Keywords

1 Introduction

Both ATC21S [2] and OECD [14] have addressed collaborative problem solving as one of the 21st century skills for students [7]. In parallel with this research focus, several studies on science education have taken the learning-by-building approach in digital networks (e.g., [6]) as well as in the combination of the digital and physical worlds (e.g., [4, 5]). Participation in collaborative problem-solving tasks is expected to enhance the skills of sharing common understanding, taking appropriate actions, and establishing group organization. On the other hand, group work in education has frequently incurred the social loafing problem as the number of participants increases [1].

Performance in group work sometimes becomes inferior to that in individual work as the participants increase. This is caused by participants making a smaller contribution in the presence of others, and it is called social loafing, also known as the Ringelmann effect. This effect has been verified by a rope-pulling experiment as well as a clapping and shouting experiment [9, 11]. Similarly, conventional studies of the brainstorming task have pointed out that group performance sometimes produces a smaller number of ideas [3] or a lower portion of creative production [17]. On the other hand, it has also been reported that group performance is generally superior to that of an average individual, although it is often inferior to that of the best indivudual [8, 10].

Does the social loafing problem occur in a collaborative problem-solving task involving building a physical structure? This paper focuses on how group size affects performance and member satisfaction while carrying out a collaborative physical task. In this paper’s task, people were instructed to assemble a piece of furniture, a bed-side table, consisting of 6 wooden boards, 54 screws, and other hardware. We chose this task by referring to the TV-cart assembly task [13] and the large-structure assembly task [15]. Moreover, we analyzed performance in our assembly task using both behavioral and psychological indexes. In this setting, we investigated how different group sizes facilitated the individual’s involvement in the furniture-assembly task.

In our previous study, participants carried out the furniture-assembly task as individuals or in two-person or five-person groups [16]. The results of that study suggest that the effects of social loafing emerge to a greater degree in five-person groups than in two-person groups. In this paper, we additionally had three- and four-person groups carry out this task. We compared three behavioral indexes (degree of completion, time-to-completion, and duration of interaction with materials) and three psychological indexes (degree of member satisfaction, degree of member contribution, and degree of familiarity with other members) using five group sizes, from individual to five-person group.

2 Method

2.1 Predictions

 

Physical performance.:

We predict that five-person groups will complete the task within a shorter time than groups of four or fewer persons because the members of five-person groups have many hands. Consequently, the degree of completion will show a higher ratio with five-person groups than with groups of four or fewer persons.

Psychological evaluation.:

We predict that the members of five-person groups will feel lower degrees of contribution, satisfaction, and familiarity than will the members of the smaller groups because they will engage in the task for a shorter duration of interaction with the materials.

 

2.2 Participants

A total of 96 graduate and undergraduate students (mean age: 20.625 years, SD: 1.481) participated in the experiment. They were randomly assigned to individuals, two-person, three-person, four-person or five-person groups. No participant was assigned a particular role in the task. Six individuals, nine two-person, six three-person, six four-person, and six five-person groups took part in the furniture-assembly task.

2.3 Procedure

Each group was instructed to assemble the furniture as soon as possible (Fig. 1 (right)). They had to build a bed-side table, OLTEDAL of IKEA International Group, using 6 boards and 8 kinds of screws and other hardware, 54 parts in total, with an electric screwdriver according to graphical instructions (Fig. 1 (middle)).

The characteristics of the task were (a) it took 30 min by a three-person group in a preliminary experiment, (b) it required the division of roles, e.g., carrying the boards, turning the screws with the screwdriver, and checking the instructions, and (c) it was difficult to explain the graphical instructions without any caption text.

Fig. 1.
figure 1

Materials, a portion of the instructions and completed bed-side table for assembling this furniture (OLTEDAL of IKEA International Group)

Full size image

2.4 Materials

We prepared a bed-side table consisting of flat wooden boards as the target object of this furniture-assembly task. Figure 1 (left) shows the experimental materials, consisting of 6 boards, 8 kinds of screws, fasteners, and castors (54 parts in total) as well as an electric screwdriver and an instruction sheet.

2.5 Parameters

 

Behavioral Indexes.:

The experimenter calculated the following individual or group performances as behavioral indexes.

1.:

Degree of completion. Whether participants persevered to succeed in building the bed-side table or gave up building it.

2.:

Time-to-completion (TTC). The amount of time required to complete the furniture-assembly task.

3.:

Duration of interaction with materials. The amount of time duration of interaction with the materials, i.e., the boards, the screws, the screwdriver, and the instruction, per minute for an individual, one two-person group and one five-person group. We extracted interaction behaviors of handling materials using the annotation software ELAN (EUDICO Linguistic Annotator [12]).

Psychological Indexes.:

The experimenter examined the results of the following questions, which were answered on a seven-point scale through a post-experiment questionnaire.

1.:

Degree of contribution. Four questions on the participant’s degree of contribution to the task.

2.:

Degree of satisfaction. Two questions on the participant’s degree of satisfaction with the task performance.

3.:

Degree of familiarity. Eight questions on the participant’s degree of familiarity with the other participants.

 

Table 1. 2 \(\times \) 5 contingency table for successful or unsuccessful groups among five group sizes
Full size table
Fig. 2.
figure 2

Results of behavioral indexes: (a) ratio of successful and unsuccessful groups among five group sizes, (b) time-to-completion and (c) duration of interaction with materials

Full size image

3 Results

3.1 Behavioral Indexes

Table 1 and Fig. 3 show the results of a chi-square test of the degree of completion for six individuals, nine two-person, six three-person, six four-person, and six five-person groups. From this test, there were significant differences among these five group sizes \(({\chi ^{2}(2)} = 14.514, {p} < 0.05)\). The results of residual analysis suggest that three-person and five-person groups always succeeded in building the bed-side table, although the individuals usually did not succeed.

Figure 2(b) shows the results of time-to-completion (TTC) for the five group sizes. As a result of ANOVA, there was a significant tendency in TTC among the three largest group sizes (F(4) = 2.653, p = 0.054). From multiple comparisons, there were significant differences between individuals and three-person groups (p = 0.022), between individuals and four-person groups (p = 0.021), and between individuals and five-person groups (p = 0.021). These results suggest that individuals had significantly longer time-to-completion than groups of three or more persons.

Figure 2(c) shows the results of duration of interaction with materials for the five group sizes. One successful group was selected as a sample from each group size. Table 2 shows Pearson’s correlation coefficients between the group size and the duration of interaction with materials. The group size and the duration of interaction with materials correlated negatively \(({r} = -.67, {p} < .05)\). The result showed that duration of interaction with materials decreased as the group size increased.

From these results, our predictions were partly supported.

Table 2. Correlation coefficients between group size and duration of interaction with materials
Full size table

3.2 Psychological Indexes

Figure 3 shows the results of the psychological indexes, i.e., the degree of satisfaction (upper) for six individuals, nine two-person groups, six three-person groups, six four-person groups and six five-person groups, as well as the degrees of familiarity (middle) and contribution (lower) for nine two-person groups, six three-person groups, six four-person groups and six five-person groups.

Figure 3(d) shows the results for degree of satisfaction. As a result of ANOVA, a significant tendency was found in the degree of satisfaction (F(4,91) = 2.103, p = 0.087). From multiple comparisons, there were significant differences between individuals and two-person groups (p = 0.038), three-person groups (p = 0.005), four-person groups (p = 0.032), and five-person groups (p = 0.015). These results suggest that the individuals felt less satisfaction through the furniture-assembly task than did the members of the groups.

Fig. 3.
figure 3

Results of psychological indexes: (d) degree of satisfaction, (e) degree of familiarity, and (f) degree of contribution

Full size image

Figure 3(e) shows the results for degree of familiarity. As a result of ANOVA, a significant tendency was found in the degree of familiarity (F(3,86) = 2.466, p = 0.068). From multiple comparisons, there were significant differences between two-person groups and four-person groups (p = 0.016) and between four-person groups and five-person groups (p = 0.039) as well as a significant tendency between three-person groups and five-person groups (p = 0.065). Furthermore, these results suggest that the members of four-person groups felt less familiarity in the furniture-assembly task than did the members of the other groups.

Figure 3(f) shows the results of degree of contribution. As a result of ANOVA, a significant tendency was found in the degree of contribution (F(3,86) = 1.353, p = 0.263).

From these results, our predictions were partly supported.

4 Conclusion

In this study, we examined how group size affects performance and member satisfaction in the furniture-assembly task through trials with individuals, two-person, three-person, four-person and five-person groups. The results show that both group performance and member satisfaction were superior to those for individuals. On the other hand, a case study on the distribution of duration of interaction with materials suggests the possibility that some members of four-person and five-person groups exhibited the phenomenon of social loafing.

We will further examine the kinds of behaviors participants produced when they did not interact with materials, e.g., providing directions with speech or hand gestures, cheering on other participants, or just watching the status of task completion.

As our future work, we will look for ways to increase the contribution of group members who engage in social loafing. For example, an observation robot or system might emit a signal, using sound or vibration, to provoke the awareness of members showing lower contribution as well as the other participants.