1.00)
	Grade 7	−0.006 (0.95)	0.002 (1.05)	−0.13 (0.96)	−0.056 (1.00)	−0.12 (0.92)
	Grade 8	−0.15 (0.85)	0.035 (0.81)	0.029 (1.00)	−0.13 (1.00)	0.31 (1.02)
	Significance^c	p = .17	p = .90	p = .26	p = .10	p = .002^e
	Partial η^2 (η^2p)					0.04
Sex Mean (SD)	Boys	0.040 (0.97)	0.030 (1.11)	−0.009 (1.00)	−0.055 (0.92)	0.008 (1.01)
	Girls	−0.029 (1.02)	−0.021 (0.91)	0.007 (0.99)	0.039 (1.05)	−0.006 (0.99)
	Significance^d	p = .54	p = .65	p = .89	p = .40	p = .90

The numbers in parentheses are standard deviations (SD), except for “Cognitive Ability,” where interquartile ranges (IQR) are shown instead.

^asingle sample t-tests

^bMann-Whitney tests

^cANOVAs

^dindependent sample t-tests

^ep < .01

When thinking about a collaborative project, the Support of others was more important to the CHA than GC when they expected it to be difficult and that they would need help to achieve the grade they wanted (CHA Mdn = 0.55, GC Mdn = −0.006, U = 5253, p < .001, Hedges’ g = 0.48). The CHA group’s greater pragmatism was also evident in the differences found in the PCS-A theme scores above for “Optimizing the Outcome.” It seems that the CHA group’s preferences for working alone and with others was significantly influenced by the likelihood they would need and benefit from other’s contributions, or if they felt they would do better on their own.

Grade

A statistically significant difference with a small effect size was found only among the mean theme scores of students in Grades 6, 7, and 8 for students’ concerns regarding Assessment when collaborating on difficult projects (PCS-O; F[2, 325] = 6.32, p < .002, η²_p = 0.04). Post hoc comparisons of the grade level means for Assessment showed the Grade 8 students’ mean (M = 0.31, SD = 1.02) was significantly higher than the means for students in Grade 6 (M = −0.12, SD = 1.00) and Grade 7 (M = −0.12, SD = 0.92). Receiving recognition for their individual contribution to a difficult group project and self-assessment appeared to be more important to students in Grade 8 than their younger schoolmates.

Sex

Independent t-tests revealed no statistically significant differences in the contextual considerations related to students’ sex. This tells us their influence on the preferences of girls and boys were similar.

Conclusions and discussion

The task-specific findings reported here provide additional evidence that the appeal of individual or shared projects depended on the school subject and a number of other contextual considerations. Some arose in the findings of earlier investigations; some were familiar but nuanced.

General or context-specific project preferences for working alone or with others

It was not surprising that a large majority of these middle school students preferred neither working alone nor with others on a challenging project. Their preferences were dynamic and varied in response to features of the context. Arguments and evidence supporting the context-sensitivity of learning preferences related to the desire to work individually or collaboratively have accumulated to the point that they can be considered trustworthy (Cantwell & Andrews, 2002; French & Shore, 2009; French et al., 2011; Kanevsky, 2011, 2015; Koutrouba et al., 2012; Samardzija & Peterson, 2015; Walker & Shore, 2015; Walker et al., 2011; Wismath & Orr, 2015). It appears students’ preferences result from weighing the risks and benefits associated with the context.

Subject-specific project preferences

Sometimes the school subject involved in the project was a consideration. Like Cowan (2014), we found group projects were popular in social studies. Our middle school students also found them attractive in science but not in art. Individual art projects were even more desirable to girls than boys. We suspect the opportunity to express ideas freely and creatively may be more important than in art than in other subjects and this resulted in the popularity of individual projects. Further, working alone is a way to avoid having to compromise or engage in uncomfortable conversations regarding differences of opinion. The appeal of collaborative projects in science differs from Li and Adamson’s (1992) whose high IQ (WISC-R IQ > 120) secondary students preferred to work individually in science. This may be due to differences in students’ experiences, age and ability, or all of the above.

In math, group differences in project preferences related to ability and sex emerged. Like the high IQ students in Li and Adamson (1992) study, the CHA students reported a greater desire to work on projects alone in math. This may reflect a greater concern regarding potential free-riders, social loafing, sucker effects, and status differentials that might arise when collaborating because they would be the most capable member of the group (Neber et al., 2001; Orbell & Dawes, 1981; Robinson, 2003; Salomon & Globerson, 1989). If working alone, their grade and the quality of their experience would not suffer if others did not understand the material or care about it or their grade as much as they did. This concern might also contribute to boys’ slight preference for individual projects in math. On the other hand, girls’ slight preference for collaborative math projects may reflect having lower confidence in their ability to complete a challenging math project without the support of others, or perhaps a desire to help others, or both. Socialization into sex-related roles may also have contributed to these results.

Contextual considerations underlying students’ project preferences

Overall, themes from the principal component analyses revealed interaction among features of the context. Collectively, they suggested students’ default “preference” was to work alone when they felt their ideas, feelings, relationships, or their grade might be at in peril. Some of their concerns (e.g., potential conflict, unfair assessment) have appeared consistently in 30 years of research on this topic and studies of cooperative and collaborative learning (e.g., Hooper & Hannafin, 1991; Neber et al., 2001; Orbell & Dawes, 1981; Salomon & Globerson, 1989). Unfortunately, their presence in these findings indicate they persist in students’ experiences and continue to influence their preferences.

Although students may feel working alone was socially and academically safer, they preferred collaborative projects when the conditionsfelt right. These conditions were: when they could work with a friend or familiar classmates, when they felt members of their group trusted each other and would include others’ ideas, when they could rely on them because they were smarter and would maintain the focus, when their contributions resulted in a better grade, and when the project’s assessment recognized the group’s accomplishments as well as individual contributions. The opportunity to choose workmates has consistently made collaboration attractive to students in this age group (Fisher & Frey, 2012; Kanevsky, 2011; Koutrouba et al., 2012; Walker & Shore, 2015), however secondary students became less enthusiastic as result of tensions between peer pressure and concern for their grade (Mitchell et al., 2004).

Our findings and others indicate high ability learners enjoyed working alone more than peers when the project or task is challenging, difficult, interesting and complex (Diezmann & Watters, 2001; Kanevsky, 1992; Walker & Shore, 2015). In contrast, all students, regardless of ability, ranked “fun” assignments as most preferred group work in Walker and Shore’s study. They attributed the difference in preferences on “difficult” or “difficult but interesting” and fun assignments to differences in the impact students perceived each type of task would have on their grade. Difficult tasks were expected to have greater consequences so students preferred to do them alone. This preference was believed to be stronger for high performers because their academic orientation was stronger (Schapiro, Schneider, Shore, Margison, & Udvari, 2009). Fun assignments were expected to have fewer consequences so working with others was more attractive because the “stakes” were lower. Students in our study likely felt a “challenging project” would also be a difficult assignment so these findings align with Walker and Shore’s but with an additional nuance. The enjoyment these high ability participants associated with working alone on a difficult project also involved having their teacher available to help if needed. Just knowing that assistance is available may provide a safety net that protects or enhances a greater sense of “flow” (Nakamura & Csikszentmihalyi, 2005) or accomplishment (Kanevsky, 1992) learners with high ability derive from challenges involving content and a topic they enjoy.

The preferences of the CHA students were also more sensitive to the pragmatism evident in their desire to “optimize the outcome” by working alone on a challenging project and also access the support provided by workmates when collaborating on a difficult project. Both of these findings may be related to gifted students’ stronger academic task-orientation (Schapiro et al., 2009) as mentioned above. French et al. (2011) also found the support of others enhanced the appeal of working with others more for high achieving and gifted learners than others. The findings of our studies complement each other but differ significantly in their meaning. In French et al., “support” meant the feeling “that people around you (for example, parents, teachers, or classmates) appreciate your work (think your work is valuable or important)” (p. 158) In our results, “support” represented the types of assistance and expertise provided by collaborators. It made working with others more attractive to CHA students’ but not GCs. Differences in the support sought or valued by students who differ in academic ability deserves further investigation.

We found concerns regarding the assessment of joint projects increased with students’ age as they have elsewhere (French et al., 2011; Kanevsky, 2015). Older students were more interested in collaborating if they assessed their own work and received a group grade as well as a separate grade from a teacher for their contribution. It may be that negative experiences with group dynamics and grading accumulate over time so older students are more skeptical of group work than younger.

Implications

Theoretical

Although the items on the PCS were not based on self-determination theory, the contextual considerations emerging from the data resonated with the basic psychological needs to feel competent, autonomous, and connected to others identified in it (Reeve et al., 2004 ; Ryan & Deci, 2000). Individual projects were preferred when students felt their potential workmates would be less competent, when the lack of individual accountability in a group project would result in a low grade that did not accurately reflect their individual efforts and competence, when relationships were weak (they lacked trust and acceptance), unreliable, or uncomfortable (they expected disagreements), or when they wanted to protect their autonomy. In contrast, working with others was attractive and enhanced their competence (and grade) when they knew their workmates well and felt safe with them, when they felt they were trustworthy and reliable, and when they were able to exercise their autonomy (e.g., they could choose workmates and felt they had a voice in the project’s assessment). This alignment with self-determination theory suggests it holds promise for explaining and investigating the benefits and challenges associated with the development of intrinsic motivation by offering students opportunities to find and enact their project preferences. In addition, the repertoire of evidence-based practices self-determination theory has inspired may assist researchers and educators efforts to understand and address students’ preferences.

Educational

The appeal of an individual or collaborative assignment depends on students’ thoughts and feelings about dimensions of the context, i.e., the subject, task, social dynamics, and setting). Educators may find these highlights from our findings helpful when planning for project work.

Working alone on a challenging project may be preferred:

• In art by most students and by high ability learners in math,

• When they expect working alone on a project will be more enjoyable, result in a better experience and outcome, and is a way to manage the risks associated with collaborating,

• By high ability learners more than others because enjoying and optimizing the project are more important to them.

Working with others on a challenging project may be preferred:

• In social studies and science,

• When students feel included and trusted, when they will need others knowledge and support, when they know or choose their workmate(s) and their work is assessed fairly,

• By high ability learners more than others in the class when they feel they will need the knowledge and support provided by collaborators to complete the project to their satisfaction.

In light of the fluidity of students’ preferences, we highly recommend teachers consult their students to accurately determine their preferences in their contexts.

The persistence and prevalence of students’ concerns regarding interpersonal dynamics and individual accountability in collaborative projects indicate they are still problematic (Clark, 2017; Cohen, 1994; Wismath & Orr, 2015). Weak learning communities and relying solely on a “group grade” diminishes academic benefits, social relationships, and intrinsic motivation. The development of the safe learning environment that is prerequisite to cooperative and collaborative learning involves preparatory training and community-building activities in which students become familiar with and trusting each other (see Abrami et al., 1995; Allen, 2012; Gillies, 2007; Johnson & Johnson, 1998; Kagan & Kagan, 2009). Professional development for educators is needed to prepare them to develop those activities and assignments that are most appropriate for individual and group work.

Future research

Many questions remain regarding students’ preference for working alone and with others. Given their sensitivity to context, research needs to be undertaken in the contexts in which learners’ preferences develop. Intentionally varying the four aspects of the context and monitoring their interactions in the experiences of diverse students and teachers over time would create opportunities explore their complex dynamics as they influence and are influenced by learning, motivation and relationships among those involved. Rigorously designed implementation studies are needed to determine short- and long-term consequences of activities that are and are not matched with students’ preferences (Pashler et al., 2008). Students’ expectations of learning alone and with others as well as definitions for key words students use to describe the context and support should be considered as well (Cera Guy et al., 2019; Williams, Cera Guy, & Shore, 2019). These words include difficulty, interest, fun, support and collaboration. It will also be valuable to collect descriptions of the nature and extent of their experiences working alone and with others on assignments that vary in size, duration, roles, responsibilities, and impact on their grades and relationships.

Limitations

Our findings are specific to one academic task, a challenging project, and thus are limited to that assignment. The nature of the sample also constrains the results. The restricted age range (Grades 6, 7, and 8) likely explains the small number of age-related differences. Further, although the sample included students from diverse cultures, it was predominantly Anglophone Canadian and was drawn from one school district. As a result, our findings are limited to students with similar project-related experiences, in the same grades, and in instructional contexts similar to those involved in this study. Finally, CHA and GC groups were formed using students’ scores on an assessment of reasoning abilities, the CogAT7. This should be considered when interpreting findings related to ability. Despite the psychometric shortcomings of the PCS, the themes emerging from students’ responses were relatively consistent with the findings of earlier studies and offer a few nuanced insights.

Conclusion

The tight focus on a single challenging task and the use of CogAT7 reasoning scores to define ability groups adds precision to the findings of previous efforts to untangle the complexities of learners’ preferences for working alone and with others. The results of this study affirm those indicating the general belief that all learners with high ability prefer to work alone all of the time is false, however it appears true for difficult math activities. Across school subjects, the considerations influencing students’ preferences aligned well with previous findings while the items within some themes introduced new insights. Those that influenced the preferences of students with high ability more than their agemates included some that highlighted their greater concern for the quality of their experience, project and grade. More than others their age, access to a teacher’s help contributed to the enjoyment students with high ability associated with working alone on a challenging project. They were also more conscientious, perhaps pragmatic, preferring to work alone when they felt the process and their grade would be better and collaborating when others would be needed to complete the project well. No matter their ability, students’ concerns for fairness and safety in collaborative projects played significant roles in their appeal. Interdependent social, emotional, ethical and academic aspects of the context collide in a student’s preference to work alone or with others on a particular assignment. Additional research is needed to continue to elucidate the ways in which they contribute students’ preferences, motivation and their roles in achieving the benefits of challenging individual and collaborative learning experiences.

Acknowledgement

The authors are grateful for the assistance and support provided by the following: School District #43 (Coquitlam, BC), Mrs. Louise Malfesi (Coquitlam’s District Coordinator of Gifted Education), and the students and teachers who participated in the study; Zahra Rajan for her data collection and management; and the contributions of Shun Fu Hu (UBC) and Ian Bercovitz (SFU) to the analysis and interpretation of the data.

Disclosure statement

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

Additional information

Funding

This work was supported by the Social Sciences and Humanities Research Council of Canada/Simon Fraser University [Grant number Small-2011-631004].This funding is provided by an grant program internal to Simon Fraser University so many not appear in the Open Funder Registry.