Group Investigation Flipped Learning in Achieving of Students’ Critical and Creative Thinking Viewed from Their Cognitive Engagement in Learning Physics

involvement is higher than students who have low cognitive involvement; there is no interactive effect between learning models and students' cognitive ingagement on critical thinking and creative thinking. The implication of this research is that in order to achieve optimal critical thinking and creative thinking, physics learning will be better if using the GrIFL model. This is an open access article under the CC BY-SA license. Copyright © 2022 by Author. Published by Universitas Pendidikan Ganesha.


INTRODUCTION
The 21st century is a digital era marked by the rapid development of science and technology (Nurtanto et al., 2020;Trisnawati & Sari, 2019). In this century, all advances in information and communication have become media that can help every human activity (Rubini et al., 2019;Sadaf & Gezer, 2020). However, the 21st century is not an easy matter to deal with. This is caused by the development of science and technology that is not limited to starting various changes in the order of human life. These changes resulted in the emergence of global competition that cannot be avoided by everyone, including the people of Indonesia (Harahap et al., 2020;Sumantri, 2019). In overcoming this, the Indonesian people need to be directed at improving the quality of appropriate to be applied in schools. The learning model that tends to be used by educators in Indonesia so far is the direct instruction (DI) learning model which is teacher centered. Learning that is still teacher centered is one of the causes of students' low critical and creative thinking skills (Z. Aini et al., 2018;Gunawan et al., 2019). The teacher centered model is believed to have not been able to improve students' critical and creative thinking skills, especially in learning physics. Therefore, the role in the selection of physics learning models and media needs to be done properly in order to produce a learning process that supports the growth and development of students' critical and creative thinking skills.
Regarding the lack of accommodativeness of the DI model in physics learning, DI model is only able to empower a small number of students to play an active role in the learning process (Rante et al., 2013;Yunita et al., 2019). In other words, the DI learning model is not able to become an arena for most students in developing their critical and creative thinking skills. In the DI model, students are immediately asked to work on questions together in groups. This method makes students not have the responsibility individually in advance to solve the problems given. DI model the teacher is the only main and all-knowing source, while students only accept what is given by the teacher so that learning outcomes cannot optimally achieve learning objectives, especially learning objectives that are accommodating to bills of the 21st century, for example critical and creative thinking skills. Students only acquire theoretical knowledge and act passively, while teachers act actively in providing information. Students' critical thinking and creative thinking skills can develop if physics learning applies learning models that are able to involve reason, scientific attitude, engage in the research process and investigate. So, the group investigation flipped learning (GrIFL) learning model must be implemented in order to develop students' critical and creative thinking in learning physics in high school. In an effort to encourage students to be technology literate, the GI model can be used as e-learning content to realize the Group Investigation Flipped Learning (GrIFL) model. In this case, E-Learning has a function as a complement to flipped learning, and can be used as an alternative to innovative learning (Arianti, 2020;Saehana et al., 2021). E-Learning is a studentcentered learning. The investigative group model that is used as e-learning content so that the investigation flipped learning group learning is formed can actually develop students' critical thinking in learning physics (Wahyudi, 2017). The class that applied the investigative group learning model was better than the class that applied the Guided Inquiry model (A. N. Aini et al., 2020;Z. Aini et al., 2018). The investigative group learning model is very suitable for fields of study that require integrated project study activities including physics teaching oriented to acquisition, analysis, and synthesis of information in an effort to solve problems (Santyasa et al., 2019). In class XI high school physics learning, it was found that the critical thinking skills of students who studied with the Group Investigation Flipped Learning model were better than the critical thinking skills of students who studied with Direct Flipped Learning (Krisparinama et al., 2020). Based on the background, this study aims to analyze a difference in the main and interactive effects between the group investigation flipped learning model and the direct flipped learning model on students' critical and creative thinking viewed from their cognitive involvement in learning physics.

METHOD
The research design used is a one-way posttest only non-equivalent control group design, which is a type of research with one main independent variable treatment (Ahsanah, 2015;Syahrial et al., 2019). The main independent variable is the Group Inverstigation Flipped Learning (GrIFL) model which is juxtaposed with Direct Flipped Learning (DFL). The dependent variable that is measured is students' critical thinking and creative thinking in learning physics in high school. This study also examines the moderator independent variables, namely students' cognitive involvement as a separator over two dimensions in the analysis design, namely high cognitive involvement and low cognitive involvement.The population of the study was students of class XI MIPA SMA Negeri 1 Kediri Tabanan, totaling 5 classes or 170 students. The research sample was selected using a random class technique, obtained 4 classes or 104 students. The research variables consist of 1) Independent variables, namely Group Investigation Flipped Learning (GrIFL) compared to Direct Flipped Learning (DFL). This variable is not measured, but manipulated with the lesson plan and implementation and student worksheets. 2) Moderator variable, namely cognitive involvement, as measured by the ICAP model questionnaire (Chi & Wylie, 2014). 3) Critical thinking is measured by a physics critical thinking test (R. Ennis, 2013;Putri et al., 2019). 4) Creative thinking is measured by a physics creative thinking test (R. H. Ennis, 2018). Research data were analyzed by descriptive statistics and parametric statistics, with the meaning of each result carried out in a qualitative descriptive manner. Descriptive statistics were used to describe the mean (M) and standard deviation (SD) in each analysis cell. Decision making on the description of the average value and standard deviation uses a five-scale absolute value conversion guideline, namely M > 85 is very high, 70 < M 85 is high, 55 < M < 70 is moderate, 40 <M < 55 is poor, and M < 40 is very less. Parametric statistical analysis techniques were used to test the null hypothesis (Ho) against the research hypothesis (Ha). Decision-making uses the criteria, that the two-way MANOVA F value shows significant figures less than 0.05, both for testing the main influence and testing for interactive influences, meaning Ho is rejected, in other words Ha is accepted. However, before the two-way MANOVA, the assumptions were first tested, namely 1) the data were normally distributed, 2) the variance of the average value of the dependent variable between treatments was homogeneous, and 3) there was no collierity effect between the dependent variables. Testing the assumption of normality of data distribution using the criteria, that the F values of Kolmogorov-Smirnov and Shapiro-wilk show significant figures greater than 0.05, meaning the data is normally distributed. Testing the assumption of homogeneity of variance using the criteria, that the variance F values show significant figures greater than 0.05, meaning that the dependent variable variance between treatments is homogeneous. Testing the assumption that there is no collierity effect between the dependent variables uses the criteria that the product moment correlation coefficient r < 0.80.

Result
This study uses two-way MANOVA as data analysis. The results of data analysis are used to test hypotheses. As the assumptions of MANOVA are 1) the data distribution is normally distributed, 2) the variance between the dependent variables is homogeneous, and 3) there is no collinearity effect between the dependent variables. To test the normality of the data distribution, the Kolmogorov-Smirnov statistic and the Shapiro-Wilk statistic were used. The results of the analysis are shown in Table 1 and Table 2.   Table 2 shows the results of the analysis of the normality of the distribution of the distributed variable (DV) data based on cognitive engagement (CE). The table shows that the Kolmogorov-Smirnov statistics and Shapiro-Wilk statistics, both for DV critical thinking and DV creative thinking students have sig values. > 0.05. Thus, all DV data are normally distributed. To test the assumption that the data variance between DV is used Levene's statistic. The results of the analysis are presented in Table 3 and Table 4.  Table 3 shows the results of the analysis of the variance of the derived variable (DV) data based on the learning model. The table shows that Levene's statistical figures based on mean, median, median with adjusted df, trimmed mean, both for the DV variant of critical thinking and the DV variant of students' creative thinking between the GrIFL and DFL learning model groups have sig values. > 0.05. Thus, the variance of the DV data between the two learning models is homogeneous.  Table 4 shows the results of the analysis of the variance of the derived variable (DV) data based on the difference in CE. In the table it appears that Levene's statistical figures based on mean, median, median with adjusted df, trimmed mean, both for the DV variant of critical thinking and the DV variant of creative thinking of students who have HCE and DCE have sig values. > 0.05. Thus, the variance of the DV data between students who had HCE and LCE was homogeneous.The collierity test between DV is also an assumption of MANOVA. To test the collierity effect, the Pearson Correlation statistic was used with the criteria of r(count) < 0.80. The results of the analysis are shown in Table 5. In the table it appears that r(count) = 0.375 with sig. = 0.001 < 0.05. This value of r(count) is < 0.80, so that there is no collierity effect between the two DVs. The next MANOVA assumption is that there is no difference in Covariance Matrices DV. This assumption is tested with Box's Test, with the criteria that the Box's Test statistic has a sig value. > 0.05. The results of the analysis of these assumptions are presented in Table 6. These results show that the Box's Test statistic is F = 1.078 with sig. = 0.375 > 0.05. Thus, the DV covariance matrices are the same. The next analysis is a multivariate test of the effect of the GrIFL vs. DFL learning model on students' critical and creative thinking in learning physics with the moderator variable being students' cognitive engagement (Cognitive engagement/CE). The CE variable is divided into two categories, namely High CE (HCE) and Low CE (LCE). The results of this multivariate analysis are presented in Table 7. between the learning model and cognitive engagement on students' critical thinking and creative thinking in learning physics. The follow-up to the multivariate analysis was the Tests of Between-Subjects Effects learning model and cognitive engagement on each student's critical thinking and creative thinking in learning physics. However, the Tests of Between-Subjects Effects assumes that there is no difference in Error Variances between DVs. To test the Equality of Error Variances used Levene's Test. The results of the analysis are presented in Table 8 which shows that the Levene statistic numbers F = 1.389 with a sig. = 0.25 for critical thinking DV, and F = 2.088 with sig. = 0.110 for creative thinking DV. sig values. each statistic for each DV > 0.05, so the Error Variances between DV are the same. The results of the Tests of Between-Subjects Effects learning model (GrIFL vs. DFL) and cognitive engagement (HCE vs. LCE) on critical thinking and creative thinking of students in learning physics are presented in Table 9. Based on Table 9, the following research findings can be presented. First, based on the source of the influence of the learning model (GrIFL vs. DFL) on students' critical thinking, it was found that the statistical value of F = 197,875 with sig. = 0.001 < 0.05. These results indicate that there is a difference in the effect between GrIFL and DFL on students' critical thinking in learning physics. Based on Table 10, it appears that M(GrIFL) = 68,412; SD = 0.585, while M (DFL) = 56,765 with SD = 0.585. The difference between the two mean values is M = 11,647 with SE = 0.828 and sig. = 0.001 (11). So, the critical thinking of students who studied with the GrIFL model was significantly higher than those who studied with the DFL model. Second, based on the source of the influence of the learning model (GrIFL vs. DFL) on students' creative thinking, it was found that the statistical value of F = 5.967 with sig. = 0.017 < 0.05 (Table 9). These results indicate that there is a difference in the effect between GrIFL and DFL on students' creative thinking in learning physics. Based on Table 10, it appears that M (GrIFL) = 72,471; SD = 2,580, while M (DFL) = 63,559 with SD = 2,580. The difference between the two mean values is M = 8,912 with SE = 3,648 and sig. = 0.017 (Table 11). So, the creative thinking of students who studied with the GrIFL model was significantly higher than those who studied with the DFL model. Third, based on the source of the influence of CE (HCE vs. LCE) on critical thinking, it was found that the statistic number F = 1,136 with sig. = 0.291 > 0.05 (Table 9). These results indicate that there is no difference in the effect between HCE and LCE on students' critical thinking in learning physics. Based on Table 12 (Table  13). So, the critical thinking of students who have HCE is not significantly different compared to those who have LCE. Fourth, based on the source of the influence of CE (HCE vs. LCE) on creative thinking, it was found that the statistic number F = 6.126 with sig. = 0.016 < 0.05 (Table 9). These results indicate that there is a difference in the effect between HCE and LCE on students' creative thinking in learning physics. Based on Table  14, it appears that M(HCE) = 72,529; SD = 2,580, while M (LCE) = 63,500 with SD = 2,580. The difference between the two mean values is M = 9,029 with SE = 3,648 and sig. = 0.016 < 0.05 (Table 15). So, the creative thinking of students who have HCE is significantly higher than those who have LCE. Fifth, based on the source of interactive influence (CE*Model) on critical thinking with F = 1.822; sig. = 0.182; and students' creative thinking with F = 3,061; sig. = 0.085 (Table 9), that there is no interactive effect between the learning model (GrIFL vs. DFL) and students' cognitive engagement (HCE vs. LCE) on each DV of critical thinking and creative thinking of students in learning physics.

Discussion
This study aims to analyze the main and interactive effects between the learning model (Group Investigation Flipped Learning/GrIFL vs Direct Flipped Learning/DFL) and cognitive engagement (HCE vs LCE) on students' critical thinking and creative thinking skills in physics learning in class XI SMA Negeri 1 Kediri Tabanan. The results of the study showed the following findings. First, in learning physics of wave and optical materials, students who study with the GrIFL model show higher critical thinking and creative thinking skills than those who study with the DFL model. Because the two learning models use the same flipped learning (FL) model, what causes differences in students' critical thinking and creative thinking skills is the group investigation (GrI) model which has a greater influence as pedagogical content than direct instruction pedagogy. DI) in DFL. In other words, the GrI model in GrIFL has a greater effect than the DI in DFL on critical thinking and creative thinking skills. This finding is in accordance with the results of previous studies (Z. Aini et al., 2018;Akcay & Doymus, 2012;Astiti, 2018;Parinduri et al., 2017;Pitoyo et al., 2014;Santyasa et al., 2019;Sari, 2017;Yuandini & Sahyar., 2017). The class that applied the GrI learning model was better than the class that applied the Guided Inquiry model in achieving critical thinking and creative thinking (Z. Aini et al., 2018).
The GrI model has been tested for excellence in learning physics, motion and force materials for first semester students, compared to the DI model in achieving learning products (Akcay & Doymus, 2012). Physics learning in class XI SMA has proven that the GrI model is superior to the Jigsow model in achieving learning products (Astiti, 2018). In studying the physics concepts of temperature and heat, the GrI model is superior to the DI model in achieving conceptual understanding and science process skills (Parinduri et al., 2017). The GI model is the most superior compared to accelerated learning and role-playing models (Pitoyo et al., 2014). The GrI learning model is very suitable for students in physical learning who require project study activities oriented to acquisition, analysis, and synthesis of information in an effort to solve problems, so that students are able to display better critical thinking and creative thinking skills (Santyasa et al., 2019). In physics learning, the GrI model is superior to the DI model in achieving high school physics learning products (Sari, 2017). In physics learning in secondary schools, it has also been proven that the GrI model is superior to the DI model in achieving conceptual understanding (Yuandini & Sahyar., 2017). In an effort to encourage students to be technology literate, the GrI model can be used as e-learning content to realize the GrIFL model. In this case, Arianti (2020) stated that e-Learning has a function as a complement to flipped learning, and can be used as an alternative to innovative learning. Wahyudi (2017) states that e-Learning is a student-centered learning. Puspitasari (2018) states that the GrI model used as e-learning content so that GrIFL learning is formed can actually develop students' critical thinking to be better in learning physics. In class XI high school physics learning, it was revealed that the critical thinking skills of students who studied with the GrIFL model were higher than those of students who studied with DFL (Krisparinama, Santyasa, & Yasa, 2020).
Second, the results of the analysis of the main effect of students' cognitive engagement (HCE vs LCE) on students' critical thinking and creative thinking skills show that (a) there is no difference in critical thinking between students who have HCE and students who have LCE, (b) students who have HCE showed higher creative thinking skills compared to those with LCE. In the achievement of critical thinking, this finding is not in accordance with the results of research by Christopher et al. (2005), that high cognitive involvement is superior to low cognitive involvement in the achievement of critical thinking as a learning product. This is thought to be caused by the fact that students are not familiar with cognitive engagement strategies in learning. In other words, students do not fully understand the learning activities that must be carried out when they have high cognitive involvement, so that the activity of empowering their critical thinking and creative thinking skills has no typical difference with students who have low cognitive involvement. In achieving creative thinking skills, research shows that students who have HCE are more successful than those who have LCE. These results are in accordance with the statements of previous results (Fredricks et al., 2004;Greene, 2015;Smith et al., 2005). High cognitive engagement is characterized by processing mental connections and elaboration of higher-order cognitive knowledge, while shallow processing perpetuates rote learning caused by a lack of strong engagement with the learning material (Christenson et al., 2012), thus not supporting adequate efforts to acquire creative thinking. Cognitive engagement is a psychological investment that students make in learning, which shifts learning habits from memorization to the use of self-regulatory strategies to facilitate deep understanding (Fredricks et al., 2004). Learning based on self-regulation has the opportunity to form meaningful learning. Meaningful learning is based on the quality of cognitive engagement (Smith et al., 2005). Cognitive engagement is a hallmark of the principle of active learning which emphasizes the importance of cognitive engagement in learning. The quality of cognitive engagement has a direct influence on the quality of learning achievement (Greene, 2015). In other words, the ability to think creatively as a learning achievement will be more optimally obtained by students who have high cognitive involvement compared to students who have low cognitive involvement.Third, this study also reveals that there is no interactive effect between the learning model (GrIFL vs. DFL) and cognitive engagement (HCE vs. LCE) on students' critical thinking and creative thinking skills in learning physics. These findings indicate that both high cognitive engagement (HCE) and low cognitive engagement (LCE) are accommodated in both GrIFL and DFL models of physics learning. Cognitive involvement of students in learning physics is needed. Therefore, they must be guided and motivated to have adequate cognitive involvement in interacting with the facilitator, with other students, and most importantly interacting with the subject matter and other learning facilities. Intensive motivation and guidance from facilitators to students in terms of cognitive engagement will be a vehicle for students to change their minds from passive to active involvement, and from constructive to interactive engagement (Barlow et al., 2020). Changes in these thoughts will affect the effectiveness of the learning process and the optimization of learning products.

CONCLUSION
There is a different effect between the group investigation flipped learning model and the direct flipped learning model on students' critical thinking and creative thinking. The critical thinking and creative thinking of students who study with the group investigation flipped learning model are significantly higher than those who study with the direct flipped learning model. There is no different effect between students who have high cognitive involvement and those who have low cognitive involvement on critical thinking, but there is a different effect between students who have high cognitive involvement and those who have low cognitive involvement on creative thinking. The creative thinking of students who have high cognitive involvement is significantly higher than students who have low cognitive involvement. There is no interactive effect between the learning model (group investigation flipped learning vs. direct flipped learning model) and cognitive engagement (high cognitive involvement vs. low cognitive engagement) on students' critical thinking and creative thinking in class XI physics learning in high school. The implication of the findings of this study is that in achieving critical thinking and creative thinking skills in physics learning, the group investigation flipped learning model and the direct flipped learning model are both accommodating to students' high cognitive engagement and low cognitive engagement. In studying physics in class XI SMA on wave and optical materials, the learning process and results, especially the students' creative thinking results, will be more optimally achieved if they are facilitated with the group investigation flipped learning model. Because in learning, students are influenced by their cognitive involvement, students who have low cognitive involvement should be given more intensive guidance so that they are able to do more activities, especially in increasing their cognitive involvement in order to optimize their learning process and in achieving their creative thinking skills in learning physics.