2. Theoretical Framework
Student motivation is a key factor for their learning, as indicated by various studies [4
]. In fact, the implementation of gamification and video games in education has improved motivation for primary education students [7
], secondary education students [10
] and even higher education students [13
]. In the same way, the implementation of collaborative learning methodologies has also contributed to students’ motivation [6
Furthermore, gamification and video games in education also enable learning gains for students in terms of knowledge, skills and attitudes, for different education system stages, including primary education [19
], secondary education [22
] and higher education [16
]. At the same time, the implementation of collaborative learning methodologies also contributes to student learning [17
This motivated us to create a methodology called “collaborative learning with video games”, which brings together, in a single methodology, the advantages and criteria of implementing video games and collaborative learning in education. In that sense, it can be understood as the implementation of educational activities in which students have to work together, sharing responsibilities to achieve a goal (for instance, to do a task, to do a project, to complete a chart, to create a digital presentation, to write an essay, etc.), while discussing different perspectives and contributing with their ideas. The main resource of this activity is a video game [30
]. It is important to highlight that collaborative learning between students can happen inside the video game, outside the game, or in both spaces (inside and outside the video game) depending on the type of educational strategy or activity the teacher chooses to implement. Educational experiences using this methodology are found in the literature [31
]. Martín [35
] provided further examples.
Considering these methodologies, students feel more motivated to face the educational process, and in turn, obtain learning gains. This poses a new question: are teachers motivated to implement these new technologies in their educational practices? What are teachers’ attitudes, opinions and perspectives about it? If teachers are not motivated in their implementation, it will be difficult for students to feel motivated when these resources are presented as part of the learning process. In fact, as Tejedor and García-Valcárcel [3
] said, one of the biggest factors that influences the integration of any pedagogical innovation, new methodology or new technological resource in educational practices, is the attitude of affected teachers. Therefore, it is important to understand teachers’ viewpoints on these issues.
In this regard, we could find several studies referring to in-service and pre-service teachers’ attitudes towards video games in education [1
], and towards gamification in educational settings [39
]. In general, research shows that teachers’ attitudes towards these approaches are positive.
This article focuses on collaborative learning with video games; in particular, it reveals the creation and subsequent validation of a Likert-type attitude scale towards collaborative learning with video games, aimed at in-service primary education teachers.
3. Materials and Methods
The creation of an attitude scale requires a rigorous process in order to obtain an appropriate and validated instrument to measure the target attitude. At the same time, it is important to highlight that there are different types of attitude scales. In our specific case, we selected a Likert-type attitude scale. Likert-type attitude scales use a construction method that adapts to the measurement of different types of attitudes [43
]. We took the ideas of multiple authors into account in creating this instrument [44
Morales, Urosa and Blanco [44
] synthesized the process of creation for this type of scale into specific stages, including the definition of the specific attitude to be measured as a key step. The next steps are: (1) the preparation of the instrument through the writing of several items and the preparation of additional information; (2) the obtainment of data from an adequate sample; (3) the item analysis, the calculation of reliability, the analysis of the scale content structure and the selection of definitive items.
It is fundamental to define the attitude to be measured, which, in our case, is teachers’ attitudes towards collaborative learning with video games. This can be defined as the relatively stable predisposition of teachers to respond favorably or unfavorably to the implementation of educational activities in which students have to work together, sharing responsibilities to achieve a goal (for instance, to do a task, to do a project, to complete a chart, to create a digital presentation, or to write an essay), while discussing different perspectives and contributing with their ideas, where the main resource of this activity is a video game.
Regarding the writing of the items of the scale, it is necessary to create several sentences. As Morales [46
] said, they are usually written in the form of opinions with which the person may or may not agree. In addition, other instruments that measure the same or similar attitudes can be taken into account for elaboration purposes. For that reason, we took the measurement instrument called “Semantic differential: learning through collaborative projects with Information and Communication Technologies (ICT)” [47
] and the questionnaire “Opinion about collaborative learning methodology” [17
] into account. A total of 75 preliminary items were prepared at this time. As Morales, Urosa and Blanco [44
] pointed out, provisional items should be reviewed by more than one person, allowing for modification or elimination as needed. In our case, they were reviewed by an in-service primary education teacher (that is to say, a professional with a similar status to the final recipients of this instrument), by an expert in written comprehension and written composition, and an expert in educational technology. This process gave rise to 64 items. The following response mode was established: (1) strongly disagree (SD), (2) disagree (D), (3) indifferent (I), (4) agree (A), and (5) strongly agree (SA). As Morales, Urosa and Blanco [44
] pointed out, additional information must be prepared, so other questions (specifically, 20 questions) were also incorporated into the complete questionnaire to provide more information including, for instance, age, gender, courses and disciplines taught this year, and university degrees finished.
Once we acquired a sufficient number of items and decided how to respond to them, it was necessary to validate our preliminary instrument by means of a methodologically appropriate procedure. Specifically, the instrument developed up to this point had to be subject to a content validation process by expert judgment. As Cabero and Barroso [48
] said, the use of experts as a strategy to assess teaching materials, as well as data collection instruments (as in our case) or the methodologies used in the educational process, is quite common in the field of educational research. Six experts took part in our study. Considering that they were experts in more than one field, it is important to highlight that four were primary school teachers, five were experts in the implementation of information and communication technologies (ICT) or video games in educational settings, two were experts in collaborative learning, and two were experts in construction and validation of measuring instruments and research methods. Considering their own opinions and knowledge, the experts had to rate the validity of each item from 1 (very bad) to 5 (very good) in relation to the objective of the scale and the specific attitude to be measured. They could also submit other suggestions about the instrument in general and about specific items in a space created in the questionnaire for scale validation. Regarding the selection criteria of the items at this stage, that is, which preliminary items were kept in the instrument and which were not, we took into account the percentage of experts who considered that an item was good or very good (taking as criterion to be above 60% of experts), the average obtained (establishing as criterion an average equal to or greater than 4), and the specific comments and suggestions provided by the experts. In addition, in the case that some of the items did not meet these criteria, we took the utility of that item on the scale into account, and were able to keep those items for the next step in which they were used with a sample. As a result of this process, we kept 57 items of the initial 64, namely those that applied to in-service primary school teachers.
We then went to the next step of creating the scale, which involved obtaining data from appropriate samples. Specifically, the sample included 223 Spanish in-service primary school teachers. With this data, we proceeded to carry out item analysis, reliability calculation, analysis of the scale content structure and final selection of the items using the statistical software SPSS 22 (IBM Corporation, Armonk, NY, USA). These aspects will be presented in the following section.
4. Data Analysis and Results
Firstly, we present the results that show the sample characterization. The total sample was 223 Spanish in-service primary school teachers, and 113 were men (50.7%) and 110 were women (49.3%). The respondents were 21–60 years old, the statistical mode was 38 years old (5.4% of the sample) and the mean was 36.09. In regards to what levels they teach (considering that primary education in Spain consists of six years, from 6 to 12 years old), 79 teachers taught the first year, 71 the second year, 78 the third year, 81 the fourth year, 87 the fifth year, and 87 the sixth year. It is important to highlight that in Spain the same teacher can work across different age levels depending on the discipline. In regards to their university-level education, excluding their undergraduate degree to be a primary school teacher, 17 teachers (7.6%) were enrolled in a master’s degree and 36 teachers (16.1%) held a master’s degree at the time of answering the questionnaire. Furthermore, nine teachers (4%) were enrolled in a PhD, and five teachers (2.24%) had finished their doctoral studies.
Once we obtained data from a sample, the next step was to analyze the items and the verification of reliability. We had to check whether each item from the initial version measured the same attitude as the other items. This is fundamental, in order to know if it is possible to sum its specific item score in a total score that supposedly measures the attitude that we want to study, taking into account that the total score of each person is one that will later be interpreted. This check was done through item analysis, and we used the item–total correlation procedure, which is, properly speaking, the correlation of each item with the sum of all others, or the correlation of each item with the total minus the item, i.e., the corrected item–total correlation [44
]. We wanted to check whether scoring high on an item means, in fact, getting a high total score on the rest of the scale. When selecting our items, we had to take into account the fact that those with scores that correlated most highly with the sum of all the others are those that have more in common, and we can assume they measure the same as the rest. However, the items that show non-significant or very low correlations in relation to the rest of the items must be eliminated from the scale [44
]. In addition, it must be taken into account that the process does not have to be automatic, rather the researchers’ ideas about what they are trying to measure need to be considered, so conceptual criteria must also be taken into account. Thus, in Appendix A
we show the total-element statistics of the items in the 57 item scale version, and in Appendix B
we show the total-element statistics of the items in the final version of 33 items; that is, these are the definitive 33 items that will be part of the scale (the complete items can be seen in Appendix C
, translated from Spanish). It is important to highlight that definitive items are numbered in the appendices and in tables according to their sequence in the final instrument and in Appendix C
(from ‘item 1’ to ‘item 33’), while the eliminated items have been named through letters (from ‘eliminated item A’ to ‘eliminated item X’).
Considering the above, this gives rise to an attitude scale of 33 items, with a reliability of α = 0.947. Furthermore, we carried out factorial analysis as a method to check the construct validity of the final version with the 33 items. As Morales, Urosa and Blanco [44
] pointed out, factorial analysis with the rotated factors allows us to appreciate whether we are measuring what we say we measure, by clarifying the matters that underlie several variables, what items are defining each factor and how these factors relate to each other, helping us to clarify the structure of the instrument and the construct. In Table 1
, we show data about the factors extracted in the analysis, and in Table 2
, we show the rotated component matrix for the 33 item final version of the scale.
As we can see in Table 1
, six factors were extracted that explain 60.882% of the total variance, taking into account that, to determine the number of factors that have to be extracted, those components with eigenvalues greater than 1 are conserved [49
]. This also fulfils what was indicated by Nunnally [43
], because it is necessary to eliminate the factors in which no variable has a weight superior to 0.30, and, as can be appreciated in Table 2
, all factors have some variable with a weight greater than this. In addition, it is necessary to take into account only those factors that are defined by at least three items [44
], which is what happens with our six factors (as can be seen in Table 2
in the different columns for each component).
In order to interpret factor structure, we examined the saturations that, in each factor, obtained the items of the scale [49
] according to the results in Table 2
. We attended mainly to those items with the largest weights [44
] and chose, in those cases where there are items that were saturated in more than one factor, to place them with the factor in which they saturated the most.
As we can see in Table 2
, the first factor is integrated by items 4, 6, 10, 12, 14, 17, 18, 19, 20, 22, 23, 24 and 28, and explains 38.9% of the variance (as we can see in the third column of Table 1
). We denominated it “educational possibilities”, because the items highlight the educational possibilities of collaborative learning with video games. For instance, the greater the interaction between the teacher and the students, the greater the students’ autonomy in their learning, the development of students’ capacity for initiative and the possibility to explore ideas and concepts more fully.
The second factor is integrated by items 2, 21, 25, 27, 31, 32 and 33, and explains 6.41% of the variance. We called it “positive disposition to implement activities” by incorporating those items that include formulations showing interest, inclination or attraction towards the approach of collaborative learning activities with video games, for example, showing interest to collaborate with other teachers who implement these kinds of activities or showing interest to work in a school where this methodology was supported.
The third factor is integrated by items 1, 26, 29 and 30, and explains 5.656% of the variance. We called it “denial as educational methodology” because the items are related to the rejection of collaborative learning with video games as a possible methodology to be applied in educational practices, indicating that implementing this methodology is impossible and inappropriate. Taking this into account, it is important to highlight that all the items in this factor are negative and, in the analysis, it was necessary to reverse the score.
The fourth factor is integrated by items 5, 8 and 13, and explains 3.421% of the variance. We denominated it “concerns about neglecting the learning” by incorporating those items that are related to teachers’ concerns about the implementation of this kind of methodology and the problem of neglecting or not giving the required importance to learning by the students, such as taking learning lightly and not putting effort into educational tasks. As for the previous factor, all the items in this factor are also negative and, in the analysis, it was necessary to reverse the score.
The fifth factor is integrated by items 9, 11 and 16, and explains 3.335% of the variance. We denominated it “useful and inclusive learning strategy” by incorporating those formulations related to the idea of collaborative learning with video games methodology as a learning strategy that allows the inclusion of all students and that allows learning relevant matters for their lives in the complex and diverse world in which we live.
Finally, the sixth factor is integrated by items 3, 7 and 15, and explains 3.161% of the variance. We called it “teacher denial due to loss of time” by incorporating those formulations in which the teacher rejects this approach, considering it a waste of time in terms of class time and personal time. In this case, the three items are negative and, in the analysis, it was also necessary to reverse the score.
The analysis of items shown, and the factorial analysis carried out, led us to confirm the selection of the 33 items as elements for the final version of the scale. The scale has a reliability of α = 0.947, with the reliability of each factor as following:
Factor 1 “educational possibilities”: α = 0.921.
Factor 2 “positive disposition to implement activities”: α = 0.876
Factor 3 “denial as educational methodology”: α = 0.762
Factor 4 “concerns about neglecting the learning”: α = 0.814
Factor 5 “useful and inclusive learning strategy”: α = 0.662
Factor 6 “teacher denial due to loss of time”: α = 0.696.
Finally, it should also be noted that, although we tried to have the same number of items in the affective, cognitive and behavioural fields, the scale has the following final structure:
Twelve items related to the affective field (items 2, 3, 5, 7, 8, 13, 21, 25, 26, 27, 29 and 33): α = 0.873
Thirteen items related to the cognitive field (items 1, 4, 6, 9, 10, 12, 14, 17, 18, 20, 22, 24 and 28): α = 0.904
Eight items related to the behavioural field (items 11, 15, 16, 19, 23, 30, 31 and 32): α = 0.832.